availability dict | codec unknown | cross_refs dict | derived_from_artifact_ref stringclasses 9
values | disclosures listlengths 1 6 | huggingface dict | id stringlengths 29 84 | kind stringclasses 3
values | model_ref stringclasses 19
values | name stringlengths 16 102 | producer dict | schema_version stringclasses 1
value | scope unknown | scope_digest stringlengths 194 463 | seal unknown | sources listlengths 1 13 | weights unknown |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
{
"status": "public",
"uri": "https://huggingface.co/0xSero/GLM-5.3-Flash-EXL3-3.0bpw"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 3,
"calibration": {
"corpus": "0xSero release-calibration-manifest: 600 rows x 2048 columns over c4/code/multilingual/technical/tiny/wiki (six utf8 corpus files, each sha256'd in evidence/release-calibration-manifest.json) plus 92 random-token ro... | {
"local_ai_registry": {
"match_confidence": "probable",
"model_id": "glm-5.3-flash",
"model_instance_id": null,
"url": "https://huggingface.co/datasets/0xSero/local-ai-registry"
}
} | artifact--zai-org.glm-5.3-flash-bf16.a6c167b6 | [
{
"affects_comparability": true,
"code": "unsealed_source",
"detail": "The Dione checkpoint ships no upstream receipts, reconstruction closures or sealed reader ABI. The packed surface was decoded WITHOUT seal verification. What it DOES publish is a per-shard sha256 manifest, and all 130 shard digests w... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "0xSero/GLM-5.3-Flash-EXL3-3.0bpw",
"revision": "8b099bf276507a17faea920deff3f62d5597fb52",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/0xSero/GLM-5.3-Flash-EXL3-3.0bpw"
} | artifact--0xsero.glm-5.3-flash-exl3-3.0bpw | quant | model--zai-org.glm-5.3-flash | 0xSero GLM-5.3-Flash EXL3 3.0bpw (Dione, K3, TP4-sliced, native BF16 head) | {
"handle": "0xSero",
"is_registry_maintainer": false,
"name": "0xSero",
"role": "quantizer",
"url": "https://huggingface.co/0xSero"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "retained at source precision in the release's own retained/ shards. read from the release's OWN config.json quantization_config (quant_method=exl3_selective_tp4, ... | attn.o=native:bf16@16|attn.other=native:mixed|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:exl3-mcg@3|moe.router=native:fp32@32|moe.shared_expert=native:bf16@16|mtp=native:bf16@16|norm=native:bf16@1... | {
"note": "unsealed source; the release publishes no seal. Its 130 shard sha256s (EXL3_MANIFEST.json, schema_version 1) were recomputed in full on the measurement instance.",
"sealed": false
} | [
{
"kind": "url",
"note": "335 files, 130 safetensors; all-files sum 149,556,991,042; safetensors sum 149,402,871,912; the index's own metadata.total_size 149,325,518,712",
"uri": "https://huggingface.co/api/models/0xSero/GLM-5.3-Flash-EXL3-3.0bpw?blobs=true&revision=8b099bf276507a17faea920deff3f62d5597f... | {
"config_sha256": "0f97529bf936f823b84dc77ac0e09190e9439094e29dc8895547df8e17e4a24e",
"container": "exl3",
"index_sha256": "0fd35de9b0d5fc9428a45d3b311dc757ea891e4cec7788050b75089593ad3215",
"precision_label": "3.0 bpw",
"shard_count": 130,
"size_basis": "repo_all_files",
"size_bytes": 149556991042,
"s... |
{
"status": "public",
"uri": "https://huggingface.co/0xSero/GLM-5.3-Flash-EXL3-Q4"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "probable",
"model_id": "glm-5.3-flash",
"model_instance_id": null,
"url": "https://huggingface.co/datasets/0xSero/local-ai-registry"
}
} | artifact--zai-org.glm-5.3-flash-bf16.a6c167b6 | [
{
"affects_comparability": false,
"code": "scope_record_corrected",
"detail": "Superseded record: this artifact's scope previously read embed_tokens, attn.qkv, attn.o and lm_head as `unknown`, with the note that the release 'declares a scope policy that was not parsed into this registry'. It is parsed n... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "0xSero/GLM-5.3-Flash-EXL3-Q4",
"revision": "99cccdf0e8741715662c383828a9ea601990c125",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/0xSero/GLM-5.3-Flash-EXL3-Q4"
} | artifact--0xsero.glm-5.3-flash-exl3-q4 | quant | model--zai-org.glm-5.3-flash | 0xSero GLM-5.3-Flash EXL3 Q4 (Dione, TP4-sliced) | {
"handle": "0xSero",
"is_registry_maintainer": false,
"name": "0xSero",
"role": "quantizer",
"url": "https://huggingface.co/0xSero"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "retained at source precision in the release's own retained/ shards. read from the release's OWN config.json quantization_config (quant_method=exl3_selective_tp4, ... | attn.o=native:bf16@16|attn.other=native:mixed|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:exl3-mcg@4|moe.router=native:fp32@32|moe.shared_expert=native:bf16@16|mtp=native:bf16@16|norm=native:bf16@1... | {
"note": "unsealed source; see the unsealed_source disclosure",
"sealed": false
} | [
{
"kind": "hf_file",
"note": "malaiwah.glm53-dione-q4-packed-kld-summary.v1: dione_repo, dione_revision 99cccdf0..., dione_shard_hash_verification=full",
"sha256": "d18b37d8ed1ba90ed837d1fb2adca0b90999b2d702613f6730ef87fe23d9f9b7",
"uri": "https://huggingface.co/datasets/malaiwah/GLM-5.3-Flash-fidel... | {
"container": "exl3",
"precision_label": "Q4",
"shard_count": 217,
"size_basis": "repo_all_files",
"size_bytes": 187607584245,
"size_gb": 187.607584245,
"tensor_parallel": {
"pre_sliced": true,
"world_size": 4
}
} |
{
"status": "public",
"uri": "https://huggingface.co/brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 3,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": false,
"tokens": null,
"used": false
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revisi... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.2-bf16 | [
{
"affects_comparability": false,
"code": "tp_sliced_artifact",
"detail": "Read from bytes: the 57,600 routed-expert matrices are stored as FOUR tensor-parallel rank shards each (230,400 trellis payload groups, all at K=3, mcg codebook, 58,368 mcg objects), and the capture composed them into whole weigh... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw",
"revision": "f79c9167690ca705e877ae4dc55a841d1aae1247",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw"
} | artifact--brandonmusic.glm-5.2-exl3-tr3-3.0bpw | quant | model--zai-org.glm-5.2 | brandonmusic GLM-5.2-EXL3-TR3-3.0bpw (routed experts trellis K3, TP4 rank-sharded, rest native) | {
"handle": "brandonmusic",
"is_registry_maintainer": false,
"name": "brandonmusic",
"role": "quantizer",
"url": "https://huggingface.co/brandonmusic"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "78 tensors: model.layers.N.self_attn.o_proj.weight. read from brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw@f79c9167690ca705e877ae4dc55a841d1aae1247 model.safetensors.inde... | attn.o=native:bf16@16|attn.other=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:exl3-mcg@3|moe.router=native:mixed|moe.shared_expert=native:bf16@16|mtp=quantized:mixed|norm=native:bf16@... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision f79c9167690c; 81 shards; config.json sha256 fcde0013... declares the artifact in hybrid_tr3_tail (format exl3-trellis, codebook mcg, tp 4, bits_avg 3.0); index sha256 346227a4...",
"uri": "https://huggingface.co/brandonmusic/GLM-5.2-EXL3-TR3-3.0bpw"
},
{
... | {
"config_sha256": "fcde001350291a0048318d4a1136e0732e31f829f804a57cfbb558903e54171a",
"container": "exl3",
"index_sha256": "346227a4ea44b6063017739ee38a830319dc10305ccf714734095e27b28064c2",
"precision_label": "3.0bpw",
"shard_count": 81,
"size_basis": "repo_weight_files",
"size_bytes": 316420224008,
"... |
{
"status": "public",
"uri": "https://huggingface.co/zai-org/GLM-5.3-Flash"
} | {
"bits_per_weight_effective": 8,
"bits_per_weight_nominal": 8,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "fp8_e4m3",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-fp8 | [
{
"affects_comparability": true,
"code": "revision_unpinned",
"detail": "No measurement receipt for this artifact records a Hub revision. brandonmusic's runtime receipts record no zai-org revision.",
"severity": "caveat"
},
{
"affects_comparability": false,
"code": "record_note",
"de... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "zai-org/GLM-5.3-Flash",
"revision": null,
"revision_source": "none",
"status": "known",
"url": "https://huggingface.co/zai-org/GLM-5.3-Flash"
} | artifact--brandonmusic.glm-5.3-flash-fp8-mla-kv | quant | model--zai-org.glm-5.3-flash | GLM-5.3-Flash official FP8 weights served with FP8 MLA KV | {
"handle": "zai-org",
"is_registry_maintainer": false,
"name": "Z.ai",
"role": "quantizer",
"url": "https://huggingface.co/zai-org"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 8,
"format": "fp8_e4m3",
"layer_range": "all",
"tenso... | attn.o=quantized:fp8_e4m3@8|attn.qkv=quantized:fp8_e4m3@8|embed_tokens=native:bf16|kv_cache=quantized:fp8_e4m3@8|lm_head=native:bf16|mlp.down=quantized:fp8_e4m3@8|mlp.gate=quantized:fp8_e4m3@8|mlp.up=quantized:fp8_e4m3@8|moe.experts=quantized:fp8_e4m3@8|norm=native:bf16|head=native|kv=fp8_e4m3 | {
"sealed": false
} | [
{
"kind": "github_file",
"note": "regime: 'FP8 MLA NoPE ... eager no-MTP, full 2048-token window'",
"sha256": "409a3487925a98b40d97c174b5e44e2b3526794d14c5e7ef5a35fd5f669b3209",
"uri": "https://raw.githubusercontent.com/brandonmmusic-max/glm-5.3-flash-exl3-4bpw/main/runtime-results/v75/kld/fp8-five-... | {
"container": "safetensors",
"precision_label": "FP8 + FP8 MLA KV",
"size_basis": "unknown",
"size_bytes": null,
"size_gb": null
} |
{
"status": "unknown",
"uri": null
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "nvfp4",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-bf16.a6c167b6 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
},
{... | {
"link_type": "none",
"path": null,
"reason": "No published NVFP4 checkpoint repository was located. The artifact exists as a serving configuration inside his runtime image, not as downloadable weights.",
"repository": null,
"revision": null,
"revision_source": "none",
"status": "unavailable",
"url": n... | artifact--brandonmusic.glm-5.3-flash-nvfp4-runtime | quant | model--zai-org.glm-5.3-flash | brandonmusic GLM-5.3-Flash NVFP4 runtime build | {
"handle": "brandonmusic",
"is_registry_maintainer": false,
"name": "brandonmusic",
"role": "quantizer",
"url": "https://huggingface.co/brandonmusic"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": 4,
"format": "nvfp4",
"layer_range": "all",
"tens... | attn.o=quantized:nvfp4@4|attn.qkv=quantized:nvfp4@4|embed_tokens=unknown:unknown|kv_cache=quantized:nvfp4@4|lm_head=unknown:unknown|mlp.down=quantized:nvfp4@4|mlp.gate=quantized:nvfp4@4|mlp.up=quantized:nvfp4@4|moe.experts=quantized:nvfp4@4|head=unknown|kv=nvfp4 | {
"sealed": false
} | [
{
"kind": "github_file",
"sha256": "c01cc32afb1802eaba317edc3c1ef90ae649f368307ff5c8957f37bccac78755",
"uri": "https://raw.githubusercontent.com/brandonmmusic-max/glm-5.3-flash-exl3-4bpw/main/runtime-results/v44/kld/nvfp4-five-run-kld-receipt.json"
}
] | {
"container": "other",
"precision_label": "NVFP4 + NVFP4 MLA KV",
"size_basis": "unknown",
"size_bytes": null,
"size_gb": null
} |
{
"status": "public",
"uri": "https://huggingface.co/brandonmusic/GLM-5.3-Flash-tr3-4bpw"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-bf16.a6c167b6 | [
{
"affects_comparability": false,
"code": "scope_record_corrected",
"detail": "Superseded record: this artifact's scope previously read attn.qkv / attn.o / mlp.{gate,up,down} as quantized:exl3-mcg at the nominal rate. That was wrong. The release is routed-experts-only: its own config.json declares scope... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "brandonmusic/GLM-5.3-Flash-tr3-4bpw",
"revision": null,
"revision_source": "none",
"status": "known",
"url": "https://huggingface.co/brandonmusic/GLM-5.3-Flash-tr3-4bpw"
} | artifact--brandonmusic.glm-5.3-flash-tr3-4bpw | quant | model--zai-org.glm-5.3-flash | brandonmusic GLM-5.3-Flash tr3 4bpw | {
"handle": "brandonmusic",
"is_registry_maintainer": false,
"name": "brandonmusic",
"role": "quantizer",
"url": "https://huggingface.co/brandonmusic"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "routed-experts-only scope: not quantized",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"form... | attn.o=native:bf16@16|attn.other=native:mixed|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:exl3-mcg@4|moe.router=native:fp32@32|moe.shared_expert=native:bf16@16|mtp=quantized:exl3-mcg@4|norm=native:... | {
"note": "ships its own five-cold-run receipt and reader digest 1fb3be87...",
"sealed": true
} | [
{
"kind": "github_file",
"note": "student_checkpoint_identity_sha256 598ce08d..., student_label uniform-k4, profile k4-tp2",
"sha256": "d955bfaedad36ad9841c30808c67fc36b72017f87b720fb460d8e1c13fe75e57",
"uri": "https://raw.githubusercontent.com/brandonmmusic-max/glm-5.3-flash-exl3-4bpw/main/results/... | {
"container": "exl3",
"precision_label": "4bpw",
"size_basis": "repo_weight_files",
"size_bytes": 175642157700,
"size_gb": 175.6421577
} |
{
"status": "public",
"uri": "https://huggingface.co/cyankiwi/Qwen3.8-27B-AWQ-INT4"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "awq",
"group_size": 32
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "Scope derived from the artifact's OWN config and weight index, not guessed. quantization_config declares compressed-tensors pack-quantized, num_bits=4, group_size=32, observer=mse, and a 313-entry `ignore` list. Verified against the rea... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "cyankiwi/Qwen3.8-27B-AWQ-INT4",
"revision": "63768c10df38c0395e12ef49edac1bd539eaeeea",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/cyankiwi/Qwen3.8-27B-AWQ-INT4"
} | artifact--cyankiwi.qwen3.8-27b-awq-int4 | quant | model--qwen.qwen3.8-27b | cyankiwi Qwen3.8-27B AWQ-INT4 | {
"handle": "cyankiwi",
"is_registry_maintainer": false,
"name": "cyankiwi",
"role": "quantizer",
"url": "https://huggingface.co/cyankiwi"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 4,
"format": "int4",
"layer_range": "all",
"tensor_cl... | attn.o=quantized:int4@4|attn.qkv=quantized:int4@4|embed_tokens=native:bf16|lm_head=native:bf16|mlp.down=quantized:int4@4|mlp.gate=quantized:int4@4|mlp.up=quantized:int4@4|norm=native:bf16|head=native|kv=bf16 | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision 63768c10..., index_sha256 82b1bf79..., 2,396 index entries; scope derived from the artifact's own quantization_config + weight index by k6/tools/derive_scope.py",
"uri": "https://huggingface.co/cyankiwi/Qwen3.8-27B-AWQ-INT4"
}
] | {
"container": "safetensors",
"index_sha256": "82b1bf79f5b61333e83da17ec3bf89c9f178e29395a14c6b3ce3bbc474e1ead8",
"precision_label": "INT4",
"size_basis": "repo_all_files",
"size_bytes": 21041255795,
"size_gb": 21.041255795
} |
{
"status": "public",
"uri": "https://huggingface.co/davidsyoung/GLM-5.3-EXL3-TR3-3.0bpw"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 3,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": false,
"tokens": null,
"used": false
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revisi... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-bf16 | [
{
"affects_comparability": false,
"code": "tp_sliced_artifact",
"detail": "Read from bytes: the 57,600 routed-expert matrices are stored as FOUR tensor-parallel rank shards each (model.layers.N.mlp.experts.E.<proj>.rank{0..3}.{trellis,suh,svh,mcg}), the artifact's own hybrid_tr3_tail declares tp=4 and t... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "davidsyoung/GLM-5.3-EXL3-TR3-3.0bpw",
"revision": "eeab94eb6e95b4e4d13d94af55ab3c420d6f52d3",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/davidsyoung/GLM-5.3-EXL3-TR3-3.0bpw"
} | artifact--davidsyoung.glm-5.3-exl3-tr3-3.0bpw | quant | model--zai-org.glm-5.3 | davidsyoung GLM-5.3 EXL3 TR3 3.0bpw (routed experts trellis, TP4 rank-sharded) | {
"handle": "davidsyoung",
"is_registry_maintainer": false,
"name": "davidsyoung",
"role": "quantizer",
"url": "https://huggingface.co/davidsyoung"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "78 tensors: model.layers.N.self_attn.o_proj.weight. read from davidsyoung/GLM-5.3-EXL3-TR3-3.0bpw@eeab94eb6e95b4e4d13d94af55ab3c420d6f52d3 model.safetensors.index... | attn.o=native:bf16@16|attn.other=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:exl3-mcg@3|moe.router=native:mixed|moe.shared_expert=native:bf16@16|mtp=quantized:mixed|norm=native:bf16@... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision eeab94eb6e95; 81 shards; config.json sha256 c1deb536... carries a leftover ModelOpt quantization_config (quant_method modelopt) and the artifact's real declaration in hybrid_tr3_tail: format exl3-trellis, codebook mcg, tp 4, bits_avg 3.0, k_values [3], exllamav3_ver... | {
"config_sha256": "c1deb536b1f9ec866a99b33fdf85726fdf093c10b4f455016b2e3865b7a3a9c4",
"container": "exl3",
"index_sha256": "80b1bd429403516791107763a7043d896e1dc03e7e848fed3331bdf734741f91",
"precision_label": "3.0bpw",
"shard_count": 81,
"size_basis": "repo_weight_files",
"size_bytes": 316420269320,
"... |
{
"status": "public",
"uri": "https://huggingface.co/davidsyoung/GLM-5.3-EXL3-TR3-3.25bpw"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 3.25,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": false,
"tokens": null,
"used": false
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"rev... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-bf16 | [
{
"affects_comparability": false,
"code": "tp_sliced_artifact",
"detail": "Read from bytes: the 57,600 routed-expert matrices are stored as FOUR tensor-parallel rank shards each (model.layers.N.mlp.experts.E.<proj>.rank{0..3}.{trellis,suh,svh,mcg}), the artifact's own hybrid_tr3_tail declares tp=4 and t... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "davidsyoung/GLM-5.3-EXL3-TR3-3.25bpw",
"revision": "6d6bd738c0c1635513e0bd0fdf0302049bd820a9",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/davidsyoung/GLM-5.3-EXL3-TR3-3.25bpw"
} | artifact--davidsyoung.glm-5.3-exl3-tr3-3.25bpw | quant | model--zai-org.glm-5.3 | davidsyoung GLM-5.3 EXL3 TR3 3.25bpw (routed experts trellis, TP4 rank-sharded) | {
"handle": "davidsyoung",
"is_registry_maintainer": false,
"name": "davidsyoung",
"role": "quantizer",
"url": "https://huggingface.co/davidsyoung"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "78 tensors: model.layers.N.self_attn.o_proj.weight. read from davidsyoung/GLM-5.3-EXL3-TR3-3.25bpw@6d6bd738c0c1635513e0bd0fdf0302049bd820a9 model.safetensors.inde... | attn.o=native:bf16@16|attn.other=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:exl3-mcg@3.25|moe.router=native:mixed|moe.shared_expert=native:bf16@16|mtp=quantized:mixed|norm=native:bf... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision 6d6bd738c0c1; 81 shards; config.json sha256 bb6589b2... carries a leftover ModelOpt quantization_config (quant_method modelopt) and the artifact's real declaration in hybrid_tr3_tail: format exl3-trellis, codebook mcg, tp 4, bits_avg 3.25, k_values [3, 4], exllamav3... | {
"config_sha256": "bb6589b29dfa52cbb16d5612cb7cd60d3bedf65ca0ae4389f377e3cffd96b47f",
"container": "exl3",
"index_sha256": "ae6a9fac0679c875a5b3d4a9524b4204a6d97f0375daf40e7ebd5cd10c184753",
"precision_label": "3.25bpw",
"shard_count": 81,
"size_basis": "repo_weight_files",
"size_bytes": 339371523592,
... |
{
"status": "public",
"uri": "https://huggingface.co/davidsyoung/GLM-5.3-EXL3-TR3-3.42bpw"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 3.421875,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": false,
"tokens": null,
"used": false
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-bf16 | [
{
"affects_comparability": false,
"code": "tp_sliced_artifact",
"detail": "Read from bytes: the 57,600 routed-expert matrices are stored as FOUR tensor-parallel rank shards each (model.layers.N.mlp.experts.E.<proj>.rank{0..3}.{trellis,suh,svh,mcg}), the artifact's own hybrid_tr3_tail declares tp=4 and t... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "davidsyoung/GLM-5.3-EXL3-TR3-3.42bpw",
"revision": "99c6f951333d2b38f1efefa533c7afadf0d376e3",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/davidsyoung/GLM-5.3-EXL3-TR3-3.42bpw"
} | artifact--davidsyoung.glm-5.3-exl3-tr3-3.42bpw | quant | model--zai-org.glm-5.3 | davidsyoung GLM-5.3 EXL3 TR3 3.42bpw (routed experts trellis, TP4 rank-sharded) | {
"handle": "davidsyoung",
"is_registry_maintainer": false,
"name": "davidsyoung",
"role": "quantizer",
"url": "https://huggingface.co/davidsyoung"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "78 tensors: model.layers.N.self_attn.o_proj.weight. read from davidsyoung/GLM-5.3-EXL3-TR3-3.42bpw@99c6f951333d2b38f1efefa533c7afadf0d376e3 model.safetensors.inde... | attn.o=native:bf16@16|attn.other=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:exl3-mcg@3.421875|moe.router=native:mixed|moe.shared_expert=native:bf16@16|mtp=quantized:mixed|norm=nativ... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision 99c6f951333d; 81 shards; config.json sha256 8f0b3a4f... carries a leftover ModelOpt quantization_config (quant_method modelopt) and the artifact's real declaration in hybrid_tr3_tail: format exl3-trellis, codebook mcg, tp 4, bits_avg 3.421875, k_values [3, 4], exlla... | {
"config_sha256": "8f0b3a4febb06915c206895e5a1b321fabf1f7eb8fcc17e9a153aadcc5e50d17",
"container": "exl3",
"index_sha256": "787b45ebd8a771097cee0d594b868f51c79336570d6cb76e13275d4691cd3131",
"precision_label": "3.42bpw",
"shard_count": 81,
"size_basis": "repo_weight_files",
"size_bytes": 355150499456,
... |
{
"status": "public",
"uri": "https://huggingface.co/drowzeys/keys-GLM-5.3-EXL3"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 3,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": true
},
"family": "exl3-trellis",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revi... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "Read from bytes: the config names ONE codebook (mul1) but the checkpoint carries two -- layer 3's 768 routed-expert modules are mcg-coded and layers 4-77's 56,832 are mul1-coded, all at K=3 -- so scope.assignments carries one moe.expert... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "drowzeys/keys-GLM-5.3-EXL3",
"revision": "ebf3c8bb0ed869b8f96a6ade9c8d365a49bdbad5",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/drowzeys/keys-GLM-5.3-EXL3"
} | artifact--drowzeys.keys-glm-5.3-exl3 | quant | model--zai-org.glm-5.3 | drowzeys keys-GLM-5.3-EXL3 (routed experts trellis 3.0 bpw, mcg/mul1) | {
"handle": "drowzeys",
"is_registry_maintainer": false,
"name": "drowzeys",
"role": "quantizer",
"url": "https://huggingface.co/drowzeys"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 8,
"format": "fp8_e4m3",
"layer_range": "all",
"note": "stored fp16; values bitwise equal to fp16(dequantize_block_fp8(zai-org/GLM-5.3@187fb9ff)) on the sampled rows (10 tensors; 576 rows each; 0 differing elements;... | attn.o=quantized:fp8_e4m3@8|attn.other=native:mixed|attn.qkv=quantized:fp8_e4m3@8|embed_tokens=native:bf16@16|lm_head=native:fp16@16|mlp.down=quantized:fp8_e4m3@8|mlp.gate=quantized:fp8_e4m3@8|mlp.up=quantized:fp8_e4m3@8|moe.experts=quantized:exl3-mcg@3|moe.experts=quantized:exl3-mul1@3|moe.router=native:mixed|moe.shar... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision ebf3c8bb0ed8; 41 shards; config.json sha256 dfdb2296... declares quantization_config {quant_method exl3, bits 3, codebook mul1, head_bits 16, version 1.4.5, calibration rows 250 x cols 2048}; index sha256 af2c20bd...",
"uri": "https://huggingface.co/drowzeys/key... | {
"config_sha256": "dfdb2296ce53761a807732700df9bae2a830ebbf1e81b95c67f6949d15bcddbd",
"container": "exl3",
"index_sha256": "af2c20bd55835c09e869b6020a6d5ba452e4f06cedde98470981e64444a84ea2",
"precision_label": "3bpw",
"shard_count": 41,
"size_basis": "repo_weight_files",
"size_bytes": 330083954256,
"si... |
{
"status": "public",
"uri": "https://huggingface.co/gittensor-model-hub/Qwen3.8-27B-NVFP4-RTX5090"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "nvfp4",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": ... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
},
{... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "gittensor-model-hub/Qwen3.8-27B-NVFP4-RTX5090",
"revision": "69274a0d8dff5dd35bcee8290612f71e03b6e981",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/gittensor-model-hub/Qwen3.8-27B-NVF... | artifact--gittensor-model-hub.qwen3.8-27b-nvfp4-rtx5090 | quant | model--qwen.qwen3.8-27b | gittensor-model-hub Qwen3.8-27B NVFP4 (RTX5090) | {
"handle": "gittensor-model-hub",
"is_registry_maintainer": false,
"name": "gittensor-model-hub",
"role": "quantizer",
"url": "https://huggingface.co/gittensor-model-hub"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
... | attn.o=unknown:unknown|attn.qkv=unknown:unknown|embed_tokens=unknown:unknown|lm_head=unknown:unknown|mlp.down=quantized:nvfp4@4|mlp.gate=quantized:nvfp4@4|mlp.up=quantized:nvfp4@4|moe.experts=quantized:nvfp4@4|head=unknown|kv=bf16 | {
"sealed": false
} | [
{
"kind": "github_file",
"note": "3 candidate shard sha256 values recorded",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/kld5-1M-gt5090.json"
}
] | {
"container": "safetensors",
"precision_label": "NVFP4",
"size_basis": "repo_all_files",
"size_bytes": 20616833355,
"size_gb": 20.616833355
} |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/Qwen3.8-27B-EXL3-K5K6-hydrated"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "no_known_deviations",
"detail": "Declared by the submission receipt.",
"severity": "info"
}
] | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/Qwen3.8-27B-EXL3-K5K6-hydrated",
"revision": "853acef0b24961b269cdcf32b1ebb405649b545b",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/Qwen3.8-27B-EXL3-K5K6-hydrated"
... | artifact--hf-jobs.fbc43bce256063219c09ce70 | quant | model--qwen.qwen3.8-27b.1d4bf0f2ff60 | malaiwah/Qwen3.8-27B-EXL3-K5K6-hydrated | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "quantizer",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"note": "One BF16 text embedding tensor retained in the artifact. Scope derives from malaiwah/Qwen3.8-27B-EXL3-K5K6-hydrated@853acef0b24961b269cdcf32b1ebb405649b545b quantization_manifest.json (sha25... | attn.o=quantized:exl3-mcg@6|attn.other=quantized:mixed|attn.qkv=quantized:exl3-mcg@6|embed_tokens=native:bf16@16|lm_head=quantized:exl3-mcg@6|mlp.down=quantized:exl3-mcg@6|mlp.gate=quantized:exl3-mcg@5|mlp.up=quantized:exl3-mcg@5|mtp=quantized:mixed|norm=native:bf16@16|other=native:bf16@16|head=quantized|kv=not_applica... | {
"sealed": false
} | [
{
"kind": "url",
"uri": "https://huggingface.co/malaiwah/Qwen3.8-27B-EXL3-K5K6-hydrated"
},
{
"kind": "hf_file",
"note": "Original producing plan binds the exact serialized weight-file census and its size basis.",
"sha256": "acd80d45221beef03f16b7639c550dc35880d8bebd12944e4a369728dd63f1ca",
... | {
"config_sha256": "fbb105334da6554c10784ff1257fda5e3821d4d5426d64469cee2b2ad67ba2b3",
"container": "safetensors",
"index_sha256": null,
"precision_label": "exl3-mcg",
"shard_count": 3,
"shard_sha256": {
"model-00001-of-00003.safetensors": "792b46b54a79871030a5a029a483b8938ff509912169a1c7627913d614365f0... |
{
"status": "public",
"uri": "https://huggingface.co/incoai/GLM-5.3-NVFP4"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "nvfp4",
"group_size": 16,
"quantizer": {
"pipeline_ref": null,
"revision": nu... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "Read from bytes (the index plus a ranged read of every shard header): the 57,600 routed-expert matrices of layers 3-77 are NVFP4 modelopt component sets -- e2m1 weight U8 [out, in/2], weight_scale F8_E4M3 [out, in/16] (group 16 along th... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "incoai/GLM-5.3-NVFP4",
"revision": "54e52520606f96b3d9fc84088ad22882a61648ac",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/incoai/GLM-5.3-NVFP4"
} | artifact--incoai.glm-5.3-nvfp4 | quant | model--zai-org.glm-5.3 | incoai GLM-5.3-NVFP4 (routed experts NVFP4 e2m1 group 16, rest native) | {
"handle": "incoai",
"is_registry_maintainer": false,
"name": "incoai",
"role": "quantizer",
"url": "https://huggingface.co/incoai"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "78 tensors: model.layers.N.self_attn.o_proj.weight. read from incoai/GLM-5.3-NVFP4@54e52520606f96b3d9fc84088ad22882a61648ac model.safetensors.index.json + every s... | attn.o=native:bf16@16|attn.other=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:nvfp4@4|moe.router=native:mixed|moe.shared_expert=native:bf16@16|mtp=native:mixed|norm=native:bf16@16|hea... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision 54e52520606f; 87 shards; config.json sha256 48025a5f... declares quantization_config quant_method modelopt / quant_algo NVFP4 (producer modelopt 0.45.0); index sha256 54c4fc5d...",
"uri": "https://huggingface.co/incoai/GLM-5.3-NVFP4"
},
{
"kind": "github... | {
"config_sha256": "48025a5f559f7a747734be8499810a22d34330577e1566a28810466319f57ef4",
"container": "safetensors",
"index_sha256": "54c4fc5dc9e59691e1797e59f8d57ef33281457cbd47fc0f3877ac95e5ce736d",
"precision_label": "NVFP4",
"shard_count": 87,
"size_basis": "repo_weight_files",
"size_bytes": 46482287291... |
{
"status": "public",
"uri": "https://huggingface.co/Inferact/GLM-5.3-NVFP4"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "nvfp4",
"group_size": 16,
"quantizer": {
"pipeline_ref": null,
"revision": nu... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "Read from bytes (the index plus a ranged read of every shard header): the 57,600 routed-expert matrices of layers 3-77 are NVFP4 modelopt component sets -- e2m1 weight U8 [out, in/2], weight_scale F8_E4M3 [out, in/16] (group 16 along th... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "Inferact/GLM-5.3-NVFP4",
"revision": "ce67b36f3669192b5bb233819f0fda6c8a9837f8",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/Inferact/GLM-5.3-NVFP4"
} | artifact--inferact.glm-5.3-nvfp4 | quant | model--zai-org.glm-5.3 | Inferact GLM-5.3-NVFP4 (routed experts NVFP4 e2m1 group 16, rest native) | {
"handle": "Inferact",
"is_registry_maintainer": false,
"name": "Inferact",
"role": "quantizer",
"url": "https://huggingface.co/Inferact"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "78 tensors: model.layers.N.self_attn.o_proj.weight. read from Inferact/GLM-5.3-NVFP4@ce67b36f3669192b5bb233819f0fda6c8a9837f8 model.safetensors.index.json + every... | attn.o=native:bf16@16|attn.other=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:nvfp4@4|moe.router=native:bf16@16|moe.shared_expert=native:bf16@16|mtp=native:mixed|norm=native:bf16@16|h... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision ce67b36f3669; 88 shards; config.json sha256 44ffebdc... declares quantization_config quant_method modelopt / quant_algo NVFP4 (producer undeclared (quantization_config names quant_method modelopt and group_size 16 but no producer version)); index sha256 e5de21cf..."... | {
"config_sha256": "44ffebdc4bf4f99b74932908896da5ba1e09b88a15edebf8d065232669193d9c",
"container": "safetensors",
"index_sha256": "e5de21cffb3ec5f958646ac6923cdd6a76b39ea0cb4b31e7627fb14c561cd42f",
"precision_label": "NVFP4",
"shard_count": 88,
"size_basis": "repo_weight_files",
"size_bytes": 46482283244... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/deepseek-v4-tiny-random-bf16/tree/185723bec10ee57db86e8c269b3c6a811122a0ca"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | null | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/deepseek-v4-tiny-random-bf16",
"revision": "185723bec10ee57db86e8c269b3c6a811122a0ca",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/deepseek-v4-tiny-random-bf16"
} | artifact--malaiwah.deepseek-v4-tiny-random-bf16.185723bec10e | base | model--malaiwah.deepseek-v4-tiny-random-bf16 | deepseek-v4-tiny-random-bf16 (random test fixture) native BF16 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "9b939c6baec8fd90421bc4310ca02fba34c4eef3939a380c98f382b74b6d821f",
"uri": "https://huggingface.co/datasets/malaiwah/deepseek-v4-tiny-fidelity-root-v1/resolve/c86180f1844f932b2e685eeeed96b633e4c54cfb/fidelity-dataset.json"
},
{
"kind": "hf_file",
"sha256": "291... | {
"config_sha256": "93e0925c4eb97d80a1f5832fe957cd101cd106610a3bc9650fc37679037ae111",
"container": "safetensors",
"index_sha256": null,
"precision_label": "BF16",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "07c036238e208cf1e51b430e315592c8161bafc2ebd5c7b73ae0975f21a72558"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/GLM-5.2-SIQ-Fruit-bf16"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Every tensor is bf16 and comes from the trained checkpoint by a direct cast: export_fruit.py FRUIT_BF16=1 reads the annealed state dict and writes .to(torch.bfloat16) bits, and its `continue` skips the SI... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/GLM-5.2-SIQ-Fruit-bf16",
"revision": "ef68013aa6e16453cf52b5b77647f72fbe258c3c",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/malaiwah/GLM-5.2-SIQ-Fruit-bf16"
} | artifact--malaiwah.glm-5.2-siq-fruit-bf16 | base | model--malaiwah.glm-5.2-siq-fruit | GLM-5.2-SIQ-Fruit BF16 (the reference export) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.other=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|moe.router=native:bf16@16|moe.shared_expert=native:bf16@16|mtp=native:bf16@16|norm=native:bf16@16|he... | null | [
{
"kind": "model_card",
"uri": "https://huggingface.co/malaiwah/GLM-5.2-SIQ-Fruit-bf16"
},
{
"kind": "github_file",
"note": "pinned to the commit so the line numbers cannot drift: the unconditional bf16() helper at 262-266 (`sd[key].to(torch.bfloat16)`), the FRUIT_BF16=1 routed-expert branch at ... | {
"config_sha256": "5a19697e555fff140d1b089b852c3ef227114b196f8d76796560feeeb34dc44a",
"container": "safetensors",
"index_sha256": "86e6cc1d8548c7bdbbc117e93b85b8ae249f446de9b48d2195e51f358674ba56",
"precision_label": "BF16",
"size_basis": "repo_all_files",
"size_bytes": 10102776813,
"size_gb": 10.1027768... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/GLM-5.2-SIQ-Fruit"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 3.375,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-trellis",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--malaiwah.glm-5.2-siq-fruit-bf16 | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "declared_scheme_mismatch",
"detail": "config.json declares quantization_config quant_method=modelopt, quant_algo=NVFP4, group_size 16, W4A4, producer b300-exl3-modelopt-dispatch-shim. That block does not describe the stored bytes... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/GLM-5.2-SIQ-Fruit",
"revision": "c1798e3676fa16b4a874381171adab1e3033fbd5",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/malaiwah/GLM-5.2-SIQ-Fruit"
} | artifact--malaiwah.glm-5.2-siq-fruit.exl3-k3k4 | quant | model--malaiwah.glm-5.2-siq-fruit | GLM-5.2-SIQ-Fruit (exl3-trellis K3/K4 routed experts) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "quantizer",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "0-13",
"tensor_cl... | attn.o=native:bf16@16|attn.other=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:exl3-trellis@3.375|moe.router=native:bf16@16|moe.shared_expert=native:bf16@16|mtp=quantized:exl3-trellis@... | null | [
{
"kind": "model_card",
"uri": "https://huggingface.co/malaiwah/GLM-5.2-SIQ-Fruit"
},
{
"kind": "hf_file",
"note": "per-expert K allocation and the encoder's own expert_rel_rt_mse; keep_nvfp4 is an empty list for every layer",
"uri": "https://huggingface.co/malaiwah/GLM-5.2-SIQ-Fruit/blob/c1... | {
"config_sha256": "7df3d68ab252ffa0bff636d00f82330f56939f2808915eb7d2d209c98e0b9753",
"container": "safetensors",
"index_sha256": "5808a4b3e75c4a949a1ede42e6c6fb2576089ec1544038b77de24076e99bf3da",
"precision_label": "EXL3 K3/K4 experts, BF16 elsewhere",
"size_basis": "repo_all_files",
"size_bytes": 312552... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/GLM-5.3-Flash-TR3-6bpw"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 6,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-bf16.a6c167b6 | [
{
"affects_comparability": false,
"code": "scope_record_corrected",
"detail": "Superseded record: this artifact's scope previously read attn.qkv / attn.o / mlp.{gate,up,down} as quantized:exl3-mcg at the nominal rate. That was wrong. The release is routed-experts-only: its own config.json declares scope... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/GLM-5.3-Flash-TR3-6bpw",
"revision": null,
"revision_source": "none",
"status": "known",
"url": "https://huggingface.co/malaiwah/GLM-5.3-Flash-TR3-6bpw"
} | artifact--malaiwah.glm-5.3-flash-tr3-6bpw | quant | model--zai-org.glm-5.3-flash | malaiwah GLM-5.3-Flash TR3 6bpw (K6) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "quantizer",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "routed-experts-only scope: not quantized",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"form... | attn.o=native:bf16@16|attn.other=native:mixed|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:exl3-mcg@6|moe.router=native:fp32@32|moe.shared_expert=native:bf16@16|mtp=quantized:exl3-mcg@6|norm=native:... | {
"note": "reader ABI 3d659542..., runtime reader 1ccce446..., reader audit receipt c986a0a9...",
"receipts": [
{
"kind": "hf_file",
"sha256": "19766e5e9643dbe940c05deaee7c3085f9ee339553da35ead973c825adddfef2",
"uri": "https://huggingface.co/malaiwah/GLM-5.3-Flash-TR3-6bpw/blob/main/receipts/k... | [
{
"kind": "hf_file",
"note": "output_logical_bytes 253,536,370,680 and 120 shard sha256 values",
"uri": "https://huggingface.co/malaiwah/GLM-5.3-Flash-TR3-6bpw/blob/main/receipts/materialization-receipt.json"
},
{
"kind": "hf_file",
"note": "student_checkpoint_identity_sha256 a8668be3...",
... | {
"container": "exl3",
"precision_label": "6bpw",
"shard_count": 120,
"size_basis": "tensor_payload",
"size_bytes": 253536370680,
"size_gb": 253.53637068
} |
{
"status": "private",
"uri": null
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 8,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-bf16.a6c167b6 | [
{
"affects_comparability": false,
"code": "scope_record_corrected",
"detail": "Superseded record: this artifact's scope previously read attn.qkv / attn.o / mlp.{gate,up,down} as quantized:exl3-mcg at the nominal rate. That was wrong. The release is routed-experts-only: its own config.json declares scope... | {
"link_type": "repository",
"path": null,
"reason": "HTTP 401 unauthenticated at seeding time: private, or not yet created.",
"repository": "malaiwah/GLM-5.3-Flash-TR3-8bpw",
"revision": null,
"revision_source": "none",
"status": "unavailable",
"url": "https://huggingface.co/malaiwah/GLM-5.3-Flash-TR3-... | artifact--malaiwah.glm-5.3-flash-tr3-8bpw | quant | model--zai-org.glm-5.3-flash | malaiwah GLM-5.3-Flash TR3 8bpw (K8) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "quantizer",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "routed-experts-only scope: not quantized",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"form... | attn.o=native:bf16@16|attn.other=native:mixed|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:exl3-mcg@8|moe.router=native:fp32@32|moe.shared_expert=native:bf16@16|mtp=quantized:exl3-mcg@8|norm=native:... | {
"sealed": false
} | [
{
"kind": "private_communication",
"note": "materialization facts: output_logical_bytes 331,449,761,784; 37,152 routed choices; 1,618 native (non-routed) checkpoint tensors; bits 8; qualified_tp_sizes []",
"uri": "operator inventory, 2026-08-28"
},
{
"kind": "receipt_file",
"note": "malaiwah... | {
"container": "exl3",
"precision_label": "8bpw",
"size_basis": "tensor_payload",
"size_bytes": 331449761784,
"size_gb": 331.449761784,
"tensor_parallel": {
"pre_sliced": false,
"world_size": null
}
} |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-random-bf16/tree/a1a973330e4e912678adbd8ea41e8b86b1c9dccb"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | null | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/glm-moe-dsa-tiny-random-bf16",
"revision": "a1a973330e4e912678adbd8ea41e8b86b1c9dccb",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-random-bf16"
} | artifact--malaiwah.glm-moe-dsa-tiny-random-bf16.a1a973330e4e | base | model--malaiwah.glm-moe-dsa-tiny-random-bf16 | glm-moe-dsa-tiny-random-bf16 (random test fixture) native BF16 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "e2114cde06393fa64d22bdc1c9f126fd1d58c06a88c58461120a0729aec60f0e",
"uri": "https://huggingface.co/datasets/malaiwah/glm-moe-dsa-tiny-fidelity-root-v1/resolve/b8ed427805abdfd6b8b20db10ca5107e840d0497/fidelity-dataset.json"
},
{
"kind": "hf_file",
"sha256": "4e5... | {
"config_sha256": "7d5880677655ece868c12425a204f539243244af121bc8ad09b881a7546930d7",
"container": "safetensors",
"index_sha256": null,
"precision_label": "BF16",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "9efe50a1685e157c1b99c3824094cf4fe269c7d332bdd0006145264949c5cec1"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-random-ct-mxfp4-rtn-format/tree/4f3287b3b0d9a34b1d2b56eaf15273a36ed2f0a0"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "mxfp4",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": ... | null | artifact--malaiwah.glm-moe-dsa-tiny-random-bf16.a1a973330e4e | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/glm-moe-dsa-tiny-random-ct-mxfp4-rtn-format",
"revision": "4f3287b3b0d9a34b1d2b56eaf15273a36ed2f0a0",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-ra... | artifact--malaiwah.glm-moe-dsa-tiny-random-ct-mxfp4-rtn-format.4f3287b3b0d9 | quant | model--malaiwah.glm-moe-dsa-tiny-random-bf16 | glm-moe-dsa-tiny-random-ct-mxfp4-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 4,
"format": "mxfp4",
"layer_range": "all",
"note": "4 tensors: model.layers.N.self_attn.o_proj.weight. FORMAT fixture CPU RTN; reconstructed weights only; activation-not-captured, no optimizer/native-kernel quality... | attn.o=quantized:mxfp4@4|attn.other=quantized:mixed|attn.qkv=quantized:mixed|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:mxfp4@4|mlp.gate=quantized:mxfp4@4|mlp.up=quantized:mxfp4@4|moe.experts=quantized:mxfp4@4|moe.router=native:mixed|moe.shared_expert=quantized:mxfp4@4|norm=native:bf16@16|hea... | null | [
{
"kind": "hf_file",
"sha256": "5bbdb501d75aa389e877a6f656378b57d3778585e2c666ab88e2eef3a5391434",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-microfloat-tiny-cpu-format-v1/resolve/b1690c1b72beafc812f2286ec6a8fe2ced338edc/raw/candidates/ct-mxfp4/evidence/candidate/fidelity-dataset.json"
},
... | {
"config_sha256": "c6cc76ca973555cae99d0c42da0a9385bf1ac67c67f4748cac0b4c0778ff5d90",
"container": "safetensors",
"precision_label": "ct-mxfp4; CPU RTN format fixture",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "9222707b17727c48479eb37416d96289344b48d70eee2d919b282063ccee3b3d"
},
"size... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-random-ct-nvfp4-rtn-format/tree/f6056d80c8a421eb2f347b1f9837634c19c09e7f"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "nvfp4",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": ... | null | artifact--malaiwah.glm-moe-dsa-tiny-random-bf16.a1a973330e4e | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/glm-moe-dsa-tiny-random-ct-nvfp4-rtn-format",
"revision": "f6056d80c8a421eb2f347b1f9837634c19c09e7f",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-ra... | artifact--malaiwah.glm-moe-dsa-tiny-random-ct-nvfp4-rtn-format.f6056d80c8a4 | quant | model--malaiwah.glm-moe-dsa-tiny-random-bf16 | glm-moe-dsa-tiny-random-ct-nvfp4-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 4,
"format": "nvfp4",
"layer_range": "all",
"note": "4 tensors: model.layers.N.self_attn.o_proj.weight. FORMAT fixture CPU RTN; reconstructed weights only; activation-not-captured, no optimizer/native-kernel quality... | attn.o=quantized:nvfp4@4|attn.other=quantized:mixed|attn.qkv=quantized:nvfp4@4|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:nvfp4@4|mlp.gate=quantized:nvfp4@4|mlp.up=quantized:nvfp4@4|moe.experts=quantized:nvfp4@4|moe.router=native:mixed|moe.shared_expert=quantized:nvfp4@4|norm=native:bf16@16|h... | null | [
{
"kind": "hf_file",
"sha256": "af50f36d643b6c30fca32c369518d95a60f3a3773089939a57da5a9715ae7b3b",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-microfloat-tiny-cpu-format-v1/resolve/b1690c1b72beafc812f2286ec6a8fe2ced338edc/raw/candidates/ct-nvfp4/evidence/candidate/fidelity-dataset.json"
},
... | {
"config_sha256": "188450fd2bed8cf0c6e06d785410192143b3777bcbbff315151fcecb96cd267e",
"container": "safetensors",
"precision_label": "ct-nvfp4; CPU RTN format fixture",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "640f3d7ea214015d8834a94b56b2dee2c2326dc50245cdd0ded5778799b0d300"
},
"size... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-random-fp8-block32-rtn-format/tree/6ada7f419a5a5cca2ad651f140e13d21a8057a7b"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 8,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "fp8_e4m3",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | null | artifact--malaiwah.glm-moe-dsa-tiny-random-bf16.a1a973330e4e | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/glm-moe-dsa-tiny-random-fp8-block32-rtn-format",
"revision": "6ada7f419a5a5cca2ad651f140e13d21a8057a7b",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny... | artifact--malaiwah.glm-moe-dsa-tiny-random-fp8-block32-rtn-format.6ada7f419a5a | quant | model--malaiwah.glm-moe-dsa-tiny-random-bf16 | glm-moe-dsa-tiny-random-fp8-block32-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 8,
"format": "fp8_e4m3",
"layer_range": "all",
"note": "4 tensors: model.layers.N.self_attn.o_proj.weight. Measured payload dtype and scale presence; 32x32 blocks with partial tails. Native BF16 forward of decoded w... | attn.o=quantized:fp8_e4m3@8|attn.other=quantized:mixed|attn.qkv=quantized:fp8_e4m3@8|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:fp8_e4m3@8|mlp.gate=quantized:fp8_e4m3@8|mlp.up=quantized:fp8_e4m3@8|moe.experts=quantized:fp8_e4m3@8|moe.router=native:mixed|moe.shared_expert=quantized:fp8_e4m3@8|... | null | [
{
"kind": "hf_file",
"sha256": "8d7fe7500bf76b84f13b5f3de6b0a84428339ecfe9ebc66c68e7eed49aa9540c",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-existing-tiny-cpu-format-v1/resolve/2b5c947281a92d1f104ee17fdcc44a0104768a45/raw/candidates/fp8-block32/evidence/fp8-block32/fidelity-dataset.json"
},... | {
"config_sha256": "4f89c2d6f5698cf2ba2ff0add80273607770a708e23abbe0bafdde8db066475e",
"container": "safetensors",
"precision_label": "fp8-block32; random FORMAT fixture",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "a51f06349335c31650be9f28e05a1dcf1bc72b00e9b124fe01d88bc65c3a2652"
},
"si... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-random-fp8-ue8m0-rtn-format/tree/b35a3365ea79b5b5e72f778819ee2dc7610975ec"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 8,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "fp8_e4m3",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | null | artifact--malaiwah.glm-moe-dsa-tiny-random-bf16.a1a973330e4e | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/glm-moe-dsa-tiny-random-fp8-ue8m0-rtn-format",
"revision": "b35a3365ea79b5b5e72f778819ee2dc7610975ec",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-r... | artifact--malaiwah.glm-moe-dsa-tiny-random-fp8-ue8m0-rtn-format.b35a3365ea79 | quant | model--malaiwah.glm-moe-dsa-tiny-random-bf16 | glm-moe-dsa-tiny-random-fp8-ue8m0-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 8,
"format": "fp8_e4m3",
"layer_range": "all",
"note": "4 tensors: model.layers.N.self_attn.o_proj.weight. FORMAT fixture CPU RTN; reconstructed weights only; activation-not-captured, no optimizer/native-kernel qual... | attn.o=quantized:fp8_e4m3@8|attn.other=quantized:mixed|attn.qkv=quantized:fp8_e4m3@8|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:fp8_e4m3@8|mlp.gate=quantized:fp8_e4m3@8|mlp.up=quantized:fp8_e4m3@8|moe.experts=quantized:fp8_e4m3@8|moe.router=native:mixed|moe.shared_expert=quantized:fp8_e4m3@8|... | null | [
{
"kind": "hf_file",
"sha256": "b5890b2985844ad14a041b49eadde3210d2b8829435175d53c682e4a54d811c7",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-microfloat-tiny-cpu-format-v1/resolve/b1690c1b72beafc812f2286ec6a8fe2ced338edc/raw/candidates/fp8-ue8m0/evidence/candidate/fidelity-dataset.json"
},
... | {
"config_sha256": "cbf3fcc7cdf78aec98f645c6920345f2fdebde6764223349249f6389606229ec",
"container": "safetensors",
"precision_label": "fp8-ue8m0; CPU RTN format fixture",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "e26433482b4d0c4e0cf15cd313cfaf140708378cce3e73866bb8badc8298a159"
},
"siz... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-random-modelopt-mixed-rtn-format/tree/dc77e2001b92d4f610f7091cfe06b13e5545512e"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "mixed",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": ... | null | artifact--malaiwah.glm-moe-dsa-tiny-random-bf16.a1a973330e4e | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/glm-moe-dsa-tiny-random-modelopt-mixed-rtn-format",
"revision": "dc77e2001b92d4f610f7091cfe06b13e5545512e",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/glm-moe-dsa-t... | artifact--malaiwah.glm-moe-dsa-tiny-random-modelopt-mixed-rtn-format.dc77e2001b92 | quant | model--malaiwah.glm-moe-dsa-tiny-random-bf16 | glm-moe-dsa-tiny-random-modelopt-mixed-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 8,
"format": "fp8_e4m3",
"layer_range": "0,3",
"note": "2 tensors: model.layers.N.self_attn.o_proj.weight. FORMAT fixture CPU RTN; reconstructed weights only; activation-not-captured, no optimizer/native-kernel qual... | attn.o=quantized:fp8_e4m3@8|attn.o=quantized:fp8_e4m3@8|attn.o=quantized:nvfp4@4|attn.other=quantized:mixed|attn.qkv=quantized:mixed|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:nvfp4@4|mlp.gate=quantized:fp8_e4m3@8|mlp.up=quantized:fp8_e4m3@8|moe.experts=quantized:mixed|moe.router=native:mixed... | null | [
{
"kind": "hf_file",
"sha256": "414494a9ef12eb8e2ffb18e3fe4280d83594ca4682e47262b29fb0aedaf47a2d",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-microfloat-tiny-cpu-format-v1/resolve/b1690c1b72beafc812f2286ec6a8fe2ced338edc/raw/candidates/modelopt-mixed/evidence/candidate/fidelity-dataset.json"
... | {
"config_sha256": "33f139cf6a6c6de8e40791cff02405adc77c4447419863d9ce801642a484e958",
"container": "safetensors",
"precision_label": "modelopt-mixed; CPU RTN format fixture",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "9a5c06dffc5c8c08aa0cb0109c87274d044d8297db132405828d914762370199"
},
... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-random-modelopt-nvfp4-rtn-format/tree/c795b911ff87882db3b5f0e8c795258e6c1f1d5b"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "nvfp4",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": ... | null | artifact--malaiwah.glm-moe-dsa-tiny-random-bf16.a1a973330e4e | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/glm-moe-dsa-tiny-random-modelopt-nvfp4-rtn-format",
"revision": "c795b911ff87882db3b5f0e8c795258e6c1f1d5b",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/glm-moe-dsa-t... | artifact--malaiwah.glm-moe-dsa-tiny-random-modelopt-nvfp4-rtn-format.c795b911ff87 | quant | model--malaiwah.glm-moe-dsa-tiny-random-bf16 | glm-moe-dsa-tiny-random-modelopt-nvfp4-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 4,
"format": "nvfp4",
"layer_range": "all",
"note": "4 tensors: model.layers.N.self_attn.o_proj.weight. FORMAT fixture CPU RTN; reconstructed weights only; activation-not-captured, no optimizer/native-kernel quality... | attn.o=quantized:nvfp4@4|attn.other=quantized:mixed|attn.qkv=quantized:nvfp4@4|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:nvfp4@4|mlp.gate=quantized:nvfp4@4|mlp.up=quantized:nvfp4@4|moe.experts=quantized:nvfp4@4|moe.router=native:mixed|moe.shared_expert=quantized:nvfp4@4|norm=native:bf16@16|h... | null | [
{
"kind": "hf_file",
"sha256": "630645250806ce64351238c0ba278e9495ff81b22bd1372ca3ea5b8e4241aa53",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-microfloat-tiny-cpu-format-v1/resolve/b1690c1b72beafc812f2286ec6a8fe2ced338edc/raw/candidates/modelopt-nvfp4/evidence/candidate/fidelity-dataset.json"
... | {
"config_sha256": "932f17d4ee472ee814a296cfa3b521bf0b9daefc57cef0f3406995fdb341e7fe",
"container": "safetensors",
"precision_label": "modelopt-nvfp4; CPU RTN format fixture",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "07cf6fbb7ad67852d9fc66c58d94e7d9e09dafc04f1358523a80cd1478c58004"
},
... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-random-q4_0-rtn-format/tree/c0e73b84e074c800ea906f46d1d9cf3d04bcf88c"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "gguf-k-quant",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revi... | null | artifact--malaiwah.glm-moe-dsa-tiny-random-bf16.a1a973330e4e | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/glm-moe-dsa-tiny-random-q4_0-rtn-format",
"revision": "c0e73b84e074c800ea906f46d1d9cf3d04bcf88c",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-random... | artifact--malaiwah.glm-moe-dsa-tiny-random-q4-0-rtn-format.c0e73b84e074 | quant | model--malaiwah.glm-moe-dsa-tiny-random-bf16 | glm-moe-dsa-tiny-random-q4_0-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 4,
"format": "gguf-k-quant",
"layer_range": "all",
"note": "the attention output projection. ggml types: Q4_0 x4. MEASURED 4.5000 bits/weight over 16384 weights.",
"tensor_class": "attn.o",
"treatment": ... | attn.o=quantized:gguf-k-quant@4|attn.other=quantized:mixed|attn.qkv=quantized:mixed|embed_tokens=quantized:gguf-k-quant@4|lm_head=quantized:gguf-k-quant@4|mlp.down=quantized:gguf-k-quant@4|mlp.gate=quantized:gguf-k-quant@4|mlp.up=quantized:gguf-k-quant@4|moe.experts=quantized:gguf-k-quant@4|moe.router=quantized:mixed|m... | null | [
{
"kind": "hf_file",
"sha256": "bd4f505fb44c110bdecf19f1f3e148c56b7bc562f2ddc366fdf51659751277ab",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-existing-tiny-cpu-format-v1/resolve/2b5c947281a92d1f104ee17fdcc44a0104768a45/raw/candidates/q4_0/evidence/q4_0/fidelity-dataset.json"
},
{
"kind... | {
"config_sha256": "7d5880677655ece868c12425a204f539243244af121bc8ad09b881a7546930d7",
"container": "gguf",
"precision_label": "q4_0; random FORMAT fixture",
"shard_count": 1,
"shard_sha256": {
"glm-moe-dsa-tiny-Q4_0.gguf": "6ef2e650af5d019e5fec726627a4900fc8cea7341e49136a11113d4d5016576b"
},
"size_ba... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-random-q8_0-rtn-format/tree/057550276019fa2ee503d175c415a093f07f0ba9"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 8,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "gguf-k-quant",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revi... | null | artifact--malaiwah.glm-moe-dsa-tiny-random-bf16.a1a973330e4e | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/glm-moe-dsa-tiny-random-q8_0-rtn-format",
"revision": "057550276019fa2ee503d175c415a093f07f0ba9",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/glm-moe-dsa-tiny-random... | artifact--malaiwah.glm-moe-dsa-tiny-random-q8-0-rtn-format.057550276019 | quant | model--malaiwah.glm-moe-dsa-tiny-random-bf16 | glm-moe-dsa-tiny-random-q8_0-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 8,
"format": "gguf-k-quant",
"layer_range": "all",
"note": "the attention output projection. ggml types: Q8_0 x4. MEASURED 8.5000 bits/weight over 16384 weights.",
"tensor_class": "attn.o",
"treatment": ... | attn.o=quantized:gguf-k-quant@8|attn.other=quantized:mixed|attn.qkv=quantized:mixed|embed_tokens=quantized:gguf-k-quant@8|lm_head=quantized:gguf-k-quant@8|mlp.down=quantized:gguf-k-quant@8|mlp.gate=quantized:gguf-k-quant@8|mlp.up=quantized:gguf-k-quant@8|moe.experts=quantized:gguf-k-quant@8|moe.router=quantized:mixed|m... | null | [
{
"kind": "hf_file",
"sha256": "a2edbe12bc4ad047d478d318cc66f92a37738dfed000804fb1d656116494a4c4",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-existing-tiny-cpu-format-v1/resolve/2b5c947281a92d1f104ee17fdcc44a0104768a45/raw/candidates/q8_0/evidence/q8_0/fidelity-dataset.json"
},
{
"kind... | {
"config_sha256": "7d5880677655ece868c12425a204f539243244af121bc8ad09b881a7546930d7",
"container": "gguf",
"precision_label": "q8_0; random FORMAT fixture",
"shard_count": 1,
"shard_sha256": {
"glm-moe-dsa-tiny-Q8_0.gguf": "d8c8229aae6f3481ac59ebb567a4e9fcfb63df5b7f9cdf09c440219b92a8be3e"
},
"size_ba... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/glm5-next-tiny-random-bf16/tree/4c31348e3beb1a8a6bd73d055464362953d768ea"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | null | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/glm5-next-tiny-random-bf16",
"revision": "4c31348e3beb1a8a6bd73d055464362953d768ea",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/glm5-next-tiny-random-bf16"
} | artifact--malaiwah.glm5-next-tiny-random-bf16.4c31348e3beb | base | model--malaiwah.glm5-next-tiny-random-bf16 | glm5-next-tiny-random-bf16 (random test fixture) native BF16 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "2e2bdbee80e2921d2b1067c59e2ec4e4631d41b9db7dc70655f7881e5df88fa5",
"uri": "https://huggingface.co/datasets/malaiwah/glm5-next-tiny-fidelity-root-v1/resolve/3990bcf5b82667cc359046898a652ee8a84921a7/fidelity-dataset.json"
},
{
"kind": "hf_file",
"sha256": "6c191... | {
"config_sha256": "c1037e7d8c788c45b57360b4930643fac8ca3220fc278c862a2854d257818d46",
"container": "safetensors",
"index_sha256": null,
"precision_label": "BF16",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "c40f4a5cae0c08752ee954f32d3f37f7bfcf77e6facbb238f62de97529783c19"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/k2-horizon-tiny-random-bf16/tree/50627d8f2166ee04a172f4a5144dc0b3385b9ee9"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | null | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/k2-horizon-tiny-random-bf16",
"revision": "50627d8f2166ee04a172f4a5144dc0b3385b9ee9",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/k2-horizon-tiny-random-bf16"
} | artifact--malaiwah.k2-horizon-tiny-random-bf16.50627d8f2166 | base | model--malaiwah.k2-horizon-tiny-random-bf16 | k2-horizon-tiny-random-bf16 (random test fixture) native BF16 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "17ffcb82bbd8a7dadb1994c1b251f65b4c8407fd5f993dabc9b2e59b2c43bdf5",
"uri": "https://huggingface.co/datasets/malaiwah/k2-horizon-tiny-fidelity-root-v1/resolve/9131871ea54fbe9589a05fbb68ed4d26f236ec32/fidelity-dataset.json"
},
{
"kind": "hf_file",
"sha256": "93a1... | {
"config_sha256": "8fa78c70b4e29d824ab8780ef4e2836b329cb54e793a50b5e86e93181bc1257b",
"container": "safetensors",
"index_sha256": null,
"precision_label": "BF16",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "533d5a7b34264d1ccf3e83fe2e9da7e44756075de28bac15f9258a263e0a8ca1"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/kimi-k25-tiny-random-bf16/tree/721c97dd956f23b6b5c1040d633539a162a19094"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | null | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/kimi-k25-tiny-random-bf16",
"revision": "721c97dd956f23b6b5c1040d633539a162a19094",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/kimi-k25-tiny-random-bf16"
} | artifact--malaiwah.kimi-k25-tiny-random-bf16.721c97dd956f | base | model--malaiwah.kimi-k25-tiny-random-bf16 | kimi-k25-tiny-random-bf16 (random test fixture) native BF16 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "a617a9a8a9048c90c63ac1b62bf8d7b036e6c44465e2af8d00af547ccc2b02d9",
"uri": "https://huggingface.co/datasets/malaiwah/kimi-k25-tiny-fidelity-root-v1/resolve/ff643f85eef7c3b5e0f382a0e19e1a0bd06f7382/fidelity-dataset.json"
},
{
"kind": "hf_file",
"sha256": "4fdcf2... | {
"config_sha256": "5a25e66471e703670ef7f69820ca27085742d9931260f9f5de9b3f7efbbe45e7",
"container": "safetensors",
"index_sha256": null,
"precision_label": "BF16",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "54873e1c4b4873d32f956b06436b8afb448e3765ee01984b48eaf0518be1bb05"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/kimi-k3-tiny-random-bf16/tree/3c1e2f22e5ccfe37addc9906bfb0fdea6ad00e16"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | null | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/kimi-k3-tiny-random-bf16",
"revision": "3c1e2f22e5ccfe37addc9906bfb0fdea6ad00e16",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/kimi-k3-tiny-random-bf16"
} | artifact--malaiwah.kimi-k3-tiny-random-bf16.3c1e2f22e5cc | base | model--malaiwah.kimi-k3-tiny-random-bf16 | kimi-k3-tiny-random-bf16 (random test fixture) native BF16 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "396acdda045e4ae8d9400a732d5722913987cf07369c77259b74f063abc877a2",
"uri": "https://huggingface.co/datasets/malaiwah/kimi-k3-tiny-fidelity-root-v1/resolve/74f15b550582c96fb26c40c87a839298b914721b/fidelity-dataset.json"
},
{
"kind": "hf_file",
"sha256": "e8d3ebb... | {
"config_sha256": "064463931db1c0c2d08ef87346bfe1586b9ca7d21a34f6dc050fa4be2e25fdfd",
"container": "safetensors",
"index_sha256": null,
"precision_label": "BF16",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "572cf29b76b1811bdfcc277382c42fd3c9f4f3c7962a704f14d41ccab06bd747"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/minimax-m2-tiny-random-bf16/tree/d4825603496f1c636385af403e891aaf20e9afe6"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | null | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/minimax-m2-tiny-random-bf16",
"revision": "d4825603496f1c636385af403e891aaf20e9afe6",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/minimax-m2-tiny-random-bf16"
} | artifact--malaiwah.minimax-m2-tiny-random-bf16.d4825603496f | base | model--malaiwah.minimax-m2-tiny-random-bf16 | minimax-m2-tiny-random-bf16 (random test fixture) native BF16 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "1cbc135335c4d6863bd412e8f68a664956a7f1248c99b4db9a7451d83a5d6a3d",
"uri": "https://huggingface.co/datasets/malaiwah/minimax-m2-tiny-fidelity-root-v1/resolve/0429feb8bc3f427e89e0ba2a341d28b82d431b0e/fidelity-dataset.json"
},
{
"kind": "hf_file",
"sha256": "f24a... | {
"config_sha256": "938204c4fe555d9229c16772e9a6670a344206a9ed17f9ebaca97adea1d4d869",
"container": "safetensors",
"index_sha256": null,
"precision_label": "BF16",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "68de109d1391135242bc28b6972f0a3c67b24eb2ace6c2b0f8816aace1affe72"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/minimax-m3-tiny-random-bf16/tree/ca904dff2941004a3f5bd5af62064fcfb34da8b9"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | null | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/minimax-m3-tiny-random-bf16",
"revision": "ca904dff2941004a3f5bd5af62064fcfb34da8b9",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/minimax-m3-tiny-random-bf16"
} | artifact--malaiwah.minimax-m3-tiny-random-bf16.ca904dff2941 | base | model--malaiwah.minimax-m3-tiny-random-bf16 | minimax-m3-tiny-random-bf16 (random test fixture) native BF16 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "a4556925a0841410e24e8212b154eafb849dd5e00a7bede521247c71830306b0",
"uri": "https://huggingface.co/datasets/malaiwah/minimax-m3-tiny-fidelity-root-v1/resolve/4bdcedf286fa2de29bc11102e90cac47cd77cd14/fidelity-dataset.json"
},
{
"kind": "hf_file",
"sha256": "e407... | {
"config_sha256": "7fb52d928d3ab4a376f7af7b838c92b434f66995c1be01cd6a0f911fff1f4c90",
"container": "safetensors",
"index_sha256": null,
"precision_label": "BF16",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "c3ae6e37101836c9dbaf85e82e91fe4174707b1c007d1486ebe7f0b3de360f1c"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen3-5-gguf-tiny-random-bf16/tree/490767e58441a55c5a9f6375ea0e31e2cdc9da76"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | null | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen3-5-gguf-tiny-random-bf16",
"revision": "490767e58441a55c5a9f6375ea0e31e2cdc9da76",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen3-5-gguf-tiny-random-bf16"
} | artifact--malaiwah.qwen3-5-gguf-tiny-random-bf16.490767e58441 | base | model--malaiwah.qwen3-5-gguf-tiny-random-bf16 | qwen3-5-gguf-tiny-random-bf16 (random test fixture) native BF16 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "7e7512737df876ec95ce5c2df4d352a55b1f04fe6f8fa3a59a34904bd14c2425",
"uri": "https://huggingface.co/datasets/malaiwah/qwen3-5-gguf-tiny-fidelity-root-v1/resolve/af1922b4fb5c619e0bf2e92c1bfcd68d8a83cb32/fidelity-dataset.json"
},
{
"kind": "hf_file",
"sha256": "6f... | {
"config_sha256": "af379f490737e9834a34adbf85c995fc8b834a675b9d90494c58fd17afa62ebd",
"container": "safetensors",
"index_sha256": null,
"precision_label": "BF16",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "170f6305c16e5dc1704bae7735fe46d9fa64b7988adf43542eaacf85ef53d7f6"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-awq-gemm-g32-rtn-format/tree/50e3ae8d2999bd381ccefad3662e264b450a43ca"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "awq",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": nu... | null | artifact--malaiwah.qwen3-5-tiny-random-bf16.a430e41d5814 | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen3-5-tiny-random-awq-gemm-g32-rtn-format",
"revision": "50e3ae8d2999bd381ccefad3662e264b450a43ca",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen3-5-tiny-random... | artifact--malaiwah.qwen3-5-tiny-random-awq-gemm-g32-rtn-format.50e3ae8d2999 | quant | model--malaiwah.qwen3-5-tiny-random-bf16 | qwen3-5-tiny-random-awq-gemm-g32-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "Unchanged from pinned root; no experts in this dense model",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:int4@4|mlp.gate=quantized:int4@4|mlp.up=quantized:int4@4|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "5ae2a74f1f5d28f19c0d3f6ff3797c7e7f8c61f72b023f6afac8a7da52144f2d",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-affine-tiny-cpu-format-v1/resolve/2a1f5fb39eedbe038eb66e851ef3d19b0daa4493/raw/candidates/awq-gemm-g32/evidence/candidate/fidelity-dataset.json"
},
... | {
"config_sha256": "32e891c3497959793e48265661a86a3488d3f81290970e5b1ccef52569a8202e",
"container": "safetensors",
"precision_label": "awq-gemm-g32; RTN packing, optimizer-not-run",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "c4f82d2fb16ec879f283418feeea2b44752cb28692217cd157591e6605a32039"
... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-bf16/tree/a430e41d5814ba5e7bfa5ee29d86935aac56d95d"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | null | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen3-5-tiny-random-bf16",
"revision": "a430e41d5814ba5e7bfa5ee29d86935aac56d95d",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-bf16"
} | artifact--malaiwah.qwen3-5-tiny-random-bf16.a430e41d5814 | base | model--malaiwah.qwen3-5-tiny-random-bf16 | qwen3-5-tiny-random-bf16 (random test fixture) native BF16 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "37ab958da203eb7567c43e842e3b440afac2bad4cdd1ac3f3644fc6485560465",
"uri": "https://huggingface.co/datasets/malaiwah/qwen3-5-tiny-fidelity-root-v1/resolve/7e1a755b7c9ee43d2b95c981de177392241e780d/fidelity-dataset.json"
},
{
"kind": "hf_file",
"sha256": "d21c46a... | {
"config_sha256": "265b5ab2c5f5b1026eaa1de1bedf0c9e8b8ac64402fdb5bd01023c4186e8d44d",
"container": "safetensors",
"index_sha256": null,
"precision_label": "BF16",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "e039d547c13f1294447e09eed708b0953bc4a9e22c4c179bfd2873f9f271abf0"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-ct-asymmetric-g32-rtn-format/tree/35882a286c17e1959ccb936618490d3fef4e3d3e"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "int4",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": n... | null | artifact--malaiwah.qwen3-5-tiny-random-bf16.a430e41d5814 | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen3-5-tiny-random-ct-asymmetric-g32-rtn-format",
"revision": "35882a286c17e1959ccb936618490d3fef4e3d3e",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen3-5-tiny-r... | artifact--malaiwah.qwen3-5-tiny-random-ct-asymmetric-g32-rtn-format.35882a286c17 | quant | model--malaiwah.qwen3-5-tiny-random-bf16 | qwen3-5-tiny-random-ct-asymmetric-g32-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "Unchanged from pinned root; no experts in this dense model",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:int4@4|mlp.gate=quantized:int4@4|mlp.up=quantized:int4@4|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "553e507c90a80a9e3940eb13df99ea5c6725dbdb121bcc92dbed0fd17bd9fa2c",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-affine-tiny-cpu-format-v1/resolve/2a1f5fb39eedbe038eb66e851ef3d19b0daa4493/raw/candidates/ct-asymmetric-g32/evidence/candidate/fidelity-dataset.json"
... | {
"config_sha256": "a85a9fd380b970972534a7a8fa0f6867d665096816ffd0399f5d6683ead73f64",
"container": "safetensors",
"precision_label": "ct-asymmetric-g32; RTN packing, optimizer-not-run",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "20c30777a80925049729b57b1e7f23627a2713b8a1798536982f7e0fa1b31... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-ct-int8-symmetric-g64-rtn-format/tree/7f59af6246a3a5d4a3684357b8612e292eed2ff0"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 8,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "int8",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": n... | null | artifact--malaiwah.qwen3-5-tiny-random-bf16.a430e41d5814 | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen3-5-tiny-random-ct-int8-symmetric-g64-rtn-format",
"revision": "7f59af6246a3a5d4a3684357b8612e292eed2ff0",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen3-5-ti... | artifact--malaiwah.qwen3-5-tiny-random-ct-int8-symmetric-g64-rtn-format.7f59af6246a3 | quant | model--malaiwah.qwen3-5-tiny-random-bf16 | qwen3-5-tiny-random-ct-int8-symmetric-g64-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "Unchanged from pinned root; no experts in this dense model",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:int8@8|mlp.gate=quantized:int8@8|mlp.up=quantized:int8@8|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "eea1964a5d387e90f57e5f007d56ba0f2e4a2e3181080cf4e74643db6d4dff16",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-affine-tiny-cpu-format-v1/resolve/2a1f5fb39eedbe038eb66e851ef3d19b0daa4493/raw/candidates/ct-int8-symmetric-g64/evidence/candidate/fidelity-dataset.json... | {
"config_sha256": "eaf4106b93db0d74ed8ce5f1d6a4259c4866dd21a648f4355d5a8a50647ab013",
"container": "safetensors",
"precision_label": "ct-int8-symmetric-g64; RTN packing, optimizer-not-run",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "4ca46dc78794cb3ea3fe4588b8644a96f9ecb01ac68fc6a4f109d576f... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-ct-symmetric-g64-rtn-format/tree/22f25d5075c9afbb44bed9568d379bfcde79638e"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "int4",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": n... | null | artifact--malaiwah.qwen3-5-tiny-random-bf16.a430e41d5814 | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen3-5-tiny-random-ct-symmetric-g64-rtn-format",
"revision": "22f25d5075c9afbb44bed9568d379bfcde79638e",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen3-5-tiny-ra... | artifact--malaiwah.qwen3-5-tiny-random-ct-symmetric-g64-rtn-format.22f25d5075c9 | quant | model--malaiwah.qwen3-5-tiny-random-bf16 | qwen3-5-tiny-random-ct-symmetric-g64-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "Unchanged from pinned root; no experts in this dense model",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:int4@4|mlp.gate=quantized:int4@4|mlp.up=quantized:int4@4|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "ac3c1d20b5b7cc27249bb2b0c5d9558e28aa99a0945c7a0d9b07696b96c990b5",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-affine-tiny-cpu-format-v1/resolve/2a1f5fb39eedbe038eb66e851ef3d19b0daa4493/raw/candidates/ct-symmetric-g64/evidence/candidate/fidelity-dataset.json"
}... | {
"config_sha256": "2668e38fc729d8a2f764a178cc4603a16cb5f5941a9d447e406b62a5e1132c50",
"container": "safetensors",
"precision_label": "ct-symmetric-g64; RTN packing, optimizer-not-run",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "329f75118676ebb92ce1f8a600ee056e7492003b036e6980cda47459f8171e... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-gguf-f16-format/tree/95d3fe871df7afb89891cabe2841cd32f5788964"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "fp16",
"group_size": null
} | null | artifact--malaiwah.qwen3-5-gguf-tiny-random-bf16.490767e58441 | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen3-5-tiny-random-gguf-f16-format",
"revision": "95d3fe871df7afb89891cabe2841cd32f5788964",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-gguf-f1... | artifact--malaiwah.qwen3-5-tiny-random-gguf-f16-format.95d3fe871df7 | base | model--malaiwah.qwen3-5-gguf-tiny-random-bf16 | qwen3-5-tiny-random-gguf-f16-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "mixed",
"layer_range": "all",
"note": "Measured types F16,F32; payload bits/weight=16.277436970320203",
"tensor_class": "attn.other",
"treatment": "native"
},
{
"bits_per_w... | attn.other=native:mixed|embed_tokens=native:fp16@16|lm_head=native:fp16@16|mlp.down=native:fp16@16|mlp.gate=native:fp16@16|mlp.up=native:fp16@16|norm=native:fp32@32|head=native|kv=not_applicable | null | [
{
"kind": "hf_file",
"sha256": "21c77951649ac8d67f1acd2073d83f2a2ba6cabbd789c43c080d146866485405",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-qwen-gguf-tiny-cpu-format-v1/resolve/7cb03a8be6ae3ebb1791c11b625718435308ddc5/raw/candidates/f16/evidence/f16/fidelity-dataset.json"
},
{
"kind"... | {
"config_sha256": "af379f490737e9834a34adbf85c995fc8b834a675b9d90494c58fd17afa62ebd",
"container": "gguf",
"precision_label": "canonical dense Qwen35 GGUF f16",
"shard_count": 1,
"shard_sha256": {
"qwen35-tiny-F16.gguf": "a78942fc323229a7f7259baa00943dbfcb56c30ce51f762ac4a8c77d2dac6544"
},
"size_basi... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-gguf-f32-format/tree/27678e2b2c3a95235e4dad9a2ac9a24d2b257408"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "fp32",
"group_size": null
} | null | artifact--malaiwah.qwen3-5-gguf-tiny-random-bf16.490767e58441 | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen3-5-tiny-random-gguf-f32-format",
"revision": "27678e2b2c3a95235e4dad9a2ac9a24d2b257408",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-gguf-f3... | artifact--malaiwah.qwen3-5-tiny-random-gguf-f32-format.27678e2b2c3a | base | model--malaiwah.qwen3-5-gguf-tiny-random-bf16 | qwen3-5-tiny-random-gguf-f32-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 32,
"format": "fp32",
"layer_range": "all",
"note": "Measured types F32; payload bits/weight=32.0",
"tensor_class": "attn.other",
"treatment": "native"
},
{
"bits_per_weight": 32,
"fo... | attn.other=native:fp32@32|embed_tokens=native:fp32@32|lm_head=native:fp32@32|mlp.down=native:fp32@32|mlp.gate=native:fp32@32|mlp.up=native:fp32@32|norm=native:fp32@32|head=native|kv=not_applicable | null | [
{
"kind": "hf_file",
"sha256": "723a42fff03b053fc55985679637b71976cf04ecd8b9925a6ea9797178559a10",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-qwen-gguf-tiny-cpu-format-v1/resolve/7cb03a8be6ae3ebb1791c11b625718435308ddc5/raw/candidates/f32/evidence/f32/fidelity-dataset.json"
},
{
"kind"... | {
"config_sha256": "af379f490737e9834a34adbf85c995fc8b834a675b9d90494c58fd17afa62ebd",
"container": "gguf",
"precision_label": "canonical dense Qwen35 GGUF f32",
"shard_count": 1,
"shard_sha256": {
"qwen35-tiny-F32.gguf": "bdca1575048d3e9f0a07e40e95c22fe9b861a00940ca4650646ec0f6b39f762d"
},
"size_basi... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-gguf-q4_0-format/tree/62077782002e9586b747244fe64593bacbd5eee5"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "gguf-k-quant",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revi... | null | artifact--malaiwah.qwen3-5-gguf-tiny-random-bf16.490767e58441 | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen3-5-tiny-random-gguf-q4_0-format",
"revision": "62077782002e9586b747244fe64593bacbd5eee5",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-gguf-q... | artifact--malaiwah.qwen3-5-tiny-random-gguf-q4-0-format.62077782002e | quant | model--malaiwah.qwen3-5-gguf-tiny-random-bf16 | qwen3-5-tiny-random-gguf-q4_0-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "mixed",
"layer_range": "all",
"note": "Measured types F32,Q4_0; payload bits/weight=4.976844792737846",
"tensor_class": "attn.other",
"treatment": "quantized"
},
{
"bits_pe... | attn.other=quantized:mixed|embed_tokens=quantized:gguf-k-quant@4|lm_head=quantized:gguf-k-quant@4|mlp.down=quantized:gguf-k-quant@4|mlp.gate=quantized:gguf-k-quant@4|mlp.up=quantized:gguf-k-quant@4|norm=native:fp32@32|head=quantized|kv=not_applicable | null | [
{
"kind": "hf_file",
"sha256": "fbd727876589287b6669dc6bea6f69b3ab53fbb45823210883004069545af214",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-qwen-gguf-tiny-cpu-format-v1/resolve/7cb03a8be6ae3ebb1791c11b625718435308ddc5/raw/candidates/q4_0/evidence/q4_0/fidelity-dataset.json"
},
{
"kin... | {
"config_sha256": "af379f490737e9834a34adbf85c995fc8b834a675b9d90494c58fd17afa62ebd",
"container": "gguf",
"precision_label": "canonical dense Qwen35 GGUF q4_0",
"shard_count": 1,
"shard_sha256": {
"qwen35-tiny-Q4_0.gguf": "e368e1be98c4ad0ea259578fc21d4847afbfc6b37608f1c5ce9dfb9f5b6763d1"
},
"size_ba... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-gguf-q8_0-format/tree/775bd7604f838e983a23ef00606fa0f592a38932"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 8,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "gguf-k-quant",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revi... | null | artifact--malaiwah.qwen3-5-gguf-tiny-random-bf16.490767e58441 | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen3-5-tiny-random-gguf-q8_0-format",
"revision": "775bd7604f838e983a23ef00606fa0f592a38932",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-gguf-q... | artifact--malaiwah.qwen3-5-tiny-random-gguf-q8-0-format.775bd7604f83 | quant | model--malaiwah.qwen3-5-gguf-tiny-random-bf16 | qwen3-5-tiny-random-gguf-q8_0-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "mixed",
"layer_range": "all",
"note": "Measured types F32,Q8_0; payload bits/weight=8.907485550157796",
"tensor_class": "attn.other",
"treatment": "quantized"
},
{
"bits_pe... | attn.other=quantized:mixed|embed_tokens=quantized:gguf-k-quant@8|lm_head=quantized:gguf-k-quant@8|mlp.down=quantized:gguf-k-quant@8|mlp.gate=quantized:gguf-k-quant@8|mlp.up=quantized:gguf-k-quant@8|norm=native:fp32@32|head=quantized|kv=not_applicable | null | [
{
"kind": "hf_file",
"sha256": "dace5182d5d42202954cae8471a72a880a0387aa17a320c200808492898b393f",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-qwen-gguf-tiny-cpu-format-v1/resolve/7cb03a8be6ae3ebb1791c11b625718435308ddc5/raw/candidates/q8_0/evidence/q8_0/fidelity-dataset.json"
},
{
"kin... | {
"config_sha256": "af379f490737e9834a34adbf85c995fc8b834a675b9d90494c58fd17afa62ebd",
"container": "gguf",
"precision_label": "canonical dense Qwen35 GGUF q8_0",
"shard_count": 1,
"shard_sha256": {
"qwen35-tiny-Q8_0.gguf": "163812796913d6167b52f40ca7a59c685da9c18be1c3dd3e97a6ca32c1da31fe"
},
"size_ba... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-gptq-v1-g32-rtn-format/tree/143b652b954d296a9c81ad45947fa5812bd719a4"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "gptq",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": n... | null | artifact--malaiwah.qwen3-5-tiny-random-bf16.a430e41d5814 | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen3-5-tiny-random-gptq-v1-g32-rtn-format",
"revision": "143b652b954d296a9c81ad45947fa5812bd719a4",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-... | artifact--malaiwah.qwen3-5-tiny-random-gptq-v1-g32-rtn-format.143b652b954d | quant | model--malaiwah.qwen3-5-tiny-random-bf16 | qwen3-5-tiny-random-gptq-v1-g32-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "Unchanged from pinned root; no experts in this dense model",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:int4@4|mlp.gate=quantized:int4@4|mlp.up=quantized:int4@4|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "fcb97a090210c49b3f892b405e3ff016f1a85cd7d71bede9db206b662faba0d8",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-affine-tiny-cpu-format-v1/resolve/2a1f5fb39eedbe038eb66e851ef3d19b0daa4493/raw/candidates/gptq-v1-g32/evidence/candidate/fidelity-dataset.json"
},
{... | {
"config_sha256": "b59585cbf1c25339c8767e5f9ee06408d790fb630e4eec593f829fb646f21cde",
"container": "safetensors",
"precision_label": "gptq-v1-g32; RTN packing, optimizer-not-run",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "679584e13e2d029958fd54d1e9cc407b16090151e7d23338d4fea912afe99ac5"
... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-gptq-v2-desc-g64-rtn-format/tree/8e4cf325e62aed0c111f22785e75a1b890d59f77"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "gptq",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": n... | null | artifact--malaiwah.qwen3-5-tiny-random-bf16.a430e41d5814 | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen3-5-tiny-random-gptq-v2-desc-g64-rtn-format",
"revision": "8e4cf325e62aed0c111f22785e75a1b890d59f77",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen3-5-tiny-ra... | artifact--malaiwah.qwen3-5-tiny-random-gptq-v2-desc-g64-rtn-format.8e4cf325e62a | quant | model--malaiwah.qwen3-5-tiny-random-bf16 | qwen3-5-tiny-random-gptq-v2-desc-g64-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "Unchanged from pinned root; no experts in this dense model",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:int4@4|mlp.gate=quantized:int4@4|mlp.up=quantized:int4@4|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "275ef2dfa8f167f41c63e4fa0a9e3c66c3d608510ea114d4b70de114a01caf04",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-affine-tiny-cpu-format-v1/resolve/2a1f5fb39eedbe038eb66e851ef3d19b0daa4493/raw/candidates/gptq-v2-desc-g64/evidence/candidate/fidelity-dataset.json"
}... | {
"config_sha256": "4c33891268dd56559ec41db16bb9beefc138664342421542f80c3a7d332d4ccb",
"container": "safetensors",
"precision_label": "gptq-v2-desc-g64; RTN packing, optimizer-not-run",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "cfbb6579f9d987c30ff846d9d509ae7c179b58dad5ad94373dc90a749496cc... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-mlx4-g32-rtn-format/tree/7e124e9a5672de3770272cbc754b39b8d9b6d508"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "mlx-affine",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revisi... | null | artifact--malaiwah.qwen3-5-tiny-random-bf16.a430e41d5814 | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen3-5-tiny-random-mlx4-g32-rtn-format",
"revision": "7e124e9a5672de3770272cbc754b39b8d9b6d508",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-mlx... | artifact--malaiwah.qwen3-5-tiny-random-mlx4-g32-rtn-format.7e124e9a5672 | quant | model--malaiwah.qwen3-5-tiny-random-bf16 | qwen3-5-tiny-random-mlx4-g32-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "Unchanged from pinned root; no experts in this dense model",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:int4@4|mlp.gate=quantized:int4@4|mlp.up=quantized:int4@4|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "fc99b3320ec84105d50952674b1946c5172cac5c38c45f939472c011f180ceae",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-affine-tiny-cpu-format-v1/resolve/2a1f5fb39eedbe038eb66e851ef3d19b0daa4493/raw/candidates/mlx4-g32/evidence/candidate/fidelity-dataset.json"
},
{
... | {
"config_sha256": "8a15aeed40881d2555650c5065763bbb11fb39dd757b62291fb9d7b39742463c",
"container": "safetensors",
"precision_label": "mlx4-g32; RTN packing, optimizer-not-run",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "ce8e8367b541ad90a4816c451249835814f8b68780a9579d755a917ce944e328"
},... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-mlx8-g64-rtn-format/tree/f6a62fe2495dd9da2eec403105b626b3cb598a32"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 8,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "mlx-affine",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revisi... | null | artifact--malaiwah.qwen3-5-tiny-random-bf16.a430e41d5814 | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen3-5-tiny-random-mlx8-g64-rtn-format",
"revision": "f6a62fe2495dd9da2eec403105b626b3cb598a32",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen3-5-tiny-random-mlx... | artifact--malaiwah.qwen3-5-tiny-random-mlx8-g64-rtn-format.f6a62fe2495d | quant | model--malaiwah.qwen3-5-tiny-random-bf16 | qwen3-5-tiny-random-mlx8-g64-rtn-format (FORMAT fixture) | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "Unchanged from pinned root; no experts in this dense model",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:int8@8|mlp.gate=quantized:int8@8|mlp.up=quantized:int8@8|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "4634b95d27b4a2279fe840e7c31e9cd0a029c1e2054cb72b5658b52a5f4cc759",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-affine-tiny-cpu-format-v1/resolve/2a1f5fb39eedbe038eb66e851ef3d19b0daa4493/raw/candidates/mlx8-g64/evidence/candidate/fidelity-dataset.json"
},
{
... | {
"config_sha256": "6546388d044a364140f113a47c2b410bef26ddb8d6855ca4cc0077c938b94656",
"container": "safetensors",
"precision_label": "mlx8-g64; RTN packing, optimizer-not-run",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "06f4050299eee8e5d20582225b7453ff49f43c4a562893ae4952533dbf03619a"
},... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/Qwen3.8-27B-EXL3-K5K6"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 5,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": true,
"code": "revision_unpinned",
"detail": "No measurement receipt for this artifact records a Hub revision. Every kld5 receipt records model_revision=null / model_revision_source='none'. Identity rests on index_sha256 and the per-shard sha256 map the receipt carries.",
... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/Qwen3.8-27B-EXL3-K5K6",
"revision": null,
"revision_source": "none",
"status": "known",
"url": "https://huggingface.co/malaiwah/Qwen3.8-27B-EXL3-K5K6"
} | artifact--malaiwah.qwen3.8-27b-exl3-k5k6 | quant | model--qwen.qwen3.8-27b | malaiwah Qwen3.8-27B EXL3 K5K6 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "quantizer",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor... | attn.o=native:bf16|attn.qkv=native:bf16|embed_tokens=native:bf16|lm_head=quantized:exl3-mcg@6|mlp.down=quantized:exl3-mcg@6|mlp.gate=quantized:exl3-mcg@5|mlp.up=quantized:exl3-mcg@5|mtp=native:bf16|norm=native:bf16|head=quantized|kv=bf16 | {
"sealed": true
} | [
{
"kind": "github_file",
"note": "candidate_index_sha256 f8ca5af9...",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/kld5-10M-k5k6.json"
},
{
"kind": "model_card",
"note": "MODEL_CARD-K5K6.md: 'Attention weights ship in B... | {
"container": "exl3",
"precision_label": "K5/K5/K6",
"size_basis": "repo_all_files",
"size_bytes": 30597231933,
"size_gb": 30.597231933
} |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/Qwen3.8-27B-EXL3-K5K6-context"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 5,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": true,
"code": "revision_unpinned",
"detail": "No measurement receipt for this artifact records a Hub revision. Every kld5 receipt records model_revision=null / model_revision_source='none'. Identity rests on index_sha256 and the per-shard sha256 map the receipt carries.",
... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/Qwen3.8-27B-EXL3-K5K6-context",
"revision": null,
"revision_source": "none",
"status": "known",
"url": "https://huggingface.co/malaiwah/Qwen3.8-27B-EXL3-K5K6-context"
} | artifact--malaiwah.qwen3.8-27b-exl3-k5k6-context | quant | model--qwen.qwen3.8-27b | malaiwah Qwen3.8-27B EXL3 K5K6 context | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "quantizer",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 5,
"format": "exl3-mcg",
"layer_range": "all",
"tenso... | attn.o=quantized:exl3-mcg@5|attn.qkv=quantized:exl3-mcg@5|embed_tokens=native:bf16|lm_head=quantized:exl3-mcg@6|mlp.down=quantized:exl3-mcg@6|mlp.gate=quantized:exl3-mcg@5|mlp.up=quantized:exl3-mcg@5|mtp=quantized:exl3-mcg@5|norm=native:bf16|head=quantized|kv=bf16 | {
"sealed": true
} | [
{
"kind": "github_file",
"note": "candidate_index_sha256 cd53b8e4...",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/kld5-10M-ctx.json"
},
{
"kind": "model_card",
"uri": "https://huggingface.co/malaiwah/Qwen3.8-27B-EXL3-K... | {
"container": "exl3",
"precision_label": "K5/K5/K6 + K5 attention",
"size_basis": "repo_all_files",
"size_bytes": 20696053306,
"size_gb": 20.696053306
} |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/Qwen3.8-27B-EXL3-K5K6-hydrated"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 5,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": true,
"code": "revision_unpinned",
"detail": "No measurement receipt for this artifact records a Hub revision. Every kld5 receipt records model_revision=null / model_revision_source='none'. Identity rests on index_sha256 and the per-shard sha256 map the receipt carries.",
... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/Qwen3.8-27B-EXL3-K5K6-hydrated",
"revision": null,
"revision_source": "none",
"status": "known",
"url": "https://huggingface.co/malaiwah/Qwen3.8-27B-EXL3-K5K6-hydrated"
} | artifact--malaiwah.qwen3.8-27b-exl3-k5k6-hydrated | quant | model--qwen.qwen3.8-27b | malaiwah Qwen3.8-27B EXL3 K5K6 hydrated | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "quantizer",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 6,
"format": "exl3-mcg",
"layer_range": "all",
"tenso... | attn.o=quantized:exl3-mcg@6|attn.qkv=quantized:exl3-mcg@6|embed_tokens=native:bf16|lm_head=quantized:exl3-mcg@6|mlp.down=quantized:exl3-mcg@6|mlp.gate=quantized:exl3-mcg@5|mlp.up=quantized:exl3-mcg@5|mtp=quantized:exl3-mcg@5|norm=native:bf16|head=quantized|kv=bf16 | {
"sealed": true
} | [
{
"kind": "github_file",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/kld5-10M-hyd.json"
},
{
"kind": "model_card",
"note": "attention EXL3 K6 serialized on disk (calibrated), quantized MTP",
"uri": "https://huggingface.... | {
"container": "exl3",
"precision_label": "K5/K5/K6 + K6 attention",
"size_basis": "repo_all_files",
"size_bytes": 21610933884,
"size_gb": 21.610933884
} |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/Qwen3.8-27B-EXL3-K6-parity"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 6,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "Revision a34ebcea is the publication receipt's; the Hub head has since moved by 40,966 B of card and doc edits only.",
"severity": "info"
}
] | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/Qwen3.8-27B-EXL3-K6-parity",
"revision": "a34ebcea909e43b3eb5b66b43782d9a509bda14b",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/malaiwah/Qwen3.8-27B-EXL3-K6-parity"
} | artifact--malaiwah.qwen3.8-27b-exl3-k6-parity | quant | model--qwen.qwen3.8-27b | malaiwah Qwen3.8-27B EXL3 K6-parity | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "quantizer",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 6,
"format": "exl3-mcg",
"layer_range": "all",
"tenso... | attn.o=quantized:exl3-mcg@6|attn.qkv=quantized:exl3-mcg@6|embed_tokens=native:bf16|lm_head=quantized:exl3-mcg@6|mlp.down=quantized:exl3-mcg@6|mlp.gate=quantized:exl3-mcg@6|mlp.up=quantized:exl3-mcg@6|mtp=quantized:exl3-mcg@6|norm=native:bf16|head=quantized|kv=bf16 | {
"sealed": true
} | [
{
"kind": "github_file",
"note": "candidate_index_sha256 a35eb2fe...",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/kld5-1M-k6parity.json"
},
{
"kind": "model_card",
"note": "MODEL_CARD-K6-parity.md: full_attention and l... | {
"container": "exl3",
"precision_label": "K6",
"size_basis": "repo_all_files",
"size_bytes": 23059333816,
"size_gb": 23.059333816
} |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/Qwen3.8-27B-K4"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": true,
"code": "revision_unpinned",
"detail": "No measurement receipt for this artifact records a Hub revision. Every kld5 receipt records model_revision=null / model_revision_source='none'. Identity rests on index_sha256 and the per-shard sha256 map the receipt carries.",
... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/Qwen3.8-27B-K4",
"revision": null,
"revision_source": "none",
"status": "known",
"url": "https://huggingface.co/malaiwah/Qwen3.8-27B-K4"
} | artifact--malaiwah.qwen3.8-27b-k4 | quant | model--qwen.qwen3.8-27b | malaiwah Qwen3.8-27B K4 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "quantizer",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"note":... | attn.o=native:bf16|attn.qkv=native:bf16|embed_tokens=native:bf16|lm_head=quantized:exl3-mcg@6|mlp.down=quantized:exl3-mcg@4|mlp.gate=quantized:exl3-mcg@4|mlp.up=quantized:exl3-mcg@4|mtp=native:bf16|norm=native:bf16|head=quantized|kv=bf16 | {
"sealed": true
} | [
{
"kind": "github_file",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/kld5-10M-k4.json"
},
{
"kind": "model_card",
"note": "MODEL_CARD-K4.md: MLP all K4, lm_head K6, attention BF16 on disk",
"uri": "https://huggingface.c... | {
"container": "exl3",
"precision_label": "K4",
"size_basis": "repo_all_files",
"size_bytes": 28345369355,
"size_gb": 28.345369355
} |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/qwen4-exp-tiny-random-bf16/tree/90fcdd52d45a200a1b6d446886638af14957117c"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | null | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/qwen4-exp-tiny-random-bf16",
"revision": "90fcdd52d45a200a1b6d446886638af14957117c",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/qwen4-exp-tiny-random-bf16"
} | artifact--malaiwah.qwen4-exp-tiny-random-bf16.90fcdd52d45a | base | model--malaiwah.qwen4-exp-tiny-random-bf16 | qwen4-exp-tiny-random-bf16 (random test fixture) native BF16 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "cf5464e611a5ee1975599dd7ebad7b0ee292fa18ff42f1e72ae9c17332be68c6",
"uri": "https://huggingface.co/datasets/malaiwah/qwen4-exp-tiny-fidelity-root-v1/resolve/c6e1af0383df78006816ca1af067b4c5374bd7c1/fidelity-dataset.json"
},
{
"kind": "hf_file",
"sha256": "4a1fd... | {
"config_sha256": "e205718d15ca6188eaff9f62d2d15e0615714cfbe18f83b0bcff0d568782a6dc",
"container": "safetensors",
"index_sha256": null,
"precision_label": "BF16",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "d4e68baa72275b3cf509602542c4d5c56a4a9229a13f396dfb451dc59d2af536"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/SmolLM2-135M-QFS-gptq-int4-g32"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "int4",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": n... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "no_known_deviations",
"detail": "Declared by the submission receipt.",
"severity": "info"
}
] | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/SmolLM2-135M-QFS-gptq-int4-g32",
"revision": "0a3e5c421fbb3d3ab6c7a56d585c9adf29fcf405",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/SmolLM2-135M-QFS-gptq-int4-g32"
... | artifact--malaiwah.smollm2-135m-qfs-gptq-int4-g32 | quant | model--malaiwah.smollm2-135m-qfs-native-bf16 | malaiwah/SmolLM2-135M-QFS-gptq-int4-g32 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "quantizer",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": null,
"note": "Untouched native BF16 tensors.",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": null,
... | attn.o=quantized:int4@4|attn.qkv=quantized:int4@4|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:int4@4|mlp.gate=quantized:int4@4|mlp.up=quantized:int4@4|norm=native:bf16@16|head=native|kv=bf16 | {
"sealed": false
} | [
{
"kind": "url",
"uri": "https://huggingface.co/malaiwah/SmolLM2-135M-QFS-gptq-int4-g32"
},
{
"kind": "hf_file",
"note": "Original producing plan binds the exact serialized weight-file census and its size basis.",
"sha256": "ffc98900096b009f54a1c64c6caca86e3f8dd593a3ea80e119d8df524b732312",
... | {
"config_sha256": "8776884d55e201f48448b258db3e2eb4716375342812c1ed8436e5d2a82a5d50",
"container": "safetensors",
"index_sha256": null,
"precision_label": "int4",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "6865ebf24efadc8fa5df7f60a376fc178c58687675e51b5dfaf49de02245b877"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/SmolLM2-135M-QFS-gptq-int4-g64"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "int4",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": n... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "no_known_deviations",
"detail": "Declared by the submission receipt.",
"severity": "info"
}
] | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/SmolLM2-135M-QFS-gptq-int4-g64",
"revision": "ed14272c4db67b2ece2a8a2a479350b9eea8fa3a",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/SmolLM2-135M-QFS-gptq-int4-g64"
... | artifact--malaiwah.smollm2-135m-qfs-gptq-int4-g64 | quant | model--malaiwah.smollm2-135m-qfs-native-bf16 | malaiwah/SmolLM2-135M-QFS-gptq-int4-g64 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "quantizer",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": null,
"note": "Untouched native BF16 tensors.",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": null,
... | attn.o=quantized:int4@4|attn.qkv=quantized:int4@4|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:int4@4|mlp.gate=quantized:int4@4|mlp.up=quantized:int4@4|norm=native:bf16@16|head=native|kv=bf16 | {
"sealed": false
} | [
{
"kind": "url",
"uri": "https://huggingface.co/malaiwah/SmolLM2-135M-QFS-gptq-int4-g64"
},
{
"kind": "hf_file",
"note": "Original producing plan binds the exact serialized weight-file census and its size basis.",
"sha256": "ea5537879cc6d611274234498824ee1a6776fcca69ac59249123c02f29a47076",
... | {
"config_sha256": "d6e21170f4e7f8cbe0d6a99c9ca30459fececd38781f5eee52b427729cff07f9",
"container": "safetensors",
"index_sha256": null,
"precision_label": "int4",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "fe9ba8a70f29b6330ede69f9856cbe81ab049e507285656ad79e5be162d11b5b"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/SmolLM2-135M-QFS-native-bf16/tree/5dc8abe390e2db0102e6afd1509f68fbb5d54e6c"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | null | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Native HF Jobs root. Receipt integrity and qualification were reviewed; the registry owner did not independently reproduce this run.",
"severity": "info",
"sources": [
{
"kind": "hf_... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/SmolLM2-135M-QFS-native-bf16",
"revision": "5dc8abe390e2db0102e6afd1509f68fbb5d54e6c",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/SmolLM2-135M-QFS-native-bf16"
} | artifact--malaiwah.smollm2-135m-qfs-native-bf16.5dc8abe390e2 | base | model--malaiwah.smollm2-135m-qfs-native-bf16 | SmolLM2-135M native BF16 (unchanged source mirror) native BF16 | {
"handle": "HuggingFaceTB",
"is_registry_maintainer": false,
"name": "HuggingFaceTB",
"role": "model-publisher",
"url": "https://huggingface.co/HuggingFaceTB"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "05a10fecf8cc00e2ab418c77cf02dd815ce4f16803bccdb1b7a8063860539a92",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-evidence-d3dc69602aeb981f06bd9f4c726937f9-public/resolve/be1e39bf8b25aa3a5c9ba77c2fd180359f357689/first/capture/manifest.json"
},
{
"kind": "hf_... | {
"config_sha256": "1d556eab73b69c7f11f64c557a2f9c6f440bd4c6b89bb2584a6b498c92603843",
"container": "safetensors",
"index_sha256": null,
"precision_label": "BF16",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "80521b40281d6ce74e35c9282c22539e75aa0ac8578892b2a59955ef78d55da1"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/SmolLM2-135M-QFS-rtn-int4-g64"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "int4",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": n... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "no_known_deviations",
"detail": "Declared by the submission receipt.",
"severity": "info"
}
] | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/SmolLM2-135M-QFS-rtn-int4-g64",
"revision": "2082dbf88c86551a07712dd236e054cfb94b217a",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/SmolLM2-135M-QFS-rtn-int4-g64"
} | artifact--malaiwah.smollm2-135m-qfs-rtn-int4-g64 | quant | model--malaiwah.smollm2-135m-qfs-native-bf16 | malaiwah/SmolLM2-135M-QFS-rtn-int4-g64 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "quantizer",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": null,
"note": "Untouched native BF16 tensors.",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": null,
... | attn.o=quantized:int4@4|attn.qkv=quantized:int4@4|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:int4@4|mlp.gate=quantized:int4@4|mlp.up=quantized:int4@4|norm=native:bf16@16|head=native|kv=bf16 | {
"sealed": false
} | [
{
"kind": "url",
"uri": "https://huggingface.co/malaiwah/SmolLM2-135M-QFS-rtn-int4-g64"
},
{
"kind": "hf_file",
"note": "Original producing plan binds the exact serialized weight-file census and its size basis.",
"sha256": "af51b22671e802ae3502c4a592f4073b6c16b7bd06449a7bdd63bc2f0e6d05ca",
... | {
"config_sha256": "76e9a662712ef7b46e737eb58dc0cbe390fe5586c47d032476ecdd2b27ea44f1",
"container": "safetensors",
"index_sha256": null,
"precision_label": "int4",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "4680fa98bb183ca875b425571ae758f9c07b1c36085af4ab1b86e14433288bec"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/malaiwah/spark2-5-tiny-random-bf16/tree/c0a57b037007d475ede719e78f4a6ce24f211da0"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | null | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Synthetic/random-initialized test fixture, not a trained language model, assistant, quality benchmark, or production-kernel qualification. This native root is a reference for fixture fidelity only. Its mo... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "malaiwah/spark2-5-tiny-random-bf16",
"revision": "c0a57b037007d475ede719e78f4a6ce24f211da0",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/malaiwah/spark2-5-tiny-random-bf16"
} | artifact--malaiwah.spark2-5-tiny-random-bf16.c0a57b037007 | base | model--malaiwah.spark2-5-tiny-random-bf16 | spark2-5-tiny-random-bf16 (random test fixture) native BF16 | {
"handle": "malaiwah",
"is_registry_maintainer": true,
"name": "malaiwah",
"role": "model-publisher",
"url": "https://huggingface.co/malaiwah"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "bec09bb731fd3414ded4eaac4b3ea278d71325cd041156e25ecc8513a27941b9",
"uri": "https://huggingface.co/datasets/malaiwah/spark2-5-tiny-fidelity-root-v1/resolve/5715a4e51de6af894ac3963cd55d6ebe371dcb5a/fidelity-dataset.json"
},
{
"kind": "hf_file",
"sha256": "25cbfa... | {
"config_sha256": "b0fa5015db542fe185e0060dbcebc956ee7ee03aeecbdac5783383b62be28103",
"container": "safetensors",
"index_sha256": null,
"precision_label": "BF16",
"shard_count": 1,
"shard_sha256": {
"model.safetensors": "4619fd92be1df082a65e5cbd17013154bfb01fe9cd9f82acf2a71ce19672a436"
},
"size_bas... |
{
"status": "public",
"uri": "https://huggingface.co/Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-bf16.a6c167b6 | [
{
"affects_comparability": true,
"code": "byte_identical_redistribution",
"detail": "This is NOT an original quantization. Its 120 weight shards are byte-identical to brandonmusic/GLM-5.3-Flash-tr3-4bpw @ 5ab363a8 (artifact--brandonmusic.glm-5.3-flash-tr3-4bpw), proven two ways: equal LFS oids for all 1... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw",
"revision": "024db9f7e9871e8efdf21538ba55af7442be3cd5",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw"
} | artifact--mia-ailab.glm-5.3-flash-exl3-tr3-4bpw | quant | model--zai-org.glm-5.3-flash | Mia-AiLab GLM-5.3-Flash EXL3 TR3 4bpw (byte-identical mirror of brandonmusic's) | {
"handle": "brandonmusic",
"is_registry_maintainer": false,
"name": "brandonmusic",
"role": "quantizer",
"url": "https://huggingface.co/brandonmusic"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "routed-experts-only scope: not quantized",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"form... | attn.o=native:bf16@16|attn.other=native:mixed|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:exl3-mcg@4|moe.router=native:fp32@32|moe.shared_expert=native:bf16@16|mtp=quantized:exl3-mcg@4|norm=native:... | {
"note": "exl3-mcg-storage-abi.json + materialization-receipt.json (receipt 092be1ff..., plan a359003a...); all 12 claims recomputed from the published bytes by tr3_surface.verify_seal",
"sealed": true
} | [
{
"kind": "url",
"note": "the size on this record: 144 files, all-files sum 175,715,854,761 (weights.size_bytes, basis repo_all_files); safetensors sum 175,642,157,752, which is the quantity byte-identical to brandonmusic's original and what this row previously reported. Re-based to all-files in M4 so the f... | {
"container": "exl3",
"precision_label": "4bpw",
"shard_count": 120,
"size_basis": "repo_all_files",
"size_bytes": 175715854761,
"size_gb": 175.715854761
} |
{
"status": "public",
"uri": "https://huggingface.co/nvidia/GLM-5.2-NVFP4"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "nvfp4",
"group_size": 16,
"quantizer": {
"pipeline_ref": null,
"revision": nu... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "Read from bytes (the index plus a ranged read of every shard header): the 57,600 routed-expert matrices of layers 3-77 are NVFP4 modelopt component sets -- e2m1 weight U8 [out, in/2], weight_scale F8_E4M3 [out, in/16] (group 16 along th... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "nvidia/GLM-5.2-NVFP4",
"revision": "53e0691e21895a3863a606dfd12910c69eba94ab",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/nvidia/GLM-5.2-NVFP4"
} | artifact--nvidia.glm-5.2-nvfp4 | quant | model--zai-org.glm-5.2 | NVIDIA GLM-5.2-NVFP4 (routed experts NVFP4 e2m1 group 16, rest native) | {
"handle": "nvidia",
"is_registry_maintainer": false,
"name": "NVIDIA",
"role": "quantizer",
"url": "https://huggingface.co/nvidia"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "78 tensors: model.layers.N.self_attn.o_proj.weight. read from nvidia/GLM-5.2-NVFP4@53e0691e21895a3863a606dfd12910c69eba94ab model.safetensors.index.json + every s... | attn.o=native:bf16@16|attn.other=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:nvfp4@4|moe.router=native:mixed|moe.shared_expert=native:bf16@16|mtp=native:mixed|norm=native:bf16@16|hea... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision 53e0691e2189; 47 shards; config.json sha256 d3783a60... declares quantization_config quant_method modelopt / quant_algo NVFP4, config_groups.group_0.weights num_bits 4 group_size 16, producer modelopt 0.46.0.dev65+g977d34dc3; index sha256 2aa8397b...",
"uri": "h... | {
"config_sha256": "d3783a603e5aa9cb58eff7a5d8ac9c42d83156b229efe185c72e9e2dc8444923",
"container": "safetensors",
"index_sha256": "2aa8397b501d9f6a232d153f328feb912f813c389061aac4cf72b04914fa5b74",
"precision_label": "NVFP4",
"shard_count": 47,
"size_basis": "repo_weight_files",
"size_bytes": 46482304209... |
{
"status": "public",
"uri": "https://huggingface.co/zai-org/GLM-5.3-Flash"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-fp8 | [
{
"affects_comparability": true,
"code": "different_reference_kind",
"detail": "This is NOT a BF16 teacher. It is the official FP8 release dequantized to BF16. A student measured against it scores systematically LOWER than the same student measured against true BF16, because the FP8 error is in the refe... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "zai-org/GLM-5.3-Flash",
"revision": null,
"revision_source": "none",
"status": "known",
"url": "https://huggingface.co/zai-org/GLM-5.3-Flash"
} | artifact--orcarouter.glm-5.3-flash-fp8-dequantized | dequantized | model--zai-org.glm-5.3-flash | GLM-5.3-Flash FP8 dequantized to BF16 (orcarouter reference) | {
"handle": "orcarouter",
"is_registry_maintainer": false,
"name": "orcarouter",
"role": "toolchain-author",
"url": "https://huggingface.co/orcarouter"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor... | attn.o=native:bf16|attn.qkv=native:bf16|embed_tokens=native:bf16|lm_head=native:bf16|mlp.down=native:bf16|mlp.gate=native:bf16|mlp.up=native:bf16|moe.experts=native:bf16|norm=native:bf16|head=native|kv=bf16 | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "'All three tables compare each build against the full FP8 reference (dequantized to BF16 and run through the identical glm5_next forward, so the only variable is the quantization).'",
"uri": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
}
] | {
"container": "mlx",
"precision_label": "FP8 dequantized to BF16",
"size_basis": "unknown",
"size_bytes": 328000000000,
"size_gb": 328
} |
{
"status": "public",
"uri": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 2,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "mlx-affine",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revisi... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-fp8 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
},
{... | {
"link_type": "repository",
"path": "2-bit/",
"reason": null,
"repository": "orcarouter/GLM-5.3-Flash-MLX",
"revision": "c80f6810b1a95b5be9042761becc6aa78d189782",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
} | artifact--orcarouter.glm-5.3-flash-mlx-2bit | quant | model--zai-org.glm-5.3-flash | orcarouter GLM-5.3-Flash-MLX 2-bit | {
"handle": "orcarouter",
"is_registry_maintainer": false,
"name": "orcarouter",
"role": "quantizer",
"url": "https://huggingface.co/orcarouter"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
... | attn.o=unknown:unknown|attn.qkv=unknown:unknown|embed_tokens=unknown:unknown|lm_head=unknown:unknown|mlp.down=quantized:mlx-affine@2|mlp.gate=quantized:mlx-affine@2|mlp.up=quantized:mlx-affine@2|moe.experts=quantized:mlx-affine@2|mtp=quantized:mlx-affine@2|head=unknown|kv=unknown | {
"sealed": false
} | [
{
"kind": "model_card",
"uri": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
},
{
"kind": "url",
"note": "per-subfolder byte totals read from the Hub API",
"uri": "https://huggingface.co/api/models/orcarouter/GLM-5.3-Flash-MLX?blobs=true"
}
] | {
"container": "mlx",
"precision_label": "2-bit",
"size_basis": "repo_weight_files",
"size_bytes": 145006896343,
"size_gb": 145.006896343
} |
{
"status": "public",
"uri": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 2,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "mlx-affine",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revisi... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-fp8 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
},
{... | {
"link_type": "repository",
"path": "2bit-lite/",
"reason": null,
"repository": "orcarouter/GLM-5.3-Flash-MLX",
"revision": "c80f6810b1a95b5be9042761becc6aa78d189782",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
} | artifact--orcarouter.glm-5.3-flash-mlx-2bitlite | quant | model--zai-org.glm-5.3-flash | orcarouter GLM-5.3-Flash-MLX 2bit-lite | {
"handle": "orcarouter",
"is_registry_maintainer": false,
"name": "orcarouter",
"role": "quantizer",
"url": "https://huggingface.co/orcarouter"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
... | attn.o=unknown:unknown|attn.qkv=unknown:unknown|embed_tokens=unknown:unknown|lm_head=unknown:unknown|mlp.down=quantized:mlx-affine@2|mlp.gate=quantized:mlx-affine@2|mlp.up=quantized:mlx-affine@2|moe.experts=quantized:mlx-affine@2|mtp=quantized:mlx-affine@2|head=unknown|kv=unknown | {
"sealed": false
} | [
{
"kind": "model_card",
"uri": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
},
{
"kind": "url",
"note": "per-subfolder byte totals read from the Hub API",
"uri": "https://huggingface.co/api/models/orcarouter/GLM-5.3-Flash-MLX?blobs=true"
}
] | {
"container": "mlx",
"precision_label": "2bit-lite",
"size_basis": "repo_weight_files",
"size_bytes": 102460072185,
"size_gb": 102.460072185
} |
{
"status": "public",
"uri": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 3,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "mlx-affine",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revisi... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-fp8 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
},
{... | {
"link_type": "repository",
"path": "3-bit/",
"reason": null,
"repository": "orcarouter/GLM-5.3-Flash-MLX",
"revision": "c80f6810b1a95b5be9042761becc6aa78d189782",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
} | artifact--orcarouter.glm-5.3-flash-mlx-3bit | quant | model--zai-org.glm-5.3-flash | orcarouter GLM-5.3-Flash-MLX 3-bit | {
"handle": "orcarouter",
"is_registry_maintainer": false,
"name": "orcarouter",
"role": "quantizer",
"url": "https://huggingface.co/orcarouter"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
... | attn.o=unknown:unknown|attn.qkv=unknown:unknown|embed_tokens=unknown:unknown|lm_head=unknown:unknown|mlp.down=quantized:mlx-affine@3|mlp.gate=quantized:mlx-affine@3|mlp.up=quantized:mlx-affine@3|moe.experts=quantized:mlx-affine@3|mtp=quantized:mlx-affine@3|head=unknown|kv=unknown | {
"sealed": false
} | [
{
"kind": "model_card",
"uri": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
},
{
"kind": "url",
"note": "per-subfolder byte totals read from the Hub API",
"uri": "https://huggingface.co/api/models/orcarouter/GLM-5.3-Flash-MLX?blobs=true"
}
] | {
"container": "mlx",
"precision_label": "3-bit",
"size_basis": "repo_weight_files",
"size_bytes": 184306492535,
"size_gb": 184.306492535
} |
{
"status": "public",
"uri": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "mlx-affine",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revisi... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-fp8 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
},
{... | {
"link_type": "repository",
"path": "4-bit/",
"reason": null,
"repository": "orcarouter/GLM-5.3-Flash-MLX",
"revision": "c80f6810b1a95b5be9042761becc6aa78d189782",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
} | artifact--orcarouter.glm-5.3-flash-mlx-4bit | quant | model--zai-org.glm-5.3-flash | orcarouter GLM-5.3-Flash-MLX 4-bit | {
"handle": "orcarouter",
"is_registry_maintainer": false,
"name": "orcarouter",
"role": "quantizer",
"url": "https://huggingface.co/orcarouter"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
... | attn.o=unknown:unknown|attn.qkv=unknown:unknown|embed_tokens=unknown:unknown|lm_head=unknown:unknown|mlp.down=quantized:mlx-affine@4|mlp.gate=quantized:mlx-affine@4|mlp.up=quantized:mlx-affine@4|moe.experts=quantized:mlx-affine@4|mtp=quantized:mlx-affine@4|head=unknown|kv=unknown | {
"sealed": false
} | [
{
"kind": "model_card",
"uri": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
},
{
"kind": "url",
"note": "per-subfolder byte totals read from the Hub API",
"uri": "https://huggingface.co/api/models/orcarouter/GLM-5.3-Flash-MLX?blobs=true"
}
] | {
"container": "mlx",
"precision_label": "4-bit",
"size_basis": "repo_weight_files",
"size_bytes": 204023874943,
"size_gb": 204.023874943
} |
{
"status": "public",
"uri": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 6,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "mlx-affine",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revisi... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-fp8 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
},
{... | {
"link_type": "repository",
"path": "6-bit/",
"reason": null,
"repository": "orcarouter/GLM-5.3-Flash-MLX",
"revision": "c80f6810b1a95b5be9042761becc6aa78d189782",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
} | artifact--orcarouter.glm-5.3-flash-mlx-6bit | quant | model--zai-org.glm-5.3-flash | orcarouter GLM-5.3-Flash-MLX 6-bit | {
"handle": "orcarouter",
"is_registry_maintainer": false,
"name": "orcarouter",
"role": "quantizer",
"url": "https://huggingface.co/orcarouter"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
... | attn.o=unknown:unknown|attn.qkv=unknown:unknown|embed_tokens=unknown:unknown|lm_head=unknown:unknown|mlp.down=quantized:mlx-affine@6|mlp.gate=quantized:mlx-affine@6|mlp.up=quantized:mlx-affine@6|moe.experts=quantized:mlx-affine@6|mtp=quantized:mlx-affine@6|head=unknown|kv=unknown | {
"sealed": false
} | [
{
"kind": "model_card",
"uri": "https://huggingface.co/orcarouter/GLM-5.3-Flash-MLX"
},
{
"kind": "url",
"note": "per-subfolder byte totals read from the Hub API",
"uri": "https://huggingface.co/api/models/orcarouter/GLM-5.3-Flash-MLX?blobs=true"
}
] | {
"container": "mlx",
"precision_label": "6-bit",
"size_basis": "repo_weight_files",
"size_bytes": 295627484591,
"size_gb": 295.627484591
} |
{
"status": "public",
"uri": "https://huggingface.co/Qwen/Qwen3.8-27B"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": true,
"code": "revision_unpinned",
"detail": "No measurement receipt for this artifact records a Hub revision. Every kld5 receipt records model_revision=null / model_revision_source='none'. Identity rests on index_sha256 and the per-shard sha256 map the receipt carries.",
... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "Qwen/Qwen3.8-27B",
"revision": null,
"revision_source": "none",
"status": "known",
"url": "https://huggingface.co/Qwen/Qwen3.8-27B"
} | artifact--qwen.qwen3.8-27b-bf16 | base | model--qwen.qwen3.8-27b | Qwen3.8-27B BF16 | {
"handle": "Qwen",
"is_registry_maintainer": false,
"name": "Qwen (Alibaba)",
"role": "model-publisher",
"url": "https://huggingface.co/Qwen"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor... | attn.o=native:bf16|attn.qkv=native:bf16|embed_tokens=native:bf16|lm_head=native:bf16|mlp.down=native:bf16|mlp.gate=native:bf16|mlp.up=native:bf16|moe.experts=native:bf16|norm=native:bf16|head=native|kv=bf16 | {
"sealed": false
} | [
{
"kind": "github_file",
"note": "reference_identity index_sha256 77042094..., config_sha256 191e0af2...",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/kld5-10M-fp8.json"
}
] | {
"config_sha256": "191e0af232104ed8b65258cf3fb2b842e288008baca7633c11b82a1ac7203aab",
"container": "safetensors",
"index_sha256": "77042094076611b69791a610065f28b7013b8c621795fa86ddccc8bac7d1b9df",
"precision_label": "BF16",
"size_basis": "unknown",
"size_bytes": null,
"size_gb": null
} |
{
"status": "public",
"uri": "https://huggingface.co/Qwen/Qwen3.8-27B-FP8"
} | {
"bits_per_weight_effective": 8,
"bits_per_weight_nominal": 8,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "fp8_e4m3",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": true,
"code": "revision_unpinned",
"detail": "No measurement receipt for this artifact records a Hub revision. Every kld5 receipt records model_revision=null / model_revision_source='none'. Identity rests on index_sha256 and the per-shard sha256 map the receipt carries.",
... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "Qwen/Qwen3.8-27B-FP8",
"revision": null,
"revision_source": "none",
"status": "known",
"url": "https://huggingface.co/Qwen/Qwen3.8-27B-FP8"
} | artifact--qwen.qwen3.8-27b-fp8 | quant | model--qwen.qwen3.8-27b | Qwen3.8-27B FP8 (official) | {
"handle": "Qwen",
"is_registry_maintainer": false,
"name": "Qwen (Alibaba)",
"role": "quantizer",
"url": "https://huggingface.co/Qwen"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 8,
"format": "fp8_e4m3",
"layer_range": "all",
"tenso... | attn.o=quantized:fp8_e4m3@8|attn.qkv=quantized:fp8_e4m3@8|embed_tokens=native:bf16|lm_head=native:bf16|mlp.down=quantized:fp8_e4m3@8|mlp.gate=quantized:fp8_e4m3@8|mlp.up=quantized:fp8_e4m3@8|norm=native:bf16|head=native|kv=bf16 | {
"sealed": false
} | [
{
"kind": "github_file",
"note": "candidate_identity index_sha256 f0838c76..., config_sha256 74227dd6...",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/kld5-10M-fp8.json"
}
] | {
"container": "safetensors",
"index_sha256": "f0838c766951bdfe76d6afbdb2771a8f67aaa2231dedb3d33cebd817729843a2",
"precision_label": "FP8",
"size_basis": "repo_all_files",
"size_bytes": 30890049597,
"size_gb": 30.890049597
} |
{
"status": "public",
"uri": "https://huggingface.co/Qwen/Qwen3.8-27B/tree/1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | null | null | [
{
"affects_comparability": false,
"asserts_provenance": true,
"code": "record_note",
"detail": "Native HF Jobs root. Receipt integrity and qualification were reviewed; the registry owner did not independently reproduce this run.",
"severity": "info",
"sources": [
{
"kind": "hf_... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "Qwen/Qwen3.8-27B",
"revision": "1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/Qwen/Qwen3.8-27B"
} | artifact--qwen.qwen3.8-27b.1d4bf0f2ff60 | base | model--qwen.qwen3.8-27b.1d4bf0f2ff60 | Qwen3.8-27B native BF16 | {
"handle": "Qwen",
"is_registry_maintainer": false,
"name": "Qwen",
"role": "model-publisher",
"url": "https://huggingface.co/Qwen"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | null | [
{
"kind": "hf_file",
"sha256": "5dc5c46a8a1418349a8beddfd1724f030b71307e72817d335ebf8d9a48170b34",
"uri": "https://huggingface.co/datasets/malaiwah/qfs-evidence-6c6b45736e86851500161bfeb9f44dd8-public/resolve/395ac358b1882d14877a02a12b647e97ae6abd03/first/capture/manifest.json"
},
{
"kind": "hf_... | {
"config_sha256": "191e0af232104ed8b65258cf3fb2b842e288008baca7633c11b82a1ac7203aab",
"container": "safetensors",
"index_sha256": null,
"precision_label": "BF16",
"shard_count": 18,
"shard_sha256": {
"model-00001-of-00018.safetensors": "ba0ce20aae489ad196733da5064bcdf159a1fe84f53336648196e1ebb7751b1c",... |
{
"status": "public",
"uri": "https://huggingface.co/RadixArk/GLM-5.3-NVFP4"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "nvfp4",
"group_size": 16,
"quantizer": {
"pipeline_ref": null,
"revision": nu... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "Read from bytes (the index plus a ranged read of every shard header): the 57,600 routed-expert matrices of layers 3-77 are NVFP4 modelopt component sets -- e2m1 weight U8 [out, in/2], weight_scale F8_E4M3 [out, in/16] (group 16 along th... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "RadixArk/GLM-5.3-NVFP4",
"revision": "11af4cba759e6559eda70358a5778bd1bddddd78",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/RadixArk/GLM-5.3-NVFP4"
} | artifact--radixark.glm-5.3-nvfp4 | quant | model--zai-org.glm-5.3 | RadixArk GLM-5.3-NVFP4 (routed experts NVFP4 e2m1 group 16, rest native) | {
"handle": "RadixArk",
"is_registry_maintainer": false,
"name": "RadixArk",
"role": "quantizer",
"url": "https://huggingface.co/RadixArk"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "78 tensors: model.layers.N.self_attn.o_proj.weight. read from RadixArk/GLM-5.3-NVFP4@11af4cba759e6559eda70358a5778bd1bddddd78 model.safetensors.index.json + every... | attn.o=native:bf16@16|attn.other=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:nvfp4@4|moe.router=native:mixed|moe.shared_expert=native:bf16@16|mtp=native:mixed|norm=native:bf16@16|hea... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision 11af4cba759e; 47 shards; config.json sha256 1dd7f73e... declares quantization_config quant_method modelopt / quant_algo NVFP4 (producer modelopt 0.47.0.dev91+g7ff81dd79); index sha256 2aa8397b...",
"uri": "https://huggingface.co/RadixArk/GLM-5.3-NVFP4"
},
{
... | {
"config_sha256": "1dd7f73e24d6a980c005b7faaa6fcfbc62c74647c792b4d236cd0fa55d772a77",
"container": "safetensors",
"index_sha256": "2aa8397b501d9f6a232d153f328feb912f813c389061aac4cf72b04914fa5b74",
"precision_label": "NVFP4",
"shard_count": 47,
"size_basis": "repo_weight_files",
"size_bytes": 46482304209... |
{
"status": "public",
"uri": "https://huggingface.co/turboderp/GLM-5.3-Flash-exl3"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 2.05,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": 512000,
"used": true
},
"family": "exl3-mul1",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"re... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-fp8 | [
{
"affects_comparability": true,
"code": "unsealed_source",
"detail": "Stock exllamav3 releases ship no upstream receipts, reconstruction closures or sealed reader ABI. The packed surface was decoded WITHOUT seal verification; the immutable repo revision and the artifact's own quantization_config sha256... | {
"link_type": "repository",
"path": "branch 2.05bpw",
"reason": null,
"repository": "turboderp/GLM-5.3-Flash-exl3",
"revision": "51058cd551c7e570d87bd32a4adee720edce2349",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/turboderp/GLM-5.3-Flash-exl3"
} | artifact--turboderp.glm-5.3-flash-exl3-2.05bpw | quant | model--zai-org.glm-5.3-flash | turboderp GLM-5.3-Flash EXL3 2.05bpw (stock exllamav3, mul1, quantized head at 5 bits) | {
"handle": "turboderp",
"is_registry_maintainer": false,
"name": "turboderp",
"role": "quantizer",
"url": "https://huggingface.co/turboderp"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "read from the release's OWN quantization_config.json @ 51058cd551c7e570d87bd32a4adee720edce2349 (47,905,719 bytes, sha256 22a0eb34458b7a5951a5f20aa30290b6cd330085... | attn.o=quantized:exl3-mul1@4|attn.other=native:fp16@16|attn.qkv=quantized:exl3-mul1@4|embed_tokens=native:bf16@16|lm_head=quantized:exl3-mul1@5|mlp.down=quantized:exl3-mul1@3|mlp.gate=quantized:exl3-mul1@3|mlp.up=quantized:exl3-mul1@3|moe.experts=quantized:exl3-mul1@2|moe.router=native:fp32@32|moe.shared_expert=quantiz... | {
"note": "unsealed source; see the unsealed_source disclosure",
"sealed": false
} | [
{
"kind": "url",
"note": "24 files at revision 51058cd5; all-files sum 85,233,484,348",
"uri": "https://huggingface.co/api/models/turboderp/GLM-5.3-Flash-exl3?blobs=true"
},
{
"kind": "hf_file",
"note": "47,905,719 bytes, one tensor_storage entry per module stating bits_per_weight; the heade... | {
"container": "exl3",
"precision_label": "2.05 bpw",
"shard_count": 19,
"size_basis": "repo_all_files",
"size_bytes": 85233484348,
"size_gb": 85.233484348,
"tensor_parallel": {
"pre_sliced": false,
"world_size": null
}
} |
{
"status": "public",
"uri": "https://huggingface.co/turboderp/GLM-5.3-Flash-exl3"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4.05,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": 512000,
"used": true
},
"family": "exl3-mul1",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"re... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-fp8 | [
{
"affects_comparability": true,
"code": "unsealed_source",
"detail": "Stock exllamav3 releases ship no upstream receipts, reconstruction closures or sealed reader ABI. The packed surface was decoded WITHOUT seal verification; the immutable repo revision, the artifact's own config/index sha256 and the c... | {
"link_type": "repository",
"path": "branch 4.05bpw",
"reason": null,
"repository": "turboderp/GLM-5.3-Flash-exl3",
"revision": "2a30229e67012798ba9f0cd832bb78abf4c363d5",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/turboderp/GLM-5.3-Flash-exl3"
} | artifact--turboderp.glm-5.3-flash-exl3-4.05bpw | quant | model--zai-org.glm-5.3-flash | turboderp GLM-5.3-Flash EXL3 4.05bpw (stock exllamav3, mul1, quantized head) | {
"handle": "turboderp",
"is_registry_maintainer": false,
"name": "turboderp",
"role": "quantizer",
"url": "https://huggingface.co/turboderp"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "read from the release's OWN quantization_config.json @ 2a30229e (47.9 MB, one tensor_storage entry per module): bits = trellis.shape[-1]//16, verified identical w... | attn.o=quantized:exl3-mul1@6|attn.other=native:fp16@16|attn.qkv=quantized:exl3-mul1@6|embed_tokens=native:bf16@16|lm_head=quantized:exl3-mul1@6|mlp.down=quantized:exl3-mul1@5|mlp.gate=quantized:exl3-mul1@5|mlp.up=quantized:exl3-mul1@5|moe.experts=quantized:exl3-mul1@4|moe.router=native:fp32@32|moe.shared_expert=quantiz... | {
"note": "unsealed source; see the unsealed_source disclosure",
"sealed": false
} | [
{
"kind": "url",
"note": "31 files at revision 2a30229e; 19 safetensors shards + mtp.safetensors; all-files sum 165,151,543,361",
"uri": "https://huggingface.co/api/models/turboderp/GLM-5.3-Flash-exl3?blobs=true"
},
{
"kind": "hf_file",
"note": "47.9 MB, one tensor_storage entry per module: ... | {
"container": "exl3",
"precision_label": "4.05 bpw",
"shard_count": 19,
"size_basis": "repo_all_files",
"size_bytes": 165151543361,
"size_gb": 165.151543361,
"tensor_parallel": {
"pre_sliced": false,
"world_size": null
}
} |
{
"status": "public",
"uri": "https://huggingface.co/turboderp/Qwen3.8-27B-exl3"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 5,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
},
{... | {
"link_type": "repository",
"path": "branch:5.00bpw",
"reason": null,
"repository": "turboderp/Qwen3.8-27B-exl3",
"revision": "a35e75a73baee51da709329d19294245cbeeb5d8",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/turboderp/Qwen3.8-27B-exl3"
} | artifact--turboderp.qwen3.8-27b-exl3.5bpw | quant | model--qwen.qwen3.8-27b | turboderp Qwen3.8-27B exl3 5.00bpw | {
"handle": "turboderp",
"is_registry_maintainer": false,
"name": "turboderp",
"role": "quantizer",
"url": "https://huggingface.co/turboderp"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
... | attn.o=unknown:unknown|attn.qkv=unknown:unknown|embed_tokens=unknown:unknown|lm_head=unknown:unknown|mlp.down=quantized:exl3-mcg@5|mlp.gate=quantized:exl3-mcg@5|mlp.up=quantized:exl3-mcg@5|moe.experts=quantized:exl3-mcg@5|head=unknown|kv=bf16 | {
"sealed": false
} | [
{
"kind": "github_file",
"note": "candidate shard sha256 map recorded in the receipt",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/kld5-1M-turbo5.json"
},
{
"kind": "url",
"uri": "https://huggingface.co/turboderp/Qwen3.... | {
"container": "exl3",
"precision_label": "5.00bpw",
"size_basis": "repo_all_files",
"size_bytes": 19925543918,
"size_gb": 19.925543918
} |
{
"status": "public",
"uri": "https://huggingface.co/turboderp/Qwen3.8-27B-exl3"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 6,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
},
{... | {
"link_type": "repository",
"path": "branch:6.00bpw",
"reason": null,
"repository": "turboderp/Qwen3.8-27B-exl3",
"revision": "d32ba0bbd17de6bed8d5bbfb8c19f16f228f67ff",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/turboderp/Qwen3.8-27B-exl3"
} | artifact--turboderp.qwen3.8-27b-exl3.6bpw | quant | model--qwen.qwen3.8-27b | turboderp Qwen3.8-27B exl3 6.00bpw | {
"handle": "turboderp",
"is_registry_maintainer": false,
"name": "turboderp",
"role": "quantizer",
"url": "https://huggingface.co/turboderp"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
... | attn.o=unknown:unknown|attn.qkv=unknown:unknown|embed_tokens=unknown:unknown|lm_head=unknown:unknown|mlp.down=quantized:exl3-mcg@6|mlp.gate=quantized:exl3-mcg@6|mlp.up=quantized:exl3-mcg@6|moe.experts=quantized:exl3-mcg@6|head=unknown|kv=bf16 | {
"sealed": false
} | [
{
"kind": "github_file",
"note": "candidate shard sha256 map recorded in the receipt",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/kld5-1M-turbo6.json"
},
{
"kind": "url",
"uri": "https://huggingface.co/turboderp/Qwen3.... | {
"container": "exl3",
"precision_label": "6.00bpw",
"size_basis": "repo_all_files",
"size_bytes": 22966414310,
"size_gb": 22.96641431
} |
{
"status": "unknown",
"uri": null
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "awq",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
},
{... | {
"link_type": "none",
"path": null,
"reason": "The measurement receipt records only a local path (/models/Qwen3.8-27B-AWQ-INT4) and model_revision=null. An upstream repository id circulates in our own landscape notes but is not recorded by any receipt, so it is NOT asserted here.",
"repository": null,
"revis... | artifact--unattributed.qwen3.8-27b-awq-int4 | quant | model--qwen.qwen3.8-27b | Qwen3.8-27B AWQ-INT4 (upstream unattributed) | {
"handle": null,
"is_registry_maintainer": false,
"name": "unknown",
"role": "quantizer",
"url": null
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
... | attn.o=unknown:unknown|attn.qkv=unknown:unknown|embed_tokens=unknown:unknown|lm_head=unknown:unknown|mlp.down=quantized:awq@4|mlp.gate=quantized:awq@4|mlp.up=quantized:awq@4|moe.experts=quantized:awq@4|head=unknown|kv=bf16 | {
"sealed": false
} | [
{
"kind": "github_file",
"note": "candidate_identity.model_path /models/Qwen3.8-27B-AWQ-INT4, model_revision null",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/kld5-1M-awq.json"
}
] | {
"container": "safetensors",
"precision_label": "INT4",
"shard_sha256": {},
"size_basis": "unknown",
"size_bytes": null,
"size_gb": null
} |
{
"status": "unknown",
"uri": null
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "nvfp4",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
},
{... | {
"link_type": "none",
"path": null,
"reason": "The measurement receipt records only a local path (/models/Qwen3.8-27B-MTP-NVFP4) and model_revision=null. An upstream repository id circulates in our own landscape notes but is not recorded by any receipt, so it is NOT asserted here.",
"repository": null,
"revi... | artifact--unattributed.qwen3.8-27b-mtp-nvfp4 | quant | model--qwen.qwen3.8-27b | Qwen3.8-27B MTP-NVFP4 (upstream unattributed) | {
"handle": null,
"is_registry_maintainer": false,
"name": "unknown",
"role": "quantizer",
"url": null
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
... | attn.o=unknown:unknown|attn.qkv=unknown:unknown|embed_tokens=unknown:unknown|lm_head=unknown:unknown|mlp.down=quantized:nvfp4@4|mlp.gate=quantized:nvfp4@4|mlp.up=quantized:nvfp4@4|moe.experts=quantized:nvfp4@4|head=unknown|kv=bf16 | {
"sealed": false
} | [
{
"kind": "github_file",
"note": "candidate_identity.model_path /models/Qwen3.8-27B-MTP-NVFP4, model_revision null",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/kld5-1M-saka.json"
}
] | {
"container": "safetensors",
"precision_label": "NVFP4",
"shard_sha256": {
"model-mtp-bf16.safetensors": "90fa0e3eed5a647c035c6df9ecabc416c0f8d573ff84ac12485b085f00a7cdf2",
"model.safetensors": "0e1597fc7835a5a7578243809420f88ae06732733a716e49629392e571a62f76"
},
"size_basis": "unknown",
"size_byte... |
{
"status": "public",
"uri": "https://huggingface.co/unsloth/GLM-5.2-GGUF"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "gguf-k-quant",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revi... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "MEASURED BITS, NOT THE BUILD NAME. The 225 routed-expert tensor groups of layers 3-77 are Q4_K x148 / Q5_K x73 / Q6_K x4 -- 4.8611 measured bits/weight over 724,775,731,200 weights, not the 4 the label suggests -- while attention, the d... | {
"link_type": "repository",
"path": "subdir UD-Q4_K_XL",
"reason": null,
"repository": "unsloth/GLM-5.2-GGUF",
"revision": "abc55e72527792c6e77069c99b4cb7de16fa9f23",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/unsloth/GLM-5.2-GGUF"
} | artifact--unsloth.glm-5.2-gguf.ud-q4-k-xl | quant | model--zai-org.glm-5.2 | unsloth GLM-5.2-GGUF UD-Q4_K_XL (llama.cpp k-quant build, mixed per tensor) | {
"handle": "unsloth",
"is_registry_maintainer": false,
"name": "unsloth",
"role": "quantizer",
"url": "https://huggingface.co/unsloth"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 8.5,
"format": "gguf-k-quant",
"layer_range": "0-77",
"note": "78 GGUF tensors: layers.N.self_attn.o_proj.weight. ggml types: Q8_0 x78. MEASURED 8.5000 bits/weight over 7851737088 weights. read from unsloth/GLM-5.2-... | attn.o=quantized:gguf-k-quant@8.5|attn.other=quantized:mixed|attn.qkv=quantized:gguf-k-quant@8.5|embed_tokens=quantized:gguf-k-quant@8.5|lm_head=quantized:gguf-k-quant@8.5|mlp.down=quantized:gguf-k-quant@8.5|mlp.gate=quantized:gguf-k-quant@8.5|mlp.up=quantized:gguf-k-quant@8.5|moe.experts=quantized:gguf-k-quant@4.8611|... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision abc55e725277, subdirectory UD-Q4_K_XL: 11 GGUF files, 467,289,111,904 bytes, architecture glm-dsa. The build ships no config.json of its own, so the lane bound the official zai-org/GLM-5.2 config (sha256 185f93ee..., the digest the sealed dataset records).",
"ur... | {
"container": "gguf",
"precision_label": "UD-Q4_K_XL",
"shard_count": 11,
"size_basis": "repo_weight_files",
"size_bytes": 467289111904,
"size_gb": 467.289111904
} |
{
"status": "public",
"uri": "https://huggingface.co/unsloth/GLM-5.3-GGUF"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "gguf-k-quant",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revi... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "MEASURED BITS, NOT THE BUILD NAME. 'UD-Q4_K_XL' is a recipe label; the scope records what the tensor tables actually hold. The 225 routed-expert tensor groups of layers 3-77 are Q4_K x148 / Q5_K x73 / Q6_K x4 -- 4.8611 measured bits/wei... | {
"link_type": "repository",
"path": "subdir UD-Q4_K_XL",
"reason": null,
"repository": "unsloth/GLM-5.3-GGUF",
"revision": "346b3591c7f28d1a23716f97a065ecf12ec14771",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/unsloth/GLM-5.3-GGUF"
} | artifact--unsloth.glm-5.3-gguf.ud-q4-k-xl | quant | model--zai-org.glm-5.3 | unsloth GLM-5.3-GGUF UD-Q4_K_XL (llama.cpp k-quant build, mixed per tensor) | {
"handle": "unsloth",
"is_registry_maintainer": false,
"name": "unsloth",
"role": "quantizer",
"url": "https://huggingface.co/unsloth"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 8.5,
"format": "gguf-k-quant",
"layer_range": "0-77",
"note": "78 GGUF tensors: layers.N.self_attn.o_proj.weight. ggml types: Q8_0 x78. MEASURED 8.5000 bits/weight over 7851737088 weights. read from unsloth/GLM-5.3-... | attn.o=quantized:gguf-k-quant@8.5|attn.other=quantized:mixed|attn.qkv=quantized:gguf-k-quant@8.5|embed_tokens=quantized:gguf-k-quant@8.5|lm_head=quantized:gguf-k-quant@8.5|mlp.down=quantized:gguf-k-quant@8.5|mlp.gate=quantized:gguf-k-quant@8.5|mlp.up=quantized:gguf-k-quant@8.5|moe.experts=quantized:gguf-k-quant@4.8611|... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision 346b3591c7f2, subdirectory UD-Q4_K_XL: 11 GGUF files, 467,289,116,837 bytes, 1,809 tensors, architecture glm-dsa. The build ships no config.json of its own, so the lane bound the official zai-org/GLM-5.3-BF16 config (sha256 ca8f2f47..., the digest the sealed dataset... | {
"container": "gguf",
"precision_label": "UD-Q4_K_XL",
"shard_count": 11,
"size_basis": "repo_weight_files",
"size_bytes": 467289116837,
"size_gb": 467.289116837
} |
{
"status": "public",
"uri": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "The unquantized BF16 GGUF. It exists in this registry only as the CROSS-ENGINE FLOOR: what llama.cpp and vLLM disagree by on identical unquantized weights.",
"severity": "info"
}
] | {
"link_type": "repository",
"path": "Qwen3.8-27B-BF16-00001-of-00002.gguf",
"reason": null,
"repository": "unsloth/Qwen3.8-27B-GGUF",
"revision": "f1bfb127c64f7072bdd2cad55f258b9c8b2910fe",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
} | artifact--unsloth.qwen3.8-27b-gguf.bf16 | base | model--qwen.qwen3.8-27b | unsloth Qwen3.8-27B-GGUF BF16 | {
"handle": "unsloth",
"is_registry_maintainer": false,
"name": "unsloth",
"role": "quantizer",
"url": "https://huggingface.co/unsloth"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor... | attn.o=native:bf16|attn.qkv=native:bf16|embed_tokens=native:bf16|lm_head=native:bf16|mlp.down=native:bf16|mlp.gate=native:bf16|mlp.up=native:bf16|moe.experts=native:bf16|norm=native:bf16|head=native|kv=bf16 | {
"sealed": false
} | [
{
"kind": "github_file",
"note": "candidate_identity.shard_sha256 pins the exact .gguf file",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/gguf-report-engine-floor.json"
}
] | {
"container": "gguf",
"precision_label": "BF16",
"shard_count": 2,
"shard_sha256": {
"Qwen3.8-27B-BF16-00001-of-00002.gguf": "b9966e82b7a4d87028b5eae061d578ee826305ebf8baea5bfc6e09bad0ba191f"
},
"size_basis": "repo_weight_files",
"size_bytes": 54657735616,
"size_gb": 54.657735616
} |
{
"status": "public",
"uri": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 6,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "gguf-k-quant",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
},
{... | {
"link_type": "repository",
"path": "Qwen3.8-27B-Q6_K.gguf",
"reason": null,
"repository": "unsloth/Qwen3.8-27B-GGUF",
"revision": "f1bfb127c64f7072bdd2cad55f258b9c8b2910fe",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
} | artifact--unsloth.qwen3.8-27b-gguf.q6-k | quant | model--qwen.qwen3.8-27b | unsloth Qwen3.8-27B-GGUF Q6_K | {
"handle": "unsloth",
"is_registry_maintainer": false,
"name": "unsloth",
"role": "quantizer",
"url": "https://huggingface.co/unsloth"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
... | attn.o=unknown:unknown|attn.qkv=unknown:unknown|embed_tokens=unknown:unknown|lm_head=unknown:unknown|mlp.down=quantized:gguf-k-quant@6|mlp.gate=quantized:gguf-k-quant@6|mlp.up=quantized:gguf-k-quant@6|moe.experts=quantized:gguf-k-quant@6|head=unknown|kv=bf16 | {
"sealed": false
} | [
{
"kind": "github_file",
"note": "candidate_identity.shard_sha256 pins the exact .gguf file",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/gguf-report-q6_k.json"
}
] | {
"container": "gguf",
"precision_label": "Q6_K",
"shard_count": 1,
"shard_sha256": {
"Qwen3.8-27B-Q6_K.gguf": "562fbf760503008f118e5df38de5b3e97992d1f693f475815631198547486727"
},
"size_basis": "repo_weight_files",
"size_bytes": 22884408288,
"size_gb": 22.884408288
} |
{
"status": "public",
"uri": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 8,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "gguf-k-quant",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
},
{... | {
"link_type": "repository",
"path": "Qwen3.8-27B-Q8_0.gguf",
"reason": null,
"repository": "unsloth/Qwen3.8-27B-GGUF",
"revision": "f1bfb127c64f7072bdd2cad55f258b9c8b2910fe",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
} | artifact--unsloth.qwen3.8-27b-gguf.q8-0 | quant | model--qwen.qwen3.8-27b | unsloth Qwen3.8-27B-GGUF Q8_0 | {
"handle": "unsloth",
"is_registry_maintainer": false,
"name": "unsloth",
"role": "quantizer",
"url": "https://huggingface.co/unsloth"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
... | attn.o=unknown:unknown|attn.qkv=unknown:unknown|embed_tokens=unknown:unknown|lm_head=unknown:unknown|mlp.down=quantized:gguf-k-quant@8|mlp.gate=quantized:gguf-k-quant@8|mlp.up=quantized:gguf-k-quant@8|moe.experts=quantized:gguf-k-quant@8|head=unknown|kv=bf16 | {
"sealed": false
} | [
{
"kind": "github_file",
"note": "candidate_identity.shard_sha256 pins the exact .gguf file",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/gguf-report-q8_0.json"
}
] | {
"container": "gguf",
"precision_label": "Q8_0",
"shard_count": 1,
"shard_sha256": {
"Qwen3.8-27B-Q8_0.gguf": "a680f44a06920e5d689774823782006aa3acc8db95750323373b24139b67e348"
},
"size_basis": "repo_weight_files",
"size_bytes": 29047086048,
"size_gb": 29.047086048
} |
{
"status": "public",
"uri": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 5,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "gguf-k-quant",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
},
{... | {
"link_type": "repository",
"path": "Qwen3.8-27B-UD-Q5_K_XL.gguf",
"reason": null,
"repository": "unsloth/Qwen3.8-27B-GGUF",
"revision": "f1bfb127c64f7072bdd2cad55f258b9c8b2910fe",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/unsloth/Qwen3.8-27B-GGUF"
} | artifact--unsloth.qwen3.8-27b-gguf.ud-q5-k-xl | quant | model--qwen.qwen3.8-27b | unsloth Qwen3.8-27B-GGUF UD-Q5_K_XL | {
"handle": "unsloth",
"is_registry_maintainer": false,
"name": "unsloth",
"role": "quantizer",
"url": "https://huggingface.co/unsloth"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
... | attn.o=unknown:unknown|attn.qkv=unknown:unknown|embed_tokens=unknown:unknown|lm_head=unknown:unknown|mlp.down=quantized:gguf-k-quant@5|mlp.gate=quantized:gguf-k-quant@5|mlp.up=quantized:gguf-k-quant@5|moe.experts=quantized:gguf-k-quant@5|head=unknown|kv=bf16 | {
"sealed": false
} | [
{
"kind": "github_file",
"note": "candidate_identity.shard_sha256 pins the exact .gguf file",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/gguf-report-q5_k_xl.json"
}
] | {
"container": "gguf",
"precision_label": "UD-Q5_K_XL",
"shard_count": 1,
"shard_sha256": {
"Qwen3.8-27B-UD-Q5_K_XL.gguf": "176a6a3f034e9cdc447c10cd00329fc9b31002e6589b9295f2ad4f1eefe0f6ab"
},
"size_basis": "repo_weight_files",
"size_bytes": 20218178624,
"size_gb": 20.218178624
} |
{
"status": "public",
"uri": "https://huggingface.co/unsloth/Qwen3.8-27B-NVFP4"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "nvfp4",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": ... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--qwen.qwen3.8-27b-bf16 | [
{
"affects_comparability": true,
"code": "artifact_identity_incomplete",
"detail": "The per-tensor-class quantization recipe for this artifact was never published, so scope.assignments records 'unknown' rather than a guessed allocation. Its scope_digest shows the gap.",
"severity": "caveat"
}
] | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "unsloth/Qwen3.8-27B-NVFP4",
"revision": "9c73e2daee1d0fd494ffbd1d8753f2174a953796",
"revision_source": "reported_by_author",
"status": "known",
"url": "https://huggingface.co/unsloth/Qwen3.8-27B-NVFP4"
} | artifact--unsloth.qwen3.8-27b-nvfp4 | quant | model--qwen.qwen3.8-27b | unsloth Qwen3.8-27B NVFP4 | {
"handle": "unsloth",
"is_registry_maintainer": false,
"name": "unsloth",
"role": "quantizer",
"url": "https://huggingface.co/unsloth"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "unknown"
},
{
"bits_per_weight": null,
"format": "unknown",
"layer_range": "all",
... | attn.o=unknown:unknown|attn.qkv=unknown:unknown|embed_tokens=unknown:unknown|lm_head=unknown:unknown|mlp.down=quantized:nvfp4@4|mlp.gate=quantized:nvfp4@4|mlp.up=quantized:nvfp4@4|moe.experts=quantized:nvfp4@4|head=unknown|kv=bf16 | {
"sealed": false
} | [
{
"kind": "github_file",
"note": "candidate shard sha256 c473512c... / 1d8268aa...",
"uri": "https://raw.githubusercontent.com/malaiwah/qwen38-27b-exl3/8558b8ca3bba028f852f4b53167b79b4cd552f93/receipts/kld5-10M-nvfp4.json"
}
] | {
"container": "safetensors",
"precision_label": "NVFP4",
"size_basis": "unknown",
"size_bytes": null,
"size_gb": null
} |
{
"status": "public",
"uri": "https://huggingface.co/vcruz305/GLM-5.3-Flash-EXL3-K2"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 2,
"calibration": {
"corpus": "not published: the release ships no calibration manifest, corpus digest or token count. An EXL3 fit is calibrated by construction, so this is 'undisclosed', not 'none'.",
"overlapping_panel_refs": [],
"overl... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": true,
"code": "unsealed_source",
"detail": "The release ships no upstream receipts, no reconstruction closures, no sealed reader ABI -- and, unlike every other third-party artifact measured on this panel, no per-shard digest list of its own either: there is no SHA256SUMS and n... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "vcruz305/GLM-5.3-Flash-EXL3-K2",
"revision": "1718dd403534fd369e82d676e55a50dc19630ffc",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/vcruz305/GLM-5.3-Flash-EXL3-K2"
} | artifact--vcruz305.glm-5.3-flash-exl3-k2 | quant | model--zai-org.glm-5.3-flash | vcruz305 GLM-5.3-Flash EXL3 K2 (stock-exllamav3 HF layout, mcg, routed experts only, native BF16 head) | {
"handle": "vcruz305",
"is_registry_maintainer": false,
"name": "vcruz305",
"role": "quantizer",
"url": "https://huggingface.co/vcruz305"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "NOT quantized: the quantized scope is the routed experts only. read from the release's OWN config.json quantization_config @ 1718dd40 (bits 2, codebook mcg, head_... | attn.o=native:bf16@16|attn.other=native:mixed|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:exl3-mcg@2|moe.router=native:fp32@32|moe.shared_expert=native:bf16@16|mtp=quantized:exl3-mcg@2|norm=native:... | {
"note": "unsealed source, and no producer digest list of any kind; see the unsealed_source disclosure for what binds the bytes",
"sealed": false
} | [
{
"kind": "url",
"note": "138 files, 120 safetensors; all-files sum 97,767,357,064; safetensors sum 97,728,721,536, which is also the index's own metadata.total_size. The weights were uploaded in one commit (a618e7ad, 2026-08-28); every later commit on this repo touches only the model card and the runtime/ ... | {
"container": "exl3",
"precision_label": "2.0 bpw",
"shard_count": 120,
"size_basis": "repo_all_files",
"size_bytes": 97767357064,
"size_gb": 97.767357064,
"tensor_parallel": {
"pre_sliced": false,
"world_size": null
}
} |
{
"status": "public",
"uri": "https://huggingface.co/willfalco/GLM-5.2-EXL3-TR3-3.25bpw"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 3.25,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": false,
"tokens": null,
"used": false
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"rev... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.2-bf16 | [
{
"affects_comparability": false,
"code": "tp_sliced_artifact",
"detail": "Read from bytes: the 57,600 routed-expert matrices are stored as four tensor-parallel rank shards each and were composed into whole weights in ascending rank order (tp_rank_payloads_composed in the sealed dataset); everything els... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "willfalco/GLM-5.2-EXL3-TR3-3.25bpw",
"revision": "a39d9254886044e621e9b6a2d5b40308548f12d9",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/willfalco/GLM-5.2-EXL3-TR3-3.25bpw"
} | artifact--willfalco.glm-5.2-exl3-tr3-3.25bpw | quant | model--zai-org.glm-5.2 | willfalco GLM-5.2-EXL3-TR3-3.25bpw (routed experts trellis mixed K, TP4 rank-sharded, rest native) | {
"handle": "willfalco",
"is_registry_maintainer": false,
"name": "willfalco",
"role": "quantizer",
"url": "https://huggingface.co/willfalco"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "78 tensors: model.layers.N.self_attn.o_proj.weight. read from willfalco/GLM-5.2-EXL3-TR3-3.25bpw@a39d9254886044e621e9b6a2d5b40308548f12d9 model.safetensors.index.... | attn.o=native:bf16@16|attn.other=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:exl3-mcg@3.25|moe.router=native:mixed|moe.shared_expert=native:bf16@16|mtp=quantized:mixed|norm=native:bf... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision a39d92548860; 81 shards; config.json sha256 d83e2d8d...; index sha256 f5dcd976...",
"uri": "https://huggingface.co/willfalco/GLM-5.2-EXL3-TR3-3.25bpw"
},
{
"kind": "github_file",
"note": "scope authored from the index bytes and shard headers by engin... | {
"config_sha256": "d83e2d8d96b6f36e94d896a05a104200d6674673daa02f788de710c1c0f94ba4",
"container": "exl3",
"index_sha256": "f5dcd976a64ca70808dd4d8bd3ad07e9610c8ca6c30e3a6ed77ddefdac4c1d21",
"precision_label": "3.25bpw",
"shard_count": 81,
"size_basis": "repo_weight_files",
"size_bytes": 339069245936,
... |
{
"status": "public",
"uri": "https://huggingface.co/wrldsuksgo2mars/GLM-5.3-EXL3-K4-v1"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 4,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "Read from bytes: the 57,600 routed-expert matrices (layers 3-77, 256 experts x 3 projections) are stock exllamav3 trellis payload groups {trellis, suh, svh, mcg} at K=4; every other 2-D projection -- attention, the three dense MLPs, the... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "wrldsuksgo2mars/GLM-5.3-EXL3-K4-v1",
"revision": "47af23347db743b4666d952e2eb48f2b01c3fede",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/wrldsuksgo2mars/GLM-5.3-EXL3-K4-v1"
} | artifact--wrldsuksgo2mars.glm-5.3-exl3-k4-v1 | quant | model--zai-org.glm-5.3 | wrldsuksgo2mars GLM-5.3 EXL3 K4 v1 (routed experts trellis K4, rest FP8) | {
"handle": "wrldsuksgo2mars",
"is_registry_maintainer": false,
"name": "wrldsuksgo2mars",
"role": "quantizer",
"url": "https://huggingface.co/wrldsuksgo2mars"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 8,
"format": "fp8_e4m3",
"layer_range": "all",
"note": "78 tensors: model.layers.N.self_attn.o_proj.weight. read from wrldsuksgo2mars/GLM-5.3-EXL3-K4-v1@47af23347db743b4666d952e2eb48f2b01c3fede model.safetensors.ind... | attn.o=quantized:fp8_e4m3@8|attn.other=quantized:mixed|attn.qkv=quantized:fp8_e4m3@8|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:fp8_e4m3@8|mlp.gate=quantized:fp8_e4m3@8|mlp.up=quantized:fp8_e4m3@8|moe.experts=quantized:exl3-mcg@4|moe.router=native:mixed|moe.shared_expert=quantized:fp8_e4m3@8|... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision 47af23347db7; 46 shards; config.json sha256 14b5c7ab... declares quantization_config {quant_method exl3, bits 4}; index sha256 6c0a0d5a...",
"uri": "https://huggingface.co/wrldsuksgo2mars/GLM-5.3-EXL3-K4-v1"
},
{
"kind": "github_file",
"note": "scope... | {
"config_sha256": "14b5c7ab15efe84146e89e4c510ddf0e9eca0754af995beba01e6e664d7a725b",
"container": "exl3",
"index_sha256": "6c0a0d5ade9cb8758b6cc412a570a954852d133ef89540bbeee88dc9bd1b565b",
"precision_label": "K4",
"shard_count": 46,
"size_basis": "repo_weight_files",
"size_bytes": 394023913872,
"size... |
{
"status": "public",
"uri": "https://huggingface.co/wrldsuksgo2mars/GLM-5.3-Flash-EXL3-K3-v1"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": 3,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": true
},
"family": "exl3-mcg",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | artifact--zai-org.glm-5.3-flash-bf16.a6c167b6 | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "Read from bytes: the 37,152 routed-expert matrices (layers 3-45 incl. the MTP block's, 288 experts x gate/up/down) are stock exllamav3 payload groups (trellis/suh/svh + an mcg marker each) at K=3; every non-routed tensor -- attention, d... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "wrldsuksgo2mars/GLM-5.3-Flash-EXL3-K3-v1",
"revision": "1e4abd26e4e1e8d58d81fbd557d6c4099352fe63",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/wrldsuksgo2mars/GLM-5.3-Flash-EXL3-K3-v1"
} | artifact--wrldsuksgo2mars.glm-5.3-flash-exl3-k3-v1 | quant | model--zai-org.glm-5.3-flash | wrldsuksgo2mars GLM-5.3-Flash EXL3 K3 v1 (routed experts trellis K3 mcg, rest bf16) | {
"handle": "wrldsuksgo2mars",
"is_registry_maintainer": false,
"name": "wrldsuksgo2mars",
"role": "quantizer",
"url": "https://huggingface.co/wrldsuksgo2mars"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"note": "45 tensors: model.language_model.layers.N.self_attn.o_proj.weight. read from wrldsuksgo2mars/GLM-5.3-Flash-EXL3-K3-v1@1e4abd26e4e1e8d58d81fbd557d6c4099352fe63 mod... | attn.o=native:bf16@16|attn.other=native:mixed|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=quantized:exl3-mcg@3|moe.router=native:mixed|moe.shared_expert=native:bf16@16|mtp=quantized:mixed|norm=native:bf16@16... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision 1e4abd26e4e1; 16 shards; config.json sha256 e445e044... declares quantization_config {quant_method exl3, bits 3, codebook mcg, module_include the routed experts of layers 3..45}; meta names quantizer gptqmodel 7.3.5 and source_revision f12e0fe1 of the BF16 release (... | {
"config_sha256": "e445e0443e7fce59943297323ff388e72c973f0e0a0f43b881be5ba55765e572",
"container": "exl3",
"index_sha256": "d751549235ef63d1954be328754e001c8e488795ed4c2ef6d5b0e4a2dc08f0dc",
"precision_label": "K3",
"shard_count": 16,
"size_basis": "repo_weight_files",
"size_bytes": 136686260192,
"size... |
{
"status": "public",
"uri": "https://huggingface.co/zai-org/GLM-5.2"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "Scope is the sealed root dataset's own scope block: every tensor bf16, lm_head [154880, 6144] bf16 with tensor content sha256 a012be05e771... The MTP block (layer index 78, 791 tensors) is present in the checkpoint and intentionally unu... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "zai-org/GLM-5.2",
"revision": "cf457fa734ab149ffef225f80893eb38c6ff5cdc",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/zai-org/GLM-5.2"
} | artifact--zai-org.glm-5.2-bf16 | base | model--zai-org.glm-5.2 | GLM-5.2 BF16 (the official full-precision release) | {
"handle": "zai-org",
"is_registry_maintainer": false,
"name": "Z.ai",
"role": "model-publisher",
"url": "https://huggingface.co/zai-org"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision cf457fa734ab; 282 shards; config.json sha256 185f93ee..., index sha256 5fd47a92..., every shard's sha256 in the root dataset's runtime/capture-runtime.json",
"uri": "https://huggingface.co/zai-org/GLM-5.2"
},
{
"kind": "dataset_card",
"note": "the re... | {
"config_sha256": "185f93ee6d12548e16a847e279dc0c3c90b1524c970b0866b42fb545747d859a",
"container": "safetensors",
"index_sha256": "5fd47a926aefce0f2c917f42523e5e0f3c87e23e389e767c3681536a62f5cf5e",
"precision_label": "BF16",
"shard_count": 282,
"size_basis": "repo_weight_files",
"size_bytes": 15066673874... |
{
"status": "public",
"uri": "https://huggingface.co/zai-org/GLM-5.2-FP8"
} | {
"bits_per_weight_effective": 8,
"bits_per_weight_nominal": 8,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "fp8_e4m3",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": ... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "Scope read from the weight index, not the README: every 2-D projection with a weight_scale_inv sibling is fp8_e4m3 (attention, the dense MLPs, all 57,600 routed expert matrices, the shared experts and 778 of the MTP block's tensors), wh... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "zai-org/GLM-5.2-FP8",
"revision": "f33c6dc501ee5a2c7e35155653b1b1abbc320951",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/zai-org/GLM-5.2-FP8"
} | artifact--zai-org.glm-5.2-fp8 | quant | model--zai-org.glm-5.2 | GLM-5.2 FP8 (the official block-scaled release) | {
"handle": "zai-org",
"is_registry_maintainer": false,
"name": "Z.ai",
"role": "quantizer",
"url": "https://huggingface.co/zai-org"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 8,
"format": "fp8_e4m3",
"layer_range": "all",
"note": "78 tensors: model.layers.N.self_attn.o_proj.weight. read from zai-org/GLM-5.2-FP8@f33c6dc501ee5a2c7e35155653b1b1abbc320951 model.safetensors.index.json: a tens... | attn.o=quantized:fp8_e4m3@8|attn.other=quantized:mixed|attn.qkv=quantized:fp8_e4m3@8|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:fp8_e4m3@8|mlp.gate=quantized:fp8_e4m3@8|mlp.up=quantized:fp8_e4m3@8|moe.experts=quantized:fp8_e4m3@8|moe.router=native:bf16@16|moe.shared_expert=quantized:fp8_e4m3@... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision f33c6dc501ee; 141 shards; config.json sha256 22e49334...; index sha256 e0fe7f28...",
"uri": "https://huggingface.co/zai-org/GLM-5.2-FP8"
},
{
"kind": "github_file",
"note": "scope authored from the checkpoint's own index bytes by engines/tools/fp8_sc... | {
"config_sha256": "22e49334abf8562fecf70ca3292ba3f5b33f5602fb2bf10b52dd64a66cfe65ff",
"container": "safetensors",
"index_sha256": "e0fe7f28c1f853d4824e4d796374e3dacf1fe470988773952c79b063768134bf",
"precision_label": "FP8",
"shard_count": 141,
"size_basis": "repo_weight_files",
"size_bytes": 755632050320... |
{
"status": "public",
"uri": "https://huggingface.co/zai-org/GLM-5.3-BF16"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "Scope is the sealed root dataset's own scope block: every tensor bf16, lm_head [154880, 6144] bf16 with tensor content sha256 864f488a0074... The MTP block (layer index 78, 791 tensors) is present in the checkpoint and intentionally unu... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "zai-org/GLM-5.3-BF16",
"revision": "304b8051cfb2b260b61ce0cbe330e02a98e73639",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/zai-org/GLM-5.3-BF16"
} | artifact--zai-org.glm-5.3-bf16 | base | model--zai-org.glm-5.3 | GLM-5.3 BF16 (the official full-precision release) | {
"handle": "zai-org",
"is_registry_maintainer": false,
"name": "Z.ai",
"role": "model-publisher",
"url": "https://huggingface.co/zai-org"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 16,
"format": "bf16",
"layer_range": "all",
"tensor_cla... | attn.o=native:bf16@16|attn.qkv=native:bf16@16|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=native:bf16@16|mlp.gate=native:bf16@16|mlp.up=native:bf16@16|moe.experts=native:bf16@16|norm=native:bf16@16|head=native|kv=bf16 | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision 304b8051cfb2; 282 shards; config.json sha256 ca8f2f47..., index sha256 5fd47a92..., every shard's sha256 in the root dataset's runtime/capture-runtime.json",
"uri": "https://huggingface.co/zai-org/GLM-5.3-BF16"
},
{
"kind": "dataset_card",
"note": "t... | {
"config_sha256": "ca8f2f47b07919a514c0ca223dc2ea2bc7445afaa5ac76c013a3784e096426ca",
"container": "safetensors",
"index_sha256": "5fd47a926aefce0f2c917f42523e5e0f3c87e23e389e767c3681536a62f5cf5e",
"precision_label": "BF16",
"shard_count": 282,
"size_basis": "repo_weight_files",
"size_bytes": 15066673874... |
{
"status": "public",
"uri": "https://huggingface.co/zai-org/GLM-5.3-Flash-BF16"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "The revision the K6 / Dione / brandonmusic chain pins as 'the BF16 teacher weights'.",
"severity": "info"
}
] | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "zai-org/GLM-5.3-Flash-BF16",
"revision": "a6c167b62691b2bac901344b65cb651a70f53e43",
"revision_source": "revision_txt",
"status": "known",
"url": "https://huggingface.co/zai-org/GLM-5.3-Flash-BF16"
} | artifact--zai-org.glm-5.3-flash-bf16.a6c167b6 | base | model--zai-org.glm-5.3-flash | GLM-5.3-Flash BF16 @a6c167b6 | {
"handle": "zai-org",
"is_registry_maintainer": false,
"name": "Z.ai",
"role": "model-publisher",
"url": "https://huggingface.co/zai-org"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor... | attn.o=native:bf16|attn.qkv=native:bf16|embed_tokens=native:bf16|lm_head=native:bf16|mlp.down=native:bf16|mlp.gate=native:bf16|mlp.up=native:bf16|moe.experts=native:bf16|norm=native:bf16|head=native|kv=bf16 | {
"sealed": false
} | [
{
"kind": "hf_file",
"note": "quant-pipeline.glm53-packed-kld-receipt.v1 pins source_revision a6c167b6...",
"sha256": "19766e5e9643dbe940c05deaee7c3085f9ee339553da35ead973c825adddfef2",
"uri": "https://huggingface.co/datasets/malaiwah/GLM-5.3-Flash-fidelity-suite-v1/resolve/main/reports/k6-packed-kl... | {
"config_sha256": "33e63ec7fe607658be712bd6dd3c16c6549960d8e7f0483d34b939881b55f943",
"container": "safetensors",
"index_sha256": "e6007bd58fb7e07f9fe69544257ee2713f252ef5855bbf685b48c991d524ef0f",
"precision_label": "BF16",
"size_basis": "unknown",
"size_bytes": null,
"size_gb": null
} |
{
"status": "public",
"uri": "https://huggingface.co/zai-org/GLM-5.3-Flash-BF16"
} | {
"bits_per_weight_effective": null,
"bits_per_weight_nominal": null,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "bf16",
"group_size": null
} | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "A DIFFERENT pinned revision of the same repository from artifact--zai-org.glm-5.3-flash-bf16.a6c167b6. index_sha256 and config_sha256 agree between the two; the Hub revisions do not. Both are kept as separate artifacts because a registr... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "zai-org/GLM-5.3-Flash-BF16",
"revision": "b1967181a3917ae70a437f4884748f6b8e3a1f4d",
"revision_source": "revision_txt",
"status": "known",
"url": "https://huggingface.co/zai-org/GLM-5.3-Flash-BF16"
} | artifact--zai-org.glm-5.3-flash-bf16.b1967181 | base | model--zai-org.glm-5.3-flash | GLM-5.3-Flash BF16 @b1967181 | {
"handle": "zai-org",
"is_registry_maintainer": false,
"name": "Z.ai",
"role": "model-publisher",
"url": "https://huggingface.co/zai-org"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor... | attn.o=native:bf16|attn.qkv=native:bf16|embed_tokens=native:bf16|lm_head=native:bf16|mlp.down=native:bf16|mlp.gate=native:bf16|mlp.up=native:bf16|moe.experts=native:bf16|norm=native:bf16|head=native|kv=bf16 | {
"sealed": false
} | [
{
"kind": "hf_file",
"note": "glm53flash-fidelity-report/2 reference_identity.model_revision = b1967181..., index_sha256 e6007bd5..., model_revision_source revision_txt",
"uri": "https://huggingface.co/datasets/malaiwah/GLM-5.3-Flash-fidelity-suite-v1/resolve/main/reports/report-fp8-vs-bf16.json"
}
] | {
"config_sha256": "33e63ec7fe607658be712bd6dd3c16c6549960d8e7f0483d34b939881b55f943",
"container": "safetensors",
"index_sha256": "e6007bd58fb7e07f9fe69544257ee2713f252ef5855bbf685b48c991d524ef0f",
"precision_label": "BF16",
"size_basis": "unknown",
"size_bytes": null,
"size_gb": null
} |
{
"status": "public",
"uri": "https://huggingface.co/zai-org/GLM-5.3-Flash"
} | {
"bits_per_weight_effective": 8,
"bits_per_weight_nominal": 8,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "fp8_e4m3",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": ... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": "glm-5.3-flash",
"model_instance_id": null,
"url": "https://huggingface.co/datasets/0xSero/local-ai-registry"
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "The Hub head has moved past the measured revision (04c4e9e9 at mining); the byte total recorded here is the one at 3f1971b7, the revision the measurements name.",
"severity": "info"
}
] | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "zai-org/GLM-5.3-Flash",
"revision": "3f1971b7b5f7a528c9c4ef6212c8785298a8c24a",
"revision_source": "revision_txt",
"status": "known",
"url": "https://huggingface.co/zai-org/GLM-5.3-Flash"
} | artifact--zai-org.glm-5.3-flash-fp8 | quant | model--zai-org.glm-5.3-flash | GLM-5.3-Flash official FP8 | {
"handle": "zai-org",
"is_registry_maintainer": false,
"name": "Z.ai",
"role": "quantizer",
"url": "https://huggingface.co/zai-org"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": null,
"format": "bf16",
"layer_range": "all",
"tensor_class": "embed_tokens",
"treatment": "native"
},
{
"bits_per_weight": 8,
"format": "fp8_e4m3",
"layer_range": "all",
"tenso... | attn.o=quantized:fp8_e4m3@8|attn.qkv=quantized:fp8_e4m3@8|embed_tokens=native:bf16|lm_head=native:bf16|mlp.down=quantized:fp8_e4m3@8|mlp.gate=quantized:fp8_e4m3@8|mlp.up=quantized:fp8_e4m3@8|moe.experts=quantized:fp8_e4m3@8|norm=native:bf16|head=native|kv=bf16 | {
"sealed": false
} | [
{
"kind": "hf_file",
"note": "candidate_identity.model_revision 3f1971b7..., index_sha256 3c3f4036...",
"uri": "https://huggingface.co/datasets/malaiwah/GLM-5.3-Flash-fidelity-suite-v1/resolve/main/reports/report-fp8-vs-bf16.json"
},
{
"kind": "url",
"note": "byte total 328,366,171,529 over ... | {
"container": "safetensors",
"index_sha256": "3c3f40366a53c3fd7974b4eab7881a365a98c2a4329150befebab99fe7c18b05",
"precision_label": "FP8",
"shard_count": 72,
"size_basis": "repo_all_files",
"size_bytes": 328366171529,
"size_gb": 328.366171529
} |
{
"status": "public",
"uri": "https://huggingface.co/zai-org/GLM-5.3"
} | {
"bits_per_weight_effective": 8,
"bits_per_weight_nominal": 8,
"calibration": {
"corpus": null,
"overlapping_panel_refs": [],
"overlaps_any_panel": null,
"tokens": null,
"used": null
},
"family": "fp8_e4m3",
"group_size": null,
"quantizer": {
"pipeline_ref": null,
"revision": ... | {
"local_ai_registry": {
"match_confidence": "unverified",
"model_id": null,
"model_instance_id": null,
"url": null
}
} | null | [
{
"affects_comparability": false,
"code": "record_note",
"detail": "Scope read from the weight index, not the README: every 2-D projection with a weight_scale_inv sibling is fp8_e4m3 (attention, dense MLP, all 57,600 routed expert matrices, the shared experts), while embed_tokens, lm_head, norms and the... | {
"link_type": "repository",
"path": null,
"reason": null,
"repository": "zai-org/GLM-5.3",
"revision": "187fb9fff6319062325ff825627ef6db084d9bc6",
"revision_source": "hf_api",
"status": "known",
"url": "https://huggingface.co/zai-org/GLM-5.3"
} | artifact--zai-org.glm-5.3-fp8 | quant | model--zai-org.glm-5.3 | GLM-5.3 FP8 (the official block-scaled release) | {
"handle": "zai-org",
"is_registry_maintainer": false,
"name": "Z.ai",
"role": "quantizer",
"url": "https://huggingface.co/zai-org"
} | quant-fidelity-registry/v1 | {
"activation_quantization": null,
"assignments": [
{
"bits_per_weight": 8,
"format": "fp8_e4m3",
"layer_range": "all",
"note": "78 tensors: model.layers.N.self_attn.o_proj.weight. read from zai-org/GLM-5.3@187fb9fff6319062325ff825627ef6db084d9bc6 model.safetensors.index.json: a tensor i... | attn.o=quantized:fp8_e4m3@8|attn.other=quantized:mixed|attn.qkv=quantized:fp8_e4m3@8|embed_tokens=native:bf16@16|lm_head=native:bf16@16|mlp.down=quantized:fp8_e4m3@8|mlp.gate=quantized:fp8_e4m3@8|mlp.up=quantized:fp8_e4m3@8|moe.experts=quantized:fp8_e4m3@8|moe.router=native:bf16@16|moe.shared_expert=quantized:fp8_e4m3@... | {
"sealed": false
} | [
{
"kind": "model_card",
"note": "revision 187fb9fff631; 141 shards; config.json sha256 3ac72612... declares quantization_config quant_method fp8, fmt e4m3, weight_block_size [128, 128], activation_scheme dynamic; index sha256 e0fe7f28...",
"uri": "https://huggingface.co/zai-org/GLM-5.3"
},
{
"ki... | {
"config_sha256": "3ac72612095574542f7fff847ada8e59d9199dd8af44bdf625d7e02615572e69",
"container": "safetensors",
"index_sha256": "e0fe7f28c1f853d4824e4d796374e3dacf1fe470988773952c79b063768134bf",
"precision_label": "FP8",
"shard_count": 141,
"size_basis": "repo_weight_files",
"size_bytes": 755632050320... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.