{ "_name_or_path": "microsoft/Phi-3-mini-4k-instruct", "architectures": "Phi3ForCausalLM", "attention_dropout": 0.0, "auto_map": { "AutoConfig": "configuration_phi3.Phi3Config", "AutoModelForCausalLM": "modeling_phi3.Phi3ForCausalLM" }, "bos_token_id": 1, "config": { "freeze": false, "spatial_cluster_rate0": 64, "spatial_cluster_rate1": 32, "spatial_cluster_rate2": 16, "temporal_cluster_rate": 0.0625, "use_cluster": 0, "vision_tune": false }, "embd_pdrop": 0.0, "eos_token_id": 32000, "freeze_mm_mlp_adapter": false, "hidden_act": "silu", "hidden_size": 3072, "image_aspect_ratio": "square", "image_grid_pinpoints": null, "initializer_range": 0.02, "intermediate_size": 8192, "max_position_embeddings": 4096, "mm_hidden_size": 1408, "mm_projector_lr": null, "mm_projector_type": "mlp2x_gelu", "mm_use_box_start_end": false, "mm_use_im_patch_token": false, "mm_use_im_start_end": false, "mm_vision_select_feature": "patch", "mm_vision_select_layer": -2, "mm_vision_tower": "OpenGVLab/InternVideo2-Stage2_1B-224p-f4/InternVideo2-stage2_1b-224p-f4.pt", "model_type": "ChatUniVi", "num_attention_heads": 32, "num_hidden_layers": 32, "num_key_value_heads": 32, "original_max_position_embeddings": 4096, "pad_token_id": 32000, "resid_pdrop": 0.0, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 10000.0, "s2": false, "s2_scales": "336,672,1008", "sliding_window": 2048, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.41.0.dev0", "tune_mm_mlp_adapter": true, "use_cache": true, "use_mm_proj": true, "vocab_size": 32064 }