| { | |
| "architectures": [ | |
| "Mistral3ForConditionalGeneration" | |
| ], | |
| "bos_token_id": 1, | |
| "eos_token_id": 2, | |
| "image_token_index": 10, | |
| "model_type": "mistral3", | |
| "multimodal_projector_bias": false, | |
| "pad_token_id": 11, | |
| "projector_hidden_act": "gelu", | |
| "quantization_config": { | |
| "_load_in_4bit": true, | |
| "_load_in_8bit": false, | |
| "bnb_4bit_compute_dtype": "bfloat16", | |
| "bnb_4bit_quant_storage": "uint8", | |
| "bnb_4bit_quant_type": "nf4", | |
| "bnb_4bit_use_double_quant": true, | |
| "llm_int8_enable_fp32_cpu_offload": false, | |
| "llm_int8_has_fp16_weight": false, | |
| "llm_int8_skip_modules": [ | |
| "lm_head", | |
| "multi_modal_projector", | |
| "merger", | |
| "modality_projection", | |
| "model.language_model.layers.10.mlp", | |
| "model.language_model.layers.8.self_attn", | |
| "model.language_model.layers.4.mlp", | |
| "model.language_model.layers.6.mlp", | |
| "model.language_model.layers.8.mlp", | |
| "model.language_model.layers.9.mlp", | |
| "model.language_model.layers.5.mlp", | |
| "model.language_model.layers.3.self_attn", | |
| "model.language_model.layers.4.self_attn", | |
| "model.language_model.layers.6.self_attn", | |
| "model.language_model.layers.7.mlp", | |
| "model.language_model.layers.1.self_attn", | |
| "model.vision_tower.transformer.layers.20.attention", | |
| "model.language_model.layers.2.self_attn", | |
| "model.vision_tower.transformer.layers.19.attention", | |
| "model.language_model.layers.7.self_attn", | |
| "model.language_model.layers.5.self_attn", | |
| "model.vision_tower.transformer.layers.18.attention", | |
| "model.vision_tower.transformer.layers.17.attention", | |
| "model.vision_tower.transformer.layers.20.feed_forward", | |
| "model.vision_tower.transformer.layers.21.attention", | |
| "model.vision_tower.transformer.layers.16.attention", | |
| "model.vision_tower.transformer.layers.21.feed_forward", | |
| "model.vision_tower.transformer.layers.15.attention", | |
| "model.vision_tower.transformer.layers.22.feed_forward", | |
| "model.multi_modal_projector", | |
| "model.vision_tower.transformer.layers.18.feed_forward", | |
| "model.vision_tower.transformer.layers.16.feed_forward", | |
| "model.vision_tower.transformer.layers.15.feed_forward", | |
| "model.language_model.layers.2.mlp", | |
| "model.vision_tower.transformer.layers.19.feed_forward", | |
| "model.vision_tower.transformer.layers.14.attention", | |
| "model.vision_tower.transformer.layers.8.feed_forward", | |
| "model.vision_tower.transformer.layers.12.feed_forward", | |
| "model.vision_tower.transformer.layers.14.feed_forward", | |
| "model.vision_tower.transformer.layers.7.feed_forward", | |
| "model.vision_tower.transformer.layers.5.feed_forward", | |
| "model.vision_tower.transformer.layers.4.feed_forward", | |
| "model.vision_tower.transformer.layers.6.feed_forward", | |
| "model.vision_tower.transformer.layers.23.attention", | |
| "model.vision_tower.transformer.layers.17.feed_forward", | |
| "model.vision_tower.transformer.layers.13.attention", | |
| "model.vision_tower.transformer.layers.11.feed_forward", | |
| "model.vision_tower.transformer.layers.22.attention", | |
| "model.vision_tower.transformer.layers.3.feed_forward", | |
| "model.vision_tower.transformer.layers.10.feed_forward", | |
| "model.vision_tower.transformer.layers.9.feed_forward", | |
| "model.vision_tower.transformer.layers.13.feed_forward", | |
| "model.vision_tower.transformer.layers.12.attention", | |
| "model.vision_tower.transformer.layers.23.feed_forward", | |
| "model.vision_tower.transformer.layers.11.attention", | |
| "model.vision_tower.transformer.layers.10.attention", | |
| "model.vision_tower.transformer.layers.8.attention", | |
| "model.vision_tower.transformer.layers.2.feed_forward", | |
| "model.vision_tower.transformer.layers.5.attention", | |
| "model.vision_tower.transformer.layers.0.feed_forward", | |
| "model.vision_tower.transformer.layers.1.feed_forward", | |
| "model.vision_tower.transformer.layers.4.attention", | |
| "model.vision_tower.transformer.layers.7.attention", | |
| "model.vision_tower.transformer.layers.6.attention", | |
| "model.vision_tower.transformer.layers.0.attention", | |
| "model.vision_tower.transformer.layers.3.attention", | |
| "model.vision_tower.transformer.layers.1.attention", | |
| "model.vision_tower.transformer.layers.9.attention", | |
| "model.vision_tower.transformer.layers.2.attention" | |
| ], | |
| "llm_int8_threshold": 6.0, | |
| "load_in_4bit": true, | |
| "load_in_8bit": false, | |
| "quant_method": "bitsandbytes" | |
| }, | |
| "spatial_merge_size": 2, | |
| "text_config": { | |
| "attention_dropout": 0.0, | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 5120, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 32768, | |
| "max_position_embeddings": 131072, | |
| "model_type": "mistral", | |
| "num_attention_heads": 32, | |
| "num_hidden_layers": 40, | |
| "num_key_value_heads": 8, | |
| "rms_norm_eps": 1e-05, | |
| "rope_theta": 1000000000.0, | |
| "sliding_window": null, | |
| "torch_dtype": "bfloat16", | |
| "use_cache": true, | |
| "vocab_size": 131072 | |
| }, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.52.4", | |
| "unsloth_fixed": true, | |
| "vision_config": { | |
| "attention_dropout": 0.0, | |
| "head_dim": 64, | |
| "hidden_act": "silu", | |
| "hidden_size": 1024, | |
| "image_size": 1540, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 4096, | |
| "model_type": "pixtral", | |
| "num_attention_heads": 16, | |
| "num_channels": 3, | |
| "num_hidden_layers": 24, | |
| "patch_size": 14, | |
| "rope_theta": 10000.0, | |
| "torch_dtype": "bfloat16" | |
| }, | |
| "vision_feature_layer": -1 | |
| } | |