Text Generation
Safetensors
Model Optimizer
gemma4
nvidia
ModelOpt
Gemma-4-31B-IT
lighthouse
quantized
NVFP4
conversational
modelopt
Instructions to use nvidia/Gemma-4-31B-IT-NVFP4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Inference
Download config.json from nvidia/Gemma-4-31B-IT-NVFP4: direct link, hf CLI and curl.
- Browser
- Download file 9.46 kB
-
https://huggingface.co/nvidia/Gemma-4-31B-IT-NVFP4/resolve/main/config.json
- Command line
-
hf download hf://nvidia/Gemma-4-31B-IT-NVFP4/config.json
-
curl -L -o config.json https://huggingface.co/nvidia/Gemma-4-31B-IT-NVFP4/resolve/main/config.json
9.46 kB
| { | |
| "architectures": [ | |
| "Gemma4ForConditionalGeneration" | |
| ], | |
| "audio_config": null, | |
| "audio_token_id": 258881, | |
| "boa_token_id": 256000, | |
| "boi_token_id": 255999, | |
| "dtype": "bfloat16", | |
| "eoa_token_id": 258883, | |
| "eoa_token_index": 258883, | |
| "eoi_token_id": 258882, | |
| "eos_token_id": [ | |
| 1, | |
| 106 | |
| ], | |
| "image_token_id": 258880, | |
| "initializer_range": 0.02, | |
| "model_type": "gemma4", | |
| "text_config": { | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "attention_k_eq_v": true, | |
| "bos_token_id": 2, | |
| "dtype": "bfloat16", | |
| "enable_moe_block": false, | |
| "eos_token_id": 1, | |
| "expert_intermediate_size": null, | |
| "final_logit_softcapping": 30.0, | |
| "global_head_dim": 512, | |
| "head_dim": 256, | |
| "hidden_activation": "gelu_pytorch_tanh", | |
| "hidden_size": 5376, | |
| "hidden_size_per_layer_input": 0, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 21504, | |
| "layer_types": [ | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "sliding_attention", | |
| "full_attention" | |
| ], | |
| "max_position_embeddings": 262144, | |
| "model_type": "gemma4_text", | |
| "num_attention_heads": 32, | |
| "num_experts": null, | |
| "num_global_key_value_heads": 4, | |
| "num_hidden_layers": 60, | |
| "num_key_value_heads": 16, | |
| "num_kv_shared_layers": 0, | |
| "pad_token_id": 0, | |
| "rms_norm_eps": 1e-06, | |
| "rope_parameters": { | |
| "full_attention": { | |
| "partial_rotary_factor": 0.25, | |
| "rope_theta": 1000000.0, | |
| "rope_type": "proportional" | |
| }, | |
| "sliding_attention": { | |
| "rope_theta": 10000.0, | |
| "rope_type": "default" | |
| } | |
| }, | |
| "sliding_window": 1024, | |
| "tie_word_embeddings": true, | |
| "top_k_experts": null, | |
| "use_bidirectional_attention": "vision", | |
| "use_cache": true, | |
| "use_double_wide_mlp": false, | |
| "vocab_size": 262144, | |
| "vocab_size_per_layer_input": 262144 | |
| }, | |
| "tie_word_embeddings": true, | |
| "transformers_version": "5.5.0.dev0", | |
| "video_token_id": 258884, | |
| "vision_config": { | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "default_output_length": 280, | |
| "dtype": "bfloat16", | |
| "global_head_dim": 72, | |
| "head_dim": 72, | |
| "hidden_activation": "gelu_pytorch_tanh", | |
| "hidden_size": 1152, | |
| "image_seq_length": 280, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 4304, | |
| "max_position_embeddings": 131072, | |
| "model_type": "gemma4_vision", | |
| "num_attention_heads": 16, | |
| "num_hidden_layers": 27, | |
| "num_key_value_heads": 16, | |
| "patch_size": 16, | |
| "pooling_kernel_size": 3, | |
| "position_embedding_size": 10240, | |
| "rms_norm_eps": 1e-06, | |
| "rope_parameters": { | |
| "rope_theta": 100.0, | |
| "rope_type": "default" | |
| }, | |
| "standardize": true, | |
| "use_clipped_linears": false, | |
| "video_seq_length": 70 | |
| }, | |
| "vision_soft_tokens_per_image": 280, | |
| "quantization_config": { | |
| "config_groups": { | |
| "group_0": { | |
| "input_activations": { | |
| "dynamic": false, | |
| "num_bits": 4, | |
| "type": "float", | |
| "group_size": 16 | |
| }, | |
| "weights": { | |
| "dynamic": false, | |
| "num_bits": 4, | |
| "type": "float", | |
| "group_size": 16 | |
| }, | |
| "targets": [ | |
| "Linear" | |
| ] | |
| } | |
| }, | |
| "ignore": [ | |
| "lm_head", | |
| "model.embed_vision*", | |
| "model.language_model.layers.0.self_attn*", | |
| "model.language_model.layers.1.self_attn*", | |
| "model.language_model.layers.10.self_attn*", | |
| "model.language_model.layers.11.self_attn*", | |
| "model.language_model.layers.12.self_attn*", | |
| "model.language_model.layers.13.self_attn*", | |
| "model.language_model.layers.14.self_attn*", | |
| "model.language_model.layers.15.self_attn*", | |
| "model.language_model.layers.16.self_attn*", | |
| "model.language_model.layers.17.self_attn*", | |
| "model.language_model.layers.18.self_attn*", | |
| "model.language_model.layers.19.self_attn*", | |
| "model.language_model.layers.2.self_attn*", | |
| "model.language_model.layers.20.self_attn*", | |
| "model.language_model.layers.21.self_attn*", | |
| "model.language_model.layers.22.self_attn*", | |
| "model.language_model.layers.23.self_attn*", | |
| "model.language_model.layers.24.self_attn*", | |
| "model.language_model.layers.25.self_attn*", | |
| "model.language_model.layers.26.self_attn*", | |
| "model.language_model.layers.27.self_attn*", | |
| "model.language_model.layers.28.self_attn*", | |
| "model.language_model.layers.29.self_attn*", | |
| "model.language_model.layers.3.self_attn*", | |
| "model.language_model.layers.30.self_attn*", | |
| "model.language_model.layers.31.self_attn*", | |
| "model.language_model.layers.32.self_attn*", | |
| "model.language_model.layers.33.self_attn*", | |
| "model.language_model.layers.34.self_attn*", | |
| "model.language_model.layers.35.self_attn*", | |
| "model.language_model.layers.36.self_attn*", | |
| "model.language_model.layers.37.self_attn*", | |
| "model.language_model.layers.38.self_attn*", | |
| "model.language_model.layers.39.self_attn*", | |
| "model.language_model.layers.4.self_attn*", | |
| "model.language_model.layers.40.self_attn*", | |
| "model.language_model.layers.41.self_attn*", | |
| "model.language_model.layers.42.self_attn*", | |
| "model.language_model.layers.43.self_attn*", | |
| "model.language_model.layers.44.self_attn*", | |
| "model.language_model.layers.45.self_attn*", | |
| "model.language_model.layers.46.self_attn*", | |
| "model.language_model.layers.47.self_attn*", | |
| "model.language_model.layers.48.self_attn*", | |
| "model.language_model.layers.49.self_attn*", | |
| "model.language_model.layers.5.self_attn*", | |
| "model.language_model.layers.50.self_attn*", | |
| "model.language_model.layers.51.self_attn*", | |
| "model.language_model.layers.52.self_attn*", | |
| "model.language_model.layers.53.self_attn*", | |
| "model.language_model.layers.54.self_attn*", | |
| "model.language_model.layers.55.self_attn*", | |
| "model.language_model.layers.56.self_attn*", | |
| "model.language_model.layers.57.self_attn*", | |
| "model.language_model.layers.58.self_attn*", | |
| "model.language_model.layers.59.self_attn*", | |
| "model.language_model.layers.6.self_attn*", | |
| "model.language_model.layers.7.self_attn*", | |
| "model.language_model.layers.8.self_attn*", | |
| "model.language_model.layers.9.self_attn*", | |
| "model.vision_tower*" | |
| ], | |
| "quant_algo": "NVFP4", | |
| "kv_cache_scheme": { | |
| "dynamic": false, | |
| "num_bits": 8, | |
| "type": "float" | |
| }, | |
| "producer": { | |
| "name": "modelopt", | |
| "version": "0.37.0" | |
| }, | |
| "quant_method": "modelopt" | |
| } | |
| } |