{ "_name_or_path": "/network/shubhra/Pixtral-Large-Instruct-2411-pr23/", "architectures": [ "LlavaForConditionalGeneration" ], "ignore_index": -100, "image_seq_length": 1, "image_token_index": 10, "model_type": "llava", "multimodal_projector_bias": false, "projector_hidden_act": "gelu", "quantization_config": { "config_groups": { "group_0": { "input_activations": { "actorder": null, "block_structure": null, "dynamic": true, "group_size": null, "num_bits": 8, "observer": null, "observer_kwargs": {}, "strategy": "token", "symmetric": true, "type": "int" }, "output_activations": null, "targets": [ "Linear" ], "weights": { "actorder": null, "block_structure": null, "dynamic": false, "group_size": null, "num_bits": 8, "observer": "minmax", "observer_kwargs": {}, "strategy": "channel", "symmetric": true, "type": "int" } } }, "format": "int-quantized", "global_compression_ratio": 1.3136315653954034, "ignore": [ "vision_tower.transformer.layers.0.feed_forward.gate_proj", "vision_tower.transformer.layers.0.feed_forward.up_proj", "vision_tower.transformer.layers.0.feed_forward.down_proj", "vision_tower.transformer.layers.0.attention.k_proj", "vision_tower.transformer.layers.0.attention.v_proj", "vision_tower.transformer.layers.0.attention.q_proj", "vision_tower.transformer.layers.0.attention.o_proj", "vision_tower.transformer.layers.1.feed_forward.gate_proj", "vision_tower.transformer.layers.1.feed_forward.up_proj", "vision_tower.transformer.layers.1.feed_forward.down_proj", "vision_tower.transformer.layers.1.attention.k_proj", "vision_tower.transformer.layers.1.attention.v_proj", "vision_tower.transformer.layers.1.attention.q_proj", "vision_tower.transformer.layers.1.attention.o_proj", "vision_tower.transformer.layers.2.feed_forward.gate_proj", "vision_tower.transformer.layers.2.feed_forward.up_proj", "vision_tower.transformer.layers.2.feed_forward.down_proj", "vision_tower.transformer.layers.2.attention.k_proj", "vision_tower.transformer.layers.2.attention.v_proj", "vision_tower.transformer.layers.2.attention.q_proj", "vision_tower.transformer.layers.2.attention.o_proj", "vision_tower.transformer.layers.3.feed_forward.gate_proj", "vision_tower.transformer.layers.3.feed_forward.up_proj", "vision_tower.transformer.layers.3.feed_forward.down_proj", "vision_tower.transformer.layers.3.attention.k_proj", "vision_tower.transformer.layers.3.attention.v_proj", "vision_tower.transformer.layers.3.attention.q_proj", "vision_tower.transformer.layers.3.attention.o_proj", "vision_tower.transformer.layers.4.feed_forward.gate_proj", "vision_tower.transformer.layers.4.feed_forward.up_proj", "vision_tower.transformer.layers.4.feed_forward.down_proj", "vision_tower.transformer.layers.4.attention.k_proj", "vision_tower.transformer.layers.4.attention.v_proj", "vision_tower.transformer.layers.4.attention.q_proj", "vision_tower.transformer.layers.4.attention.o_proj", "vision_tower.transformer.layers.5.feed_forward.gate_proj", "vision_tower.transformer.layers.5.feed_forward.up_proj", "vision_tower.transformer.layers.5.feed_forward.down_proj", "vision_tower.transformer.layers.5.attention.k_proj", "vision_tower.transformer.layers.5.attention.v_proj", "vision_tower.transformer.layers.5.attention.q_proj", "vision_tower.transformer.layers.5.attention.o_proj", "vision_tower.transformer.layers.6.feed_forward.gate_proj", "vision_tower.transformer.layers.6.feed_forward.up_proj", "vision_tower.transformer.layers.6.feed_forward.down_proj", "vision_tower.transformer.layers.6.attention.k_proj", "vision_tower.transformer.layers.6.attention.v_proj", "vision_tower.transformer.layers.6.attention.q_proj", "vision_tower.transformer.layers.6.attention.o_proj", "vision_tower.transformer.layers.7.feed_forward.gate_proj", "vision_tower.transformer.layers.7.feed_forward.up_proj", "vision_tower.transformer.layers.7.feed_forward.down_proj", "vision_tower.transformer.layers.7.attention.k_proj", "vision_tower.transformer.layers.7.attention.v_proj", "vision_tower.transformer.layers.7.attention.q_proj", "vision_tower.transformer.layers.7.attention.o_proj", "vision_tower.transformer.layers.8.feed_forward.gate_proj", "vision_tower.transformer.layers.8.feed_forward.up_proj", "vision_tower.transformer.layers.8.feed_forward.down_proj", "vision_tower.transformer.layers.8.attention.k_proj", "vision_tower.transformer.layers.8.attention.v_proj", "vision_tower.transformer.layers.8.attention.q_proj", "vision_tower.transformer.layers.8.attention.o_proj", "vision_tower.transformer.layers.9.feed_forward.gate_proj", "vision_tower.transformer.layers.9.feed_forward.up_proj", "vision_tower.transformer.layers.9.feed_forward.down_proj", "vision_tower.transformer.layers.9.attention.k_proj", "vision_tower.transformer.layers.9.attention.v_proj", "vision_tower.transformer.layers.9.attention.q_proj", "vision_tower.transformer.layers.9.attention.o_proj", "vision_tower.transformer.layers.10.feed_forward.gate_proj", "vision_tower.transformer.layers.10.feed_forward.up_proj", "vision_tower.transformer.layers.10.feed_forward.down_proj", "vision_tower.transformer.layers.10.attention.k_proj", "vision_tower.transformer.layers.10.attention.v_proj", "vision_tower.transformer.layers.10.attention.q_proj", "vision_tower.transformer.layers.10.attention.o_proj", "vision_tower.transformer.layers.11.feed_forward.gate_proj", "vision_tower.transformer.layers.11.feed_forward.up_proj", "vision_tower.transformer.layers.11.feed_forward.down_proj", "vision_tower.transformer.layers.11.attention.k_proj", "vision_tower.transformer.layers.11.attention.v_proj", "vision_tower.transformer.layers.11.attention.q_proj", "vision_tower.transformer.layers.11.attention.o_proj", "vision_tower.transformer.layers.12.feed_forward.gate_proj", "vision_tower.transformer.layers.12.feed_forward.up_proj", "vision_tower.transformer.layers.12.feed_forward.down_proj", "vision_tower.transformer.layers.12.attention.k_proj", "vision_tower.transformer.layers.12.attention.v_proj", "vision_tower.transformer.layers.12.attention.q_proj", "vision_tower.transformer.layers.12.attention.o_proj", "vision_tower.transformer.layers.13.feed_forward.gate_proj", "vision_tower.transformer.layers.13.feed_forward.up_proj", "vision_tower.transformer.layers.13.feed_forward.down_proj", "vision_tower.transformer.layers.13.attention.k_proj", "vision_tower.transformer.layers.13.attention.v_proj", "vision_tower.transformer.layers.13.attention.q_proj", "vision_tower.transformer.layers.13.attention.o_proj", "vision_tower.transformer.layers.14.feed_forward.gate_proj", "vision_tower.transformer.layers.14.feed_forward.up_proj", "vision_tower.transformer.layers.14.feed_forward.down_proj", "vision_tower.transformer.layers.14.attention.k_proj", "vision_tower.transformer.layers.14.attention.v_proj", "vision_tower.transformer.layers.14.attention.q_proj", "vision_tower.transformer.layers.14.attention.o_proj", "vision_tower.transformer.layers.15.feed_forward.gate_proj", "vision_tower.transformer.layers.15.feed_forward.up_proj", "vision_tower.transformer.layers.15.feed_forward.down_proj", "vision_tower.transformer.layers.15.attention.k_proj", "vision_tower.transformer.layers.15.attention.v_proj", "vision_tower.transformer.layers.15.attention.q_proj", "vision_tower.transformer.layers.15.attention.o_proj", "vision_tower.transformer.layers.16.feed_forward.gate_proj", "vision_tower.transformer.layers.16.feed_forward.up_proj", "vision_tower.transformer.layers.16.feed_forward.down_proj", "vision_tower.transformer.layers.16.attention.k_proj", "vision_tower.transformer.layers.16.attention.v_proj", "vision_tower.transformer.layers.16.attention.q_proj", "vision_tower.transformer.layers.16.attention.o_proj", "vision_tower.transformer.layers.17.feed_forward.gate_proj", "vision_tower.transformer.layers.17.feed_forward.up_proj", "vision_tower.transformer.layers.17.feed_forward.down_proj", "vision_tower.transformer.layers.17.attention.k_proj", "vision_tower.transformer.layers.17.attention.v_proj", "vision_tower.transformer.layers.17.attention.q_proj", "vision_tower.transformer.layers.17.attention.o_proj", "vision_tower.transformer.layers.18.feed_forward.gate_proj", "vision_tower.transformer.layers.18.feed_forward.up_proj", "vision_tower.transformer.layers.18.feed_forward.down_proj", "vision_tower.transformer.layers.18.attention.k_proj", "vision_tower.transformer.layers.18.attention.v_proj", "vision_tower.transformer.layers.18.attention.q_proj", "vision_tower.transformer.layers.18.attention.o_proj", "vision_tower.transformer.layers.19.feed_forward.gate_proj", "vision_tower.transformer.layers.19.feed_forward.up_proj", "vision_tower.transformer.layers.19.feed_forward.down_proj", "vision_tower.transformer.layers.19.attention.k_proj", "vision_tower.transformer.layers.19.attention.v_proj", "vision_tower.transformer.layers.19.attention.q_proj", "vision_tower.transformer.layers.19.attention.o_proj", "vision_tower.transformer.layers.20.feed_forward.gate_proj", "vision_tower.transformer.layers.20.feed_forward.up_proj", "vision_tower.transformer.layers.20.feed_forward.down_proj", "vision_tower.transformer.layers.20.attention.k_proj", "vision_tower.transformer.layers.20.attention.v_proj", "vision_tower.transformer.layers.20.attention.q_proj", "vision_tower.transformer.layers.20.attention.o_proj", "vision_tower.transformer.layers.21.feed_forward.gate_proj", "vision_tower.transformer.layers.21.feed_forward.up_proj", "vision_tower.transformer.layers.21.feed_forward.down_proj", "vision_tower.transformer.layers.21.attention.k_proj", "vision_tower.transformer.layers.21.attention.v_proj", "vision_tower.transformer.layers.21.attention.q_proj", "vision_tower.transformer.layers.21.attention.o_proj", "vision_tower.transformer.layers.22.feed_forward.gate_proj", "vision_tower.transformer.layers.22.feed_forward.up_proj", "vision_tower.transformer.layers.22.feed_forward.down_proj", "vision_tower.transformer.layers.22.attention.k_proj", "vision_tower.transformer.layers.22.attention.v_proj", "vision_tower.transformer.layers.22.attention.q_proj", "vision_tower.transformer.layers.22.attention.o_proj", "vision_tower.transformer.layers.23.feed_forward.gate_proj", "vision_tower.transformer.layers.23.feed_forward.up_proj", "vision_tower.transformer.layers.23.feed_forward.down_proj", "vision_tower.transformer.layers.23.attention.k_proj", "vision_tower.transformer.layers.23.attention.v_proj", "vision_tower.transformer.layers.23.attention.q_proj", "vision_tower.transformer.layers.23.attention.o_proj", "vision_tower.transformer.layers.24.feed_forward.gate_proj", "vision_tower.transformer.layers.24.feed_forward.up_proj", "vision_tower.transformer.layers.24.feed_forward.down_proj", "vision_tower.transformer.layers.24.attention.k_proj", "vision_tower.transformer.layers.24.attention.v_proj", "vision_tower.transformer.layers.24.attention.q_proj", "vision_tower.transformer.layers.24.attention.o_proj", "vision_tower.transformer.layers.25.feed_forward.gate_proj", "vision_tower.transformer.layers.25.feed_forward.up_proj", "vision_tower.transformer.layers.25.feed_forward.down_proj", "vision_tower.transformer.layers.25.attention.k_proj", "vision_tower.transformer.layers.25.attention.v_proj", "vision_tower.transformer.layers.25.attention.q_proj", "vision_tower.transformer.layers.25.attention.o_proj", "vision_tower.transformer.layers.26.feed_forward.gate_proj", "vision_tower.transformer.layers.26.feed_forward.up_proj", "vision_tower.transformer.layers.26.feed_forward.down_proj", "vision_tower.transformer.layers.26.attention.k_proj", "vision_tower.transformer.layers.26.attention.v_proj", "vision_tower.transformer.layers.26.attention.q_proj", "vision_tower.transformer.layers.26.attention.o_proj", "vision_tower.transformer.layers.27.feed_forward.gate_proj", "vision_tower.transformer.layers.27.feed_forward.up_proj", "vision_tower.transformer.layers.27.feed_forward.down_proj", "vision_tower.transformer.layers.27.attention.k_proj", "vision_tower.transformer.layers.27.attention.v_proj", "vision_tower.transformer.layers.27.attention.q_proj", "vision_tower.transformer.layers.27.attention.o_proj", "vision_tower.transformer.layers.28.feed_forward.gate_proj", "vision_tower.transformer.layers.28.feed_forward.up_proj", "vision_tower.transformer.layers.28.feed_forward.down_proj", "vision_tower.transformer.layers.28.attention.k_proj", "vision_tower.transformer.layers.28.attention.v_proj", "vision_tower.transformer.layers.28.attention.q_proj", "vision_tower.transformer.layers.28.attention.o_proj", "vision_tower.transformer.layers.29.feed_forward.gate_proj", "vision_tower.transformer.layers.29.feed_forward.up_proj", "vision_tower.transformer.layers.29.feed_forward.down_proj", "vision_tower.transformer.layers.29.attention.k_proj", "vision_tower.transformer.layers.29.attention.v_proj", "vision_tower.transformer.layers.29.attention.q_proj", "vision_tower.transformer.layers.29.attention.o_proj", "vision_tower.transformer.layers.30.feed_forward.gate_proj", "vision_tower.transformer.layers.30.feed_forward.up_proj", "vision_tower.transformer.layers.30.feed_forward.down_proj", "vision_tower.transformer.layers.30.attention.k_proj", "vision_tower.transformer.layers.30.attention.v_proj", "vision_tower.transformer.layers.30.attention.q_proj", "vision_tower.transformer.layers.30.attention.o_proj", "vision_tower.transformer.layers.31.feed_forward.gate_proj", "vision_tower.transformer.layers.31.feed_forward.up_proj", "vision_tower.transformer.layers.31.feed_forward.down_proj", "vision_tower.transformer.layers.31.attention.k_proj", "vision_tower.transformer.layers.31.attention.v_proj", "vision_tower.transformer.layers.31.attention.q_proj", "vision_tower.transformer.layers.31.attention.o_proj", "vision_tower.transformer.layers.32.feed_forward.gate_proj", "vision_tower.transformer.layers.32.feed_forward.up_proj", "vision_tower.transformer.layers.32.feed_forward.down_proj", "vision_tower.transformer.layers.32.attention.k_proj", "vision_tower.transformer.layers.32.attention.v_proj", "vision_tower.transformer.layers.32.attention.q_proj", "vision_tower.transformer.layers.32.attention.o_proj", "vision_tower.transformer.layers.33.feed_forward.gate_proj", "vision_tower.transformer.layers.33.feed_forward.up_proj", "vision_tower.transformer.layers.33.feed_forward.down_proj", "vision_tower.transformer.layers.33.attention.k_proj", "vision_tower.transformer.layers.33.attention.v_proj", "vision_tower.transformer.layers.33.attention.q_proj", "vision_tower.transformer.layers.33.attention.o_proj", "vision_tower.transformer.layers.34.feed_forward.gate_proj", "vision_tower.transformer.layers.34.feed_forward.up_proj", "vision_tower.transformer.layers.34.feed_forward.down_proj", "vision_tower.transformer.layers.34.attention.k_proj", "vision_tower.transformer.layers.34.attention.v_proj", "vision_tower.transformer.layers.34.attention.q_proj", "vision_tower.transformer.layers.34.attention.o_proj", "vision_tower.transformer.layers.35.feed_forward.gate_proj", "vision_tower.transformer.layers.35.feed_forward.up_proj", "vision_tower.transformer.layers.35.feed_forward.down_proj", "vision_tower.transformer.layers.35.attention.k_proj", "vision_tower.transformer.layers.35.attention.v_proj", "vision_tower.transformer.layers.35.attention.q_proj", "vision_tower.transformer.layers.35.attention.o_proj", "vision_tower.transformer.layers.36.feed_forward.gate_proj", "vision_tower.transformer.layers.36.feed_forward.up_proj", "vision_tower.transformer.layers.36.feed_forward.down_proj", "vision_tower.transformer.layers.36.attention.k_proj", "vision_tower.transformer.layers.36.attention.v_proj", "vision_tower.transformer.layers.36.attention.q_proj", "vision_tower.transformer.layers.36.attention.o_proj", "vision_tower.transformer.layers.37.feed_forward.gate_proj", "vision_tower.transformer.layers.37.feed_forward.up_proj", "vision_tower.transformer.layers.37.feed_forward.down_proj", "vision_tower.transformer.layers.37.attention.k_proj", "vision_tower.transformer.layers.37.attention.v_proj", "vision_tower.transformer.layers.37.attention.q_proj", "vision_tower.transformer.layers.37.attention.o_proj", "vision_tower.transformer.layers.38.feed_forward.gate_proj", "vision_tower.transformer.layers.38.feed_forward.up_proj", "vision_tower.transformer.layers.38.feed_forward.down_proj", "vision_tower.transformer.layers.38.attention.k_proj", "vision_tower.transformer.layers.38.attention.v_proj", "vision_tower.transformer.layers.38.attention.q_proj", "vision_tower.transformer.layers.38.attention.o_proj", "vision_tower.transformer.layers.39.feed_forward.gate_proj", "vision_tower.transformer.layers.39.feed_forward.up_proj", "vision_tower.transformer.layers.39.feed_forward.down_proj", "vision_tower.transformer.layers.39.attention.k_proj", "vision_tower.transformer.layers.39.attention.v_proj", "vision_tower.transformer.layers.39.attention.q_proj", "vision_tower.transformer.layers.39.attention.o_proj", "multi_modal_projector.linear_1", "multi_modal_projector.linear_2", "language_model.lm_head" ], "kv_cache_scheme": null, "quant_method": "compressed-tensors", "quantization_status": "compressed" }, "text_config": { "hidden_size": 12288, "intermediate_size": 28672, "is_composition": true, "max_position_embeddings": 131072, "model_type": "mistral", "norm_eps": 1e-05, "num_attention_heads": 96, "num_hidden_layers": 88, "num_key_value_heads": 8, "rms_norm_eps": 1e-05, "rope_theta": 1000000000.0, "sliding_window": null, "vocab_size": 32768 }, "torch_dtype": "bfloat16", "transformers_version": "4.48.2", "vision_config": { "head_dim": 88, "hidden_act": "silu", "hidden_size": 1408, "image_break_token_id": 14, "image_end_token_id": 15, "image_size": 1024, "image_token_id": 10, "intermediate_size": 6144, "model_type": "pixtral", "num_hidden_layers": 40, "patch_size": 16, "rope_theta": 10000.0 }, "vision_feature_layer": -1, "vision_feature_select_strategy": "full" }