checks_run list | error_count int64 | errors list | format_version int64 | receipt_sha256 string | schema string | structural_status string | subject string | warning_count int64 | warnings list |
|---|---|---|---|---|---|---|---|---|---|
[
"pre-seal"
] | 0 | [] | 1 | a04a056c29abb005484550879b4d42e9cdb7732dbaeb9be3d32dcbde23c04418 | malaiwah.fidelity-structural-validation.v1 | sealed | dataset:fidelity--fruit.malaiwah.quant.fp8 | 0 | [] |
fidelity--fruit.malaiwah.quant.fp8
A quant fidelity dataset in hidden form, produced by engines/tools/hf_capture.py from malaiwah/GLM-5.2-SIQ-Fruit-fp8.
The cut
the final hidden state handed to lm_head -- after the text model's final norm and immediately before the head matmul -- captured as the head module's input via torch.nn.Module.register_forward_pre_hook; replay applies the head ONLY (no final norm at replay time: the capture already sits after it). Same cut as engines/tools/hidden_replay.py and as Festr's kimi-k3 hidden-replay qualification.
What is here
| records | 16 |
| scored positions | 32752 |
| context length | 2048 |
| hidden width | 1024 |
| vocab size | 154880 |
| capture bytes (hidden form) | 67080368 |
| the same capture in logit form would be | 20290519040 bytes (302x) |
| scope policy | mixed |
| lane | streaming |
Does the model still generate sensibly
The capital of France is -> the (p=0.1148, entropy 5.525 of a possible 11.950 nats). Recorded, not enforced. One extra window through the same schedule; its hidden state was discarded and is not part of this dataset.
How to check it
bin/fidelity-dataset verify <this directory>
Full specification: https://github.com/malaiwah/quant-fidelity-suite/blob/main/docs/FIDELITY-DATASET-SPEC.md
- Downloads last month
- 36