{ "_name_or_path": "/home/mgoin/.cache/huggingface/hub/models--openbmb--MiniCPM-V-2_6/snapshots/c13ade7d9008d590524921b46f614b00135d75ee", "architectures": [ "MiniCPMV" ], "attention_dropout": 0.0, "auto_map": { "AutoConfig": "configuration_minicpm.MiniCPMVConfig", "AutoModel": "modeling_minicpmv.MiniCPMV", "AutoModelForCausalLM": "modeling_minicpmv.MiniCPMV" }, "batch_vision_input": true, "bos_token_id": 151643, "drop_vision_last_layer": false, "eos_token_id": 151645, "hidden_act": "silu", "hidden_size": 3584, "image_size": 448, "initializer_range": 0.02, "intermediate_size": 18944, "max_position_embeddings": 32768, "max_window_layers": 28, "model_type": "minicpmv", "num_attention_heads": 28, "num_hidden_layers": 28, "num_key_value_heads": 4, "patch_size": 14, "quantization_config": { "config_groups": { "group_0": { "input_activations": { "actorder": null, "block_structure": null, "dynamic": true, "group_size": null, "num_bits": 8, "observer": "memoryless", "observer_kwargs": {}, "strategy": "token", "symmetric": true, "type": "float" }, "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": "float" } } }, "format": "float-quantized", "global_compression_ratio": 1.1482456009156867, "ignore": [ "llm.lm_head", "vpm.encoder.layers.0.self_attn.k_proj", "vpm.encoder.layers.0.self_attn.v_proj", "vpm.encoder.layers.0.self_attn.q_proj", "vpm.encoder.layers.0.self_attn.out_proj", "vpm.encoder.layers.0.mlp.fc1", "vpm.encoder.layers.0.mlp.fc2", "vpm.encoder.layers.1.self_attn.k_proj", "vpm.encoder.layers.1.self_attn.v_proj", "vpm.encoder.layers.1.self_attn.q_proj", "vpm.encoder.layers.1.self_attn.out_proj", "vpm.encoder.layers.1.mlp.fc1", "vpm.encoder.layers.1.mlp.fc2", "vpm.encoder.layers.2.self_attn.k_proj", "vpm.encoder.layers.2.self_attn.v_proj", "vpm.encoder.layers.2.self_attn.q_proj", "vpm.encoder.layers.2.self_attn.out_proj", "vpm.encoder.layers.2.mlp.fc1", "vpm.encoder.layers.2.mlp.fc2", "vpm.encoder.layers.3.self_attn.k_proj", "vpm.encoder.layers.3.self_attn.v_proj", "vpm.encoder.layers.3.self_attn.q_proj", "vpm.encoder.layers.3.self_attn.out_proj", "vpm.encoder.layers.3.mlp.fc1", "vpm.encoder.layers.3.mlp.fc2", "vpm.encoder.layers.4.self_attn.k_proj", "vpm.encoder.layers.4.self_attn.v_proj", "vpm.encoder.layers.4.self_attn.q_proj", "vpm.encoder.layers.4.self_attn.out_proj", "vpm.encoder.layers.4.mlp.fc1", "vpm.encoder.layers.4.mlp.fc2", "vpm.encoder.layers.5.self_attn.k_proj", "vpm.encoder.layers.5.self_attn.v_proj", "vpm.encoder.layers.5.self_attn.q_proj", "vpm.encoder.layers.5.self_attn.out_proj", "vpm.encoder.layers.5.mlp.fc1", "vpm.encoder.layers.5.mlp.fc2", "vpm.encoder.layers.6.self_attn.k_proj", "vpm.encoder.layers.6.self_attn.v_proj", "vpm.encoder.layers.6.self_attn.q_proj", "vpm.encoder.layers.6.self_attn.out_proj", "vpm.encoder.layers.6.mlp.fc1", "vpm.encoder.layers.6.mlp.fc2", "vpm.encoder.layers.7.self_attn.k_proj", "vpm.encoder.layers.7.self_attn.v_proj", "vpm.encoder.layers.7.self_attn.q_proj", "vpm.encoder.layers.7.self_attn.out_proj", "vpm.encoder.layers.7.mlp.fc1", "vpm.encoder.layers.7.mlp.fc2", "vpm.encoder.layers.8.self_attn.k_proj", "vpm.encoder.layers.8.self_attn.v_proj", "vpm.encoder.layers.8.self_attn.q_proj", "vpm.encoder.layers.8.self_attn.out_proj", "vpm.encoder.layers.8.mlp.fc1", "vpm.encoder.layers.8.mlp.fc2", "vpm.encoder.layers.9.self_attn.k_proj", "vpm.encoder.layers.9.self_attn.v_proj", "vpm.encoder.layers.9.self_attn.q_proj", "vpm.encoder.layers.9.self_attn.out_proj", "vpm.encoder.layers.9.mlp.fc1", "vpm.encoder.layers.9.mlp.fc2", "vpm.encoder.layers.10.self_attn.k_proj", "vpm.encoder.layers.10.self_attn.v_proj", "vpm.encoder.layers.10.self_attn.q_proj", "vpm.encoder.layers.10.self_attn.out_proj", "vpm.encoder.layers.10.mlp.fc1", "vpm.encoder.layers.10.mlp.fc2", "vpm.encoder.layers.11.self_attn.k_proj", "vpm.encoder.layers.11.self_attn.v_proj", "vpm.encoder.layers.11.self_attn.q_proj", "vpm.encoder.layers.11.self_attn.out_proj", "vpm.encoder.layers.11.mlp.fc1", "vpm.encoder.layers.11.mlp.fc2", "vpm.encoder.layers.12.self_attn.k_proj", "vpm.encoder.layers.12.self_attn.v_proj", "vpm.encoder.layers.12.self_attn.q_proj", "vpm.encoder.layers.12.self_attn.out_proj", "vpm.encoder.layers.12.mlp.fc1", "vpm.encoder.layers.12.mlp.fc2", "vpm.encoder.layers.13.self_attn.k_proj", "vpm.encoder.layers.13.self_attn.v_proj", "vpm.encoder.layers.13.self_attn.q_proj", "vpm.encoder.layers.13.self_attn.out_proj", "vpm.encoder.layers.13.mlp.fc1", "vpm.encoder.layers.13.mlp.fc2", "vpm.encoder.layers.14.self_attn.k_proj", "vpm.encoder.layers.14.self_attn.v_proj", "vpm.encoder.layers.14.self_attn.q_proj", "vpm.encoder.layers.14.self_attn.out_proj", "vpm.encoder.layers.14.mlp.fc1", "vpm.encoder.layers.14.mlp.fc2", "vpm.encoder.layers.15.self_attn.k_proj", "vpm.encoder.layers.15.self_attn.v_proj", "vpm.encoder.layers.15.self_attn.q_proj", "vpm.encoder.layers.15.self_attn.out_proj", "vpm.encoder.layers.15.mlp.fc1", "vpm.encoder.layers.15.mlp.fc2", "vpm.encoder.layers.16.self_attn.k_proj", "vpm.encoder.layers.16.self_attn.v_proj", "vpm.encoder.layers.16.self_attn.q_proj", "vpm.encoder.layers.16.self_attn.out_proj", "vpm.encoder.layers.16.mlp.fc1", "vpm.encoder.layers.16.mlp.fc2", "vpm.encoder.layers.17.self_attn.k_proj", "vpm.encoder.layers.17.self_attn.v_proj", "vpm.encoder.layers.17.self_attn.q_proj", "vpm.encoder.layers.17.self_attn.out_proj", "vpm.encoder.layers.17.mlp.fc1", "vpm.encoder.layers.17.mlp.fc2", "vpm.encoder.layers.18.self_attn.k_proj", "vpm.encoder.layers.18.self_attn.v_proj", "vpm.encoder.layers.18.self_attn.q_proj", "vpm.encoder.layers.18.self_attn.out_proj", "vpm.encoder.layers.18.mlp.fc1", "vpm.encoder.layers.18.mlp.fc2", "vpm.encoder.layers.19.self_attn.k_proj", "vpm.encoder.layers.19.self_attn.v_proj", "vpm.encoder.layers.19.self_attn.q_proj", "vpm.encoder.layers.19.self_attn.out_proj", "vpm.encoder.layers.19.mlp.fc1", "vpm.encoder.layers.19.mlp.fc2", "vpm.encoder.layers.20.self_attn.k_proj", "vpm.encoder.layers.20.self_attn.v_proj", "vpm.encoder.layers.20.self_attn.q_proj", "vpm.encoder.layers.20.self_attn.out_proj", "vpm.encoder.layers.20.mlp.fc1", "vpm.encoder.layers.20.mlp.fc2", "vpm.encoder.layers.21.self_attn.k_proj", "vpm.encoder.layers.21.self_attn.v_proj", "vpm.encoder.layers.21.self_attn.q_proj", "vpm.encoder.layers.21.self_attn.out_proj", "vpm.encoder.layers.21.mlp.fc1", "vpm.encoder.layers.21.mlp.fc2", "vpm.encoder.layers.22.self_attn.k_proj", "vpm.encoder.layers.22.self_attn.v_proj", "vpm.encoder.layers.22.self_attn.q_proj", "vpm.encoder.layers.22.self_attn.out_proj", "vpm.encoder.layers.22.mlp.fc1", "vpm.encoder.layers.22.mlp.fc2", "vpm.encoder.layers.23.self_attn.k_proj", "vpm.encoder.layers.23.self_attn.v_proj", "vpm.encoder.layers.23.self_attn.q_proj", "vpm.encoder.layers.23.self_attn.out_proj", "vpm.encoder.layers.23.mlp.fc1", "vpm.encoder.layers.23.mlp.fc2", "vpm.encoder.layers.24.self_attn.k_proj", "vpm.encoder.layers.24.self_attn.v_proj", "vpm.encoder.layers.24.self_attn.q_proj", "vpm.encoder.layers.24.self_attn.out_proj", "vpm.encoder.layers.24.mlp.fc1", "vpm.encoder.layers.24.mlp.fc2", "vpm.encoder.layers.25.self_attn.k_proj", "vpm.encoder.layers.25.self_attn.v_proj", "vpm.encoder.layers.25.self_attn.q_proj", "vpm.encoder.layers.25.self_attn.out_proj", "vpm.encoder.layers.25.mlp.fc1", "vpm.encoder.layers.25.mlp.fc2", "vpm.encoder.layers.26.self_attn.k_proj", "vpm.encoder.layers.26.self_attn.v_proj", "vpm.encoder.layers.26.self_attn.q_proj", "vpm.encoder.layers.26.self_attn.out_proj", "vpm.encoder.layers.26.mlp.fc1", "vpm.encoder.layers.26.mlp.fc2", "resampler.kv_proj", "resampler.attn.out_proj" ], "kv_cache_scheme": null, "quant_method": "compressed-tensors", "quantization_status": "compressed" }, "query_num": 64, "rms_norm_eps": 1e-06, "rope_scaling": null, "rope_theta": 1000000.0, "slice_config": { "max_slice_nums": 9, "model_type": "minicpmv" }, "slice_mode": true, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.46.0.dev0", "use_cache": true, "use_image_id": true, "use_sliding_window": false, "version": 2.6, "vision_batch_size": 16, "vision_config": { "hidden_size": 1152, "image_size": 980, "intermediate_size": 4304, "model_type": "siglip_vision_model", "num_attention_heads": 16, "num_hidden_layers": 27, "patch_size": 14 }, "vocab_size": 151666 }