{ "architectures": [ "Mistral3Model" ], "dtype": "bfloat16", "image_token_index": 10, "model_type": "mistral3", "multimodal_projector_bias": false, "projector_hidden_act": "gelu", "quantization_config": { "add_skip_keys": true, "dequantize_fp32": false, "dynamic_loss_threshold": 0.0001, "group_size": 0, "is_integer": true, "is_training": false, "modules_dtype_dict": { "float10_e4m5fn": [ "vision_tower.transformer.layers.0.attention.k_proj.weight", "vision_tower.transformer.layers.0.attention.q_proj.weight" ], "float9_e3m5fn": [ "vision_tower.transformer.layers.0.attention.o_proj.weight", "vision_tower.transformer.layers.0.attention.v_proj.weight", "vision_tower.transformer.layers.0.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.0.feed_forward.up_proj.weight", "vision_tower.transformer.layers.1.attention.k_proj.weight", "vision_tower.transformer.layers.1.attention.o_proj.weight", "vision_tower.transformer.layers.1.attention.q_proj.weight", "vision_tower.transformer.layers.1.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.2.attention.o_proj.weight", "vision_tower.transformer.layers.23.feed_forward.down_proj.weight" ], "int8": [ "language_model.layers.0.mlp.down_proj.weight", "language_model.layers.0.mlp.gate_proj.weight", "language_model.layers.0.mlp.up_proj.weight", "language_model.layers.0.self_attn.k_proj.weight", "language_model.layers.0.self_attn.q_proj.weight", "language_model.layers.0.self_attn.v_proj.weight", "language_model.layers.1.mlp.down_proj.weight", "language_model.layers.1.mlp.gate_proj.weight", "language_model.layers.1.mlp.up_proj.weight", "language_model.layers.1.self_attn.k_proj.weight", "language_model.layers.1.self_attn.o_proj.weight", "language_model.layers.1.self_attn.q_proj.weight", "language_model.layers.1.self_attn.v_proj.weight", "language_model.layers.2.mlp.down_proj.weight", "language_model.layers.2.mlp.gate_proj.weight", "language_model.layers.2.mlp.up_proj.weight", "language_model.layers.2.self_attn.k_proj.weight", "language_model.layers.2.self_attn.o_proj.weight", "language_model.layers.2.self_attn.q_proj.weight", "language_model.layers.2.self_attn.v_proj.weight", "language_model.layers.3.mlp.down_proj.weight", "language_model.layers.3.mlp.gate_proj.weight", "language_model.layers.3.mlp.up_proj.weight", "language_model.layers.3.self_attn.k_proj.weight", "language_model.layers.3.self_attn.o_proj.weight", "language_model.layers.3.self_attn.q_proj.weight", "language_model.layers.3.self_attn.v_proj.weight", "language_model.layers.4.mlp.down_proj.weight", "language_model.layers.4.mlp.gate_proj.weight", "language_model.layers.4.mlp.up_proj.weight", "language_model.layers.4.self_attn.k_proj.weight", "language_model.layers.4.self_attn.o_proj.weight", "language_model.layers.4.self_attn.q_proj.weight", "language_model.layers.4.self_attn.v_proj.weight", "language_model.layers.5.mlp.down_proj.weight", "language_model.layers.5.mlp.gate_proj.weight", "language_model.layers.5.mlp.up_proj.weight", "language_model.layers.5.self_attn.k_proj.weight", "language_model.layers.5.self_attn.o_proj.weight", "language_model.layers.5.self_attn.q_proj.weight", "language_model.layers.5.self_attn.v_proj.weight", "language_model.layers.6.mlp.down_proj.weight", "language_model.layers.6.mlp.gate_proj.weight", "language_model.layers.6.mlp.up_proj.weight", "language_model.layers.6.self_attn.k_proj.weight", "language_model.layers.6.self_attn.o_proj.weight", "language_model.layers.6.self_attn.q_proj.weight", "language_model.layers.6.self_attn.v_proj.weight", "language_model.layers.7.mlp.down_proj.weight", "language_model.layers.7.mlp.gate_proj.weight", "language_model.layers.7.mlp.up_proj.weight", "language_model.layers.7.self_attn.k_proj.weight", "language_model.layers.7.self_attn.o_proj.weight", "language_model.layers.7.self_attn.q_proj.weight", "language_model.layers.7.self_attn.v_proj.weight", "language_model.layers.8.mlp.down_proj.weight", "language_model.layers.8.mlp.gate_proj.weight", "language_model.layers.8.mlp.up_proj.weight", "language_model.layers.8.self_attn.k_proj.weight", "language_model.layers.8.self_attn.o_proj.weight", "language_model.layers.8.self_attn.q_proj.weight", "language_model.layers.8.self_attn.v_proj.weight", "language_model.layers.9.mlp.down_proj.weight", "language_model.layers.9.mlp.gate_proj.weight", "language_model.layers.9.mlp.up_proj.weight", "language_model.layers.9.self_attn.k_proj.weight", "language_model.layers.9.self_attn.o_proj.weight", "language_model.layers.9.self_attn.q_proj.weight", "language_model.layers.9.self_attn.v_proj.weight", "language_model.layers.10.mlp.down_proj.weight", "language_model.layers.10.mlp.gate_proj.weight", "language_model.layers.10.mlp.up_proj.weight", "language_model.layers.10.self_attn.k_proj.weight", "language_model.layers.10.self_attn.o_proj.weight", "language_model.layers.10.self_attn.q_proj.weight", "language_model.layers.10.self_attn.v_proj.weight", "language_model.layers.11.mlp.down_proj.weight", "language_model.layers.11.mlp.gate_proj.weight", "language_model.layers.11.mlp.up_proj.weight", "language_model.layers.11.self_attn.k_proj.weight", "language_model.layers.11.self_attn.o_proj.weight", "language_model.layers.11.self_attn.q_proj.weight", "language_model.layers.11.self_attn.v_proj.weight", "language_model.layers.12.mlp.down_proj.weight", "language_model.layers.12.mlp.gate_proj.weight", "language_model.layers.12.mlp.up_proj.weight", "language_model.layers.12.self_attn.k_proj.weight", "language_model.layers.12.self_attn.o_proj.weight", "language_model.layers.12.self_attn.q_proj.weight", "language_model.layers.12.self_attn.v_proj.weight", "language_model.layers.13.mlp.down_proj.weight", "language_model.layers.13.mlp.gate_proj.weight", "language_model.layers.13.mlp.up_proj.weight", "language_model.layers.13.self_attn.k_proj.weight", "language_model.layers.13.self_attn.o_proj.weight", "language_model.layers.13.self_attn.q_proj.weight", "language_model.layers.13.self_attn.v_proj.weight", "language_model.layers.14.mlp.down_proj.weight", "language_model.layers.14.mlp.gate_proj.weight", "language_model.layers.14.mlp.up_proj.weight", "language_model.layers.14.self_attn.k_proj.weight", "language_model.layers.14.self_attn.o_proj.weight", "language_model.layers.14.self_attn.q_proj.weight", "language_model.layers.14.self_attn.v_proj.weight", "language_model.layers.15.mlp.down_proj.weight", "language_model.layers.15.mlp.gate_proj.weight", "language_model.layers.15.mlp.up_proj.weight", "language_model.layers.15.self_attn.k_proj.weight", "language_model.layers.15.self_attn.o_proj.weight", "language_model.layers.15.self_attn.q_proj.weight", "language_model.layers.15.self_attn.v_proj.weight", "language_model.layers.16.mlp.down_proj.weight", "language_model.layers.16.mlp.gate_proj.weight", "language_model.layers.16.mlp.up_proj.weight", "language_model.layers.16.self_attn.k_proj.weight", "language_model.layers.16.self_attn.o_proj.weight", "language_model.layers.16.self_attn.q_proj.weight", "language_model.layers.16.self_attn.v_proj.weight", "language_model.layers.17.mlp.down_proj.weight", "language_model.layers.17.mlp.gate_proj.weight", "language_model.layers.17.mlp.up_proj.weight", "language_model.layers.17.self_attn.k_proj.weight", "language_model.layers.17.self_attn.o_proj.weight", "language_model.layers.17.self_attn.q_proj.weight", "language_model.layers.17.self_attn.v_proj.weight", "language_model.layers.18.mlp.down_proj.weight", "language_model.layers.18.mlp.gate_proj.weight", "language_model.layers.18.mlp.up_proj.weight", "language_model.layers.18.self_attn.k_proj.weight", "language_model.layers.18.self_attn.o_proj.weight", "language_model.layers.18.self_attn.q_proj.weight", "language_model.layers.18.self_attn.v_proj.weight", "language_model.layers.19.mlp.down_proj.weight", "language_model.layers.19.mlp.gate_proj.weight", "language_model.layers.19.mlp.up_proj.weight", "language_model.layers.19.self_attn.k_proj.weight", "language_model.layers.19.self_attn.o_proj.weight", "language_model.layers.19.self_attn.q_proj.weight", "language_model.layers.19.self_attn.v_proj.weight", "language_model.layers.20.mlp.down_proj.weight", "language_model.layers.20.mlp.gate_proj.weight", "language_model.layers.20.mlp.up_proj.weight", "language_model.layers.20.self_attn.k_proj.weight", "language_model.layers.20.self_attn.o_proj.weight", "language_model.layers.20.self_attn.q_proj.weight", "language_model.layers.20.self_attn.v_proj.weight", "language_model.layers.21.mlp.down_proj.weight", "language_model.layers.21.mlp.gate_proj.weight", "language_model.layers.21.mlp.up_proj.weight", "language_model.layers.21.self_attn.k_proj.weight", "language_model.layers.21.self_attn.o_proj.weight", "language_model.layers.21.self_attn.q_proj.weight", "language_model.layers.21.self_attn.v_proj.weight", "language_model.layers.22.mlp.down_proj.weight", "language_model.layers.22.mlp.gate_proj.weight", "language_model.layers.22.mlp.up_proj.weight", "language_model.layers.22.self_attn.k_proj.weight", "language_model.layers.22.self_attn.o_proj.weight", "language_model.layers.22.self_attn.q_proj.weight", "language_model.layers.22.self_attn.v_proj.weight", "language_model.layers.23.mlp.down_proj.weight", "language_model.layers.23.mlp.gate_proj.weight", "language_model.layers.23.mlp.up_proj.weight", "language_model.layers.23.self_attn.k_proj.weight", "language_model.layers.23.self_attn.o_proj.weight", "language_model.layers.23.self_attn.q_proj.weight", "language_model.layers.23.self_attn.v_proj.weight", "language_model.layers.24.mlp.down_proj.weight", "language_model.layers.24.mlp.gate_proj.weight", "language_model.layers.24.mlp.up_proj.weight", "language_model.layers.24.self_attn.k_proj.weight", "language_model.layers.24.self_attn.o_proj.weight", "language_model.layers.24.self_attn.q_proj.weight", "language_model.layers.24.self_attn.v_proj.weight", "language_model.layers.25.mlp.down_proj.weight", "language_model.layers.25.mlp.gate_proj.weight", "language_model.layers.25.mlp.up_proj.weight", "language_model.layers.25.self_attn.k_proj.weight", "language_model.layers.25.self_attn.o_proj.weight", "language_model.layers.25.self_attn.q_proj.weight", "language_model.layers.25.self_attn.v_proj.weight", "vision_tower.transformer.layers.0.feed_forward.down_proj.weight", "vision_tower.transformer.layers.1.feed_forward.up_proj.weight", "vision_tower.transformer.layers.2.attention.k_proj.weight", "vision_tower.transformer.layers.2.attention.q_proj.weight", "vision_tower.transformer.layers.2.attention.v_proj.weight", "vision_tower.transformer.layers.2.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.2.feed_forward.up_proj.weight", "vision_tower.transformer.layers.3.attention.q_proj.weight", "vision_tower.transformer.layers.3.attention.v_proj.weight", "vision_tower.transformer.layers.3.feed_forward.down_proj.weight", "vision_tower.transformer.layers.3.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.3.feed_forward.up_proj.weight", "vision_tower.transformer.layers.4.attention.q_proj.weight", "vision_tower.transformer.layers.4.attention.v_proj.weight", "vision_tower.transformer.layers.4.feed_forward.down_proj.weight", "vision_tower.transformer.layers.4.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.4.feed_forward.up_proj.weight", "vision_tower.transformer.layers.5.attention.o_proj.weight", "vision_tower.transformer.layers.5.attention.q_proj.weight", "vision_tower.transformer.layers.5.attention.v_proj.weight", "vision_tower.transformer.layers.5.feed_forward.down_proj.weight", "vision_tower.transformer.layers.5.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.5.feed_forward.up_proj.weight", "vision_tower.transformer.layers.6.attention.k_proj.weight", "vision_tower.transformer.layers.6.attention.o_proj.weight", "vision_tower.transformer.layers.6.attention.q_proj.weight", "vision_tower.transformer.layers.6.attention.v_proj.weight", "vision_tower.transformer.layers.6.feed_forward.down_proj.weight", "vision_tower.transformer.layers.6.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.6.feed_forward.up_proj.weight", "vision_tower.transformer.layers.7.attention.k_proj.weight", "vision_tower.transformer.layers.7.attention.o_proj.weight", "vision_tower.transformer.layers.7.attention.q_proj.weight", "vision_tower.transformer.layers.7.attention.v_proj.weight", "vision_tower.transformer.layers.7.feed_forward.down_proj.weight", "vision_tower.transformer.layers.7.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.7.feed_forward.up_proj.weight", "vision_tower.transformer.layers.8.attention.k_proj.weight", "vision_tower.transformer.layers.8.attention.o_proj.weight", "vision_tower.transformer.layers.8.attention.q_proj.weight", "vision_tower.transformer.layers.8.attention.v_proj.weight", "vision_tower.transformer.layers.8.feed_forward.down_proj.weight", "vision_tower.transformer.layers.8.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.8.feed_forward.up_proj.weight", "vision_tower.transformer.layers.9.attention.k_proj.weight", "vision_tower.transformer.layers.9.attention.o_proj.weight", "vision_tower.transformer.layers.9.attention.q_proj.weight", "vision_tower.transformer.layers.9.attention.v_proj.weight", "vision_tower.transformer.layers.9.feed_forward.down_proj.weight", "vision_tower.transformer.layers.9.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.9.feed_forward.up_proj.weight", "vision_tower.transformer.layers.10.attention.k_proj.weight", "vision_tower.transformer.layers.10.attention.o_proj.weight", "vision_tower.transformer.layers.10.attention.q_proj.weight", "vision_tower.transformer.layers.10.attention.v_proj.weight", "vision_tower.transformer.layers.10.feed_forward.down_proj.weight", "vision_tower.transformer.layers.10.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.10.feed_forward.up_proj.weight", "vision_tower.transformer.layers.11.attention.k_proj.weight", "vision_tower.transformer.layers.11.attention.o_proj.weight", "vision_tower.transformer.layers.11.attention.q_proj.weight", "vision_tower.transformer.layers.11.attention.v_proj.weight", "vision_tower.transformer.layers.11.feed_forward.down_proj.weight", "vision_tower.transformer.layers.11.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.11.feed_forward.up_proj.weight", "vision_tower.transformer.layers.12.attention.k_proj.weight", "vision_tower.transformer.layers.12.attention.o_proj.weight", "vision_tower.transformer.layers.12.attention.q_proj.weight", "vision_tower.transformer.layers.12.attention.v_proj.weight", "vision_tower.transformer.layers.12.feed_forward.down_proj.weight", "vision_tower.transformer.layers.12.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.12.feed_forward.up_proj.weight", "vision_tower.transformer.layers.13.attention.k_proj.weight", "vision_tower.transformer.layers.13.attention.o_proj.weight", "vision_tower.transformer.layers.13.attention.q_proj.weight", "vision_tower.transformer.layers.13.attention.v_proj.weight", "vision_tower.transformer.layers.13.feed_forward.down_proj.weight", "vision_tower.transformer.layers.13.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.13.feed_forward.up_proj.weight", "vision_tower.transformer.layers.14.attention.k_proj.weight", "vision_tower.transformer.layers.14.attention.o_proj.weight", "vision_tower.transformer.layers.14.attention.q_proj.weight", "vision_tower.transformer.layers.14.attention.v_proj.weight", "vision_tower.transformer.layers.14.feed_forward.down_proj.weight", "vision_tower.transformer.layers.14.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.14.feed_forward.up_proj.weight", "vision_tower.transformer.layers.15.attention.k_proj.weight", "vision_tower.transformer.layers.15.attention.o_proj.weight", "vision_tower.transformer.layers.15.attention.q_proj.weight", "vision_tower.transformer.layers.15.attention.v_proj.weight", "vision_tower.transformer.layers.15.feed_forward.down_proj.weight", "vision_tower.transformer.layers.15.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.15.feed_forward.up_proj.weight", "vision_tower.transformer.layers.16.attention.k_proj.weight", "vision_tower.transformer.layers.16.attention.o_proj.weight", "vision_tower.transformer.layers.16.attention.q_proj.weight", "vision_tower.transformer.layers.16.attention.v_proj.weight", "vision_tower.transformer.layers.16.feed_forward.down_proj.weight", "vision_tower.transformer.layers.16.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.16.feed_forward.up_proj.weight", "vision_tower.transformer.layers.17.attention.k_proj.weight", "vision_tower.transformer.layers.17.attention.o_proj.weight", "vision_tower.transformer.layers.17.attention.q_proj.weight", "vision_tower.transformer.layers.17.attention.v_proj.weight", "vision_tower.transformer.layers.17.feed_forward.down_proj.weight", "vision_tower.transformer.layers.17.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.17.feed_forward.up_proj.weight", "vision_tower.transformer.layers.18.attention.k_proj.weight", "vision_tower.transformer.layers.18.attention.o_proj.weight", "vision_tower.transformer.layers.18.attention.q_proj.weight", "vision_tower.transformer.layers.18.attention.v_proj.weight", "vision_tower.transformer.layers.18.feed_forward.down_proj.weight", "vision_tower.transformer.layers.18.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.18.feed_forward.up_proj.weight", "vision_tower.transformer.layers.19.attention.k_proj.weight", "vision_tower.transformer.layers.19.attention.o_proj.weight", "vision_tower.transformer.layers.19.attention.q_proj.weight", "vision_tower.transformer.layers.19.attention.v_proj.weight", "vision_tower.transformer.layers.19.feed_forward.down_proj.weight", "vision_tower.transformer.layers.19.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.19.feed_forward.up_proj.weight", "vision_tower.transformer.layers.20.attention.k_proj.weight", "vision_tower.transformer.layers.20.attention.o_proj.weight", "vision_tower.transformer.layers.20.attention.q_proj.weight", "vision_tower.transformer.layers.20.attention.v_proj.weight", "vision_tower.transformer.layers.20.feed_forward.down_proj.weight", "vision_tower.transformer.layers.20.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.20.feed_forward.up_proj.weight", "vision_tower.transformer.layers.21.attention.k_proj.weight", "vision_tower.transformer.layers.21.attention.o_proj.weight", "vision_tower.transformer.layers.21.attention.q_proj.weight", "vision_tower.transformer.layers.21.attention.v_proj.weight", "vision_tower.transformer.layers.21.feed_forward.down_proj.weight", "vision_tower.transformer.layers.21.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.21.feed_forward.up_proj.weight", "vision_tower.transformer.layers.22.attention.k_proj.weight", "vision_tower.transformer.layers.22.attention.o_proj.weight", "vision_tower.transformer.layers.22.attention.q_proj.weight", "vision_tower.transformer.layers.22.attention.v_proj.weight", "vision_tower.transformer.layers.22.feed_forward.down_proj.weight", "vision_tower.transformer.layers.22.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.22.feed_forward.up_proj.weight", "vision_tower.transformer.layers.23.attention.k_proj.weight", "vision_tower.transformer.layers.23.attention.o_proj.weight", "vision_tower.transformer.layers.23.attention.q_proj.weight", "vision_tower.transformer.layers.23.attention.v_proj.weight", "vision_tower.transformer.layers.23.feed_forward.gate_proj.weight", "vision_tower.transformer.layers.23.feed_forward.up_proj.weight" ], "uint8": [ "language_model.layers.0.self_attn.o_proj.weight", "vision_tower.transformer.layers.1.attention.v_proj.weight", "vision_tower.transformer.layers.1.feed_forward.down_proj.weight", "vision_tower.transformer.layers.2.feed_forward.down_proj.weight", "vision_tower.transformer.layers.3.attention.k_proj.weight", "vision_tower.transformer.layers.3.attention.o_proj.weight", "vision_tower.transformer.layers.4.attention.k_proj.weight", "vision_tower.transformer.layers.4.attention.o_proj.weight", "vision_tower.transformer.layers.5.attention.k_proj.weight" ] }, "modules_quant_config": {}, "modules_to_not_convert": [ "lm_head", "time_text_embed", ".proj_out", ".emb_out", ".norm_out", ".vid_in", ".txt_in", ".x_embedder", ".time_embed", ".img_out", ".txt_out", "patch_embed", "patch_embedding", ".emb_in", ".t_embedder", ".condition_embedder", ".context_embedder", ".vid_out", "multi_modal_projector", ".y_embedder", "patch_emb", "wte", "language_model.norm", ".img_in", ".final_layer" ], "non_blocking": false, "quant_conv": false, "quant_method": "sdnq", "quantization_device": null, "quantized_matmul_dtype": null, "return_device": null, "sdnq_version": "0.1.6", "svd_rank": 32, "svd_steps": 8, "use_dynamic_quantization": true, "use_grad_ckpt": true, "use_quantized_matmul": false, "use_quantized_matmul_conv": false, "use_static_quantization": true, "use_stochastic_rounding": false, "use_svd": false, "weights_dtype": "int8" }, "spatial_merge_size": 2, "text_config": { "attention_dropout": 0.0, "bos_token_id": 1, "dtype": "bfloat16", "eos_token_id": 2, "head_dim": 128, "hidden_act": "silu", "hidden_size": 3072, "initializer_range": 0.02, "intermediate_size": 9216, "max_position_embeddings": 262144, "model_type": "ministral3", "num_attention_heads": 32, "num_hidden_layers": 26, "num_key_value_heads": 8, "pad_token_id": 11, "rms_norm_eps": 1e-05, "rope_parameters": { "beta_fast": 32.0, "beta_slow": 1.0, "factor": 16.0, "llama_4_scaling_beta": 0.1, "mscale": 1.0, "mscale_all_dim": 1.0, "original_max_position_embeddings": 16384, "rope_theta": 1000000.0, "rope_type": "yarn", "type": "yarn" }, "sliding_window": null, "tie_word_embeddings": true, "use_cache": true, "vocab_size": 131072 }, "tie_word_embeddings": true, "transformers_version": "5.3.0", "vision_config": { "attention_dropout": 0.0, "dtype": "bfloat16", "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_parameters": { "rope_theta": 10000.0, "rope_type": "default" } }, "vision_feature_layer": -1 }