{ "architectures": [ "LlavaForConditionalGeneration" ], "ignore_index": -100, "image_seq_length": 576, "image_token_index": 32000, "model_type": "llava", "projector_hidden_act": "gelu", "text_config": { "hidden_size": 5120, "intermediate_size": 14336, "max_position_embeddings": 1024000, "model_type": "mistral", "num_hidden_layers": 40, "num_key_value_heads": 8, "rms_norm_eps": 1e-05, "rope_theta": 1000000000.0, "sliding_window": null, "vocab_size": 131072 }, "torch_dtype": "bfloat16", "transformers_version": "4.45.0.dev0", "vision_config": { "head_dim": 64, "hidden_activation": "gelu", "image_size": 1024, "model_type": "pixtral", "patch_size": 16, "rope_theta": 10000.0, "tie_word_embeddings": false }, "vision_feature_layer": -2, "vision_feature_select_strategy": "default" }