| { |
| "_name_or_path": "THUDM/glm-4v-9b", |
| "add_bias_linear": false, |
| "add_qkv_bias": true, |
| "apply_query_key_layer_scaling": true, |
| "apply_residual_connection_post_layernorm": false, |
| "architectures": [ |
| "ChatGLMForConditionalGeneration" |
| ], |
| "attention_dropout": 0.0, |
| "attention_softmax_in_fp32": true, |
| "auto_map": { |
| "AutoConfig": "THUDM/glm-4v-9b--configuration_chatglm.ChatGLMConfig", |
| "AutoModel": "THUDM/glm-4v-9b--modeling_chatglm.ChatGLMForConditionalGeneration", |
| "AutoModelForCausalLM": "THUDM/glm-4v-9b--modeling_chatglm.ChatGLMForConditionalGeneration", |
| "AutoModelForSeq2SeqLM": "THUDM/glm-4v-9b--modeling_chatglm.ChatGLMForConditionalGeneration", |
| "AutoModelForSequenceClassification": "THUDM/glm-4v-9b--modeling_chatglm.ChatGLMForSequenceClassification" |
| }, |
| "bias_dropout_fusion": true, |
| "boi_token_id": 151339, |
| "classifier_dropout": null, |
| "eoi_token_id": 151340, |
| "eos_token_id": [ |
| 151329, |
| 151336, |
| 151338 |
| ], |
| "ffn_hidden_size": 13696, |
| "fp32_residual_connection": false, |
| "hidden_dropout": 0.0, |
| "hidden_size": 4096, |
| "kv_channels": 128, |
| "layernorm_epsilon": 1.5625e-07, |
| "model_type": "chatglm", |
| "multi_query_attention": true, |
| "multi_query_group_num": 2, |
| "num_attention_heads": 32, |
| "num_layers": 40, |
| "original_rope": true, |
| "pad_token_id": 151329, |
| "padded_vocab_size": 151552, |
| "post_layer_norm": true, |
| "pre_seq_len": null, |
| "prefix_projection": false, |
| "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": 1.1115688433018316, |
| "ignore": [ |
| "transformer.output_layer", |
| "transformer.vision.transformer.layers.0.attention.query_key_value", |
| "transformer.vision.transformer.layers.0.attention.dense", |
| "transformer.vision.transformer.layers.0.mlp.fc1", |
| "transformer.vision.transformer.layers.0.mlp.fc2", |
| "transformer.vision.transformer.layers.1.attention.query_key_value", |
| "transformer.vision.transformer.layers.1.attention.dense", |
| "transformer.vision.transformer.layers.1.mlp.fc1", |
| "transformer.vision.transformer.layers.1.mlp.fc2", |
| "transformer.vision.transformer.layers.2.attention.query_key_value", |
| "transformer.vision.transformer.layers.2.attention.dense", |
| "transformer.vision.transformer.layers.2.mlp.fc1", |
| "transformer.vision.transformer.layers.2.mlp.fc2", |
| "transformer.vision.transformer.layers.3.attention.query_key_value", |
| "transformer.vision.transformer.layers.3.attention.dense", |
| "transformer.vision.transformer.layers.3.mlp.fc1", |
| "transformer.vision.transformer.layers.3.mlp.fc2", |
| "transformer.vision.transformer.layers.4.attention.query_key_value", |
| "transformer.vision.transformer.layers.4.attention.dense", |
| "transformer.vision.transformer.layers.4.mlp.fc1", |
| "transformer.vision.transformer.layers.4.mlp.fc2", |
| "transformer.vision.transformer.layers.5.attention.query_key_value", |
| "transformer.vision.transformer.layers.5.attention.dense", |
| "transformer.vision.transformer.layers.5.mlp.fc1", |
| "transformer.vision.transformer.layers.5.mlp.fc2", |
| "transformer.vision.transformer.layers.6.attention.query_key_value", |
| "transformer.vision.transformer.layers.6.attention.dense", |
| "transformer.vision.transformer.layers.6.mlp.fc1", |
| "transformer.vision.transformer.layers.6.mlp.fc2", |
| "transformer.vision.transformer.layers.7.attention.query_key_value", |
| "transformer.vision.transformer.layers.7.attention.dense", |
| "transformer.vision.transformer.layers.7.mlp.fc1", |
| "transformer.vision.transformer.layers.7.mlp.fc2", |
| "transformer.vision.transformer.layers.8.attention.query_key_value", |
| "transformer.vision.transformer.layers.8.attention.dense", |
| "transformer.vision.transformer.layers.8.mlp.fc1", |
| "transformer.vision.transformer.layers.8.mlp.fc2", |
| "transformer.vision.transformer.layers.9.attention.query_key_value", |
| "transformer.vision.transformer.layers.9.attention.dense", |
| "transformer.vision.transformer.layers.9.mlp.fc1", |
| "transformer.vision.transformer.layers.9.mlp.fc2", |
| "transformer.vision.transformer.layers.10.attention.query_key_value", |
| "transformer.vision.transformer.layers.10.attention.dense", |
| "transformer.vision.transformer.layers.10.mlp.fc1", |
| "transformer.vision.transformer.layers.10.mlp.fc2", |
| "transformer.vision.transformer.layers.11.attention.query_key_value", |
| "transformer.vision.transformer.layers.11.attention.dense", |
| "transformer.vision.transformer.layers.11.mlp.fc1", |
| "transformer.vision.transformer.layers.11.mlp.fc2", |
| "transformer.vision.transformer.layers.12.attention.query_key_value", |
| "transformer.vision.transformer.layers.12.attention.dense", |
| "transformer.vision.transformer.layers.12.mlp.fc1", |
| "transformer.vision.transformer.layers.12.mlp.fc2", |
| "transformer.vision.transformer.layers.13.attention.query_key_value", |
| "transformer.vision.transformer.layers.13.attention.dense", |
| "transformer.vision.transformer.layers.13.mlp.fc1", |
| "transformer.vision.transformer.layers.13.mlp.fc2", |
| "transformer.vision.transformer.layers.14.attention.query_key_value", |
| "transformer.vision.transformer.layers.14.attention.dense", |
| "transformer.vision.transformer.layers.14.mlp.fc1", |
| "transformer.vision.transformer.layers.14.mlp.fc2", |
| "transformer.vision.transformer.layers.15.attention.query_key_value", |
| "transformer.vision.transformer.layers.15.attention.dense", |
| "transformer.vision.transformer.layers.15.mlp.fc1", |
| "transformer.vision.transformer.layers.15.mlp.fc2", |
| "transformer.vision.transformer.layers.16.attention.query_key_value", |
| "transformer.vision.transformer.layers.16.attention.dense", |
| "transformer.vision.transformer.layers.16.mlp.fc1", |
| "transformer.vision.transformer.layers.16.mlp.fc2", |
| "transformer.vision.transformer.layers.17.attention.query_key_value", |
| "transformer.vision.transformer.layers.17.attention.dense", |
| "transformer.vision.transformer.layers.17.mlp.fc1", |
| "transformer.vision.transformer.layers.17.mlp.fc2", |
| "transformer.vision.transformer.layers.18.attention.query_key_value", |
| "transformer.vision.transformer.layers.18.attention.dense", |
| "transformer.vision.transformer.layers.18.mlp.fc1", |
| "transformer.vision.transformer.layers.18.mlp.fc2", |
| "transformer.vision.transformer.layers.19.attention.query_key_value", |
| "transformer.vision.transformer.layers.19.attention.dense", |
| "transformer.vision.transformer.layers.19.mlp.fc1", |
| "transformer.vision.transformer.layers.19.mlp.fc2", |
| "transformer.vision.transformer.layers.20.attention.query_key_value", |
| "transformer.vision.transformer.layers.20.attention.dense", |
| "transformer.vision.transformer.layers.20.mlp.fc1", |
| "transformer.vision.transformer.layers.20.mlp.fc2", |
| "transformer.vision.transformer.layers.21.attention.query_key_value", |
| "transformer.vision.transformer.layers.21.attention.dense", |
| "transformer.vision.transformer.layers.21.mlp.fc1", |
| "transformer.vision.transformer.layers.21.mlp.fc2", |
| "transformer.vision.transformer.layers.22.attention.query_key_value", |
| "transformer.vision.transformer.layers.22.attention.dense", |
| "transformer.vision.transformer.layers.22.mlp.fc1", |
| "transformer.vision.transformer.layers.22.mlp.fc2", |
| "transformer.vision.transformer.layers.23.attention.query_key_value", |
| "transformer.vision.transformer.layers.23.attention.dense", |
| "transformer.vision.transformer.layers.23.mlp.fc1", |
| "transformer.vision.transformer.layers.23.mlp.fc2", |
| "transformer.vision.transformer.layers.24.attention.query_key_value", |
| "transformer.vision.transformer.layers.24.attention.dense", |
| "transformer.vision.transformer.layers.24.mlp.fc1", |
| "transformer.vision.transformer.layers.24.mlp.fc2", |
| "transformer.vision.transformer.layers.25.attention.query_key_value", |
| "transformer.vision.transformer.layers.25.attention.dense", |
| "transformer.vision.transformer.layers.25.mlp.fc1", |
| "transformer.vision.transformer.layers.25.mlp.fc2", |
| "transformer.vision.transformer.layers.26.attention.query_key_value", |
| "transformer.vision.transformer.layers.26.attention.dense", |
| "transformer.vision.transformer.layers.26.mlp.fc1", |
| "transformer.vision.transformer.layers.26.mlp.fc2", |
| "transformer.vision.transformer.layers.27.attention.query_key_value", |
| "transformer.vision.transformer.layers.27.attention.dense", |
| "transformer.vision.transformer.layers.27.mlp.fc1", |
| "transformer.vision.transformer.layers.27.mlp.fc2", |
| "transformer.vision.transformer.layers.28.attention.query_key_value", |
| "transformer.vision.transformer.layers.28.attention.dense", |
| "transformer.vision.transformer.layers.28.mlp.fc1", |
| "transformer.vision.transformer.layers.28.mlp.fc2", |
| "transformer.vision.transformer.layers.29.attention.query_key_value", |
| "transformer.vision.transformer.layers.29.attention.dense", |
| "transformer.vision.transformer.layers.29.mlp.fc1", |
| "transformer.vision.transformer.layers.29.mlp.fc2", |
| "transformer.vision.transformer.layers.30.attention.query_key_value", |
| "transformer.vision.transformer.layers.30.attention.dense", |
| "transformer.vision.transformer.layers.30.mlp.fc1", |
| "transformer.vision.transformer.layers.30.mlp.fc2", |
| "transformer.vision.transformer.layers.31.attention.query_key_value", |
| "transformer.vision.transformer.layers.31.attention.dense", |
| "transformer.vision.transformer.layers.31.mlp.fc1", |
| "transformer.vision.transformer.layers.31.mlp.fc2", |
| "transformer.vision.transformer.layers.32.attention.query_key_value", |
| "transformer.vision.transformer.layers.32.attention.dense", |
| "transformer.vision.transformer.layers.32.mlp.fc1", |
| "transformer.vision.transformer.layers.32.mlp.fc2", |
| "transformer.vision.transformer.layers.33.attention.query_key_value", |
| "transformer.vision.transformer.layers.33.attention.dense", |
| "transformer.vision.transformer.layers.33.mlp.fc1", |
| "transformer.vision.transformer.layers.33.mlp.fc2", |
| "transformer.vision.transformer.layers.34.attention.query_key_value", |
| "transformer.vision.transformer.layers.34.attention.dense", |
| "transformer.vision.transformer.layers.34.mlp.fc1", |
| "transformer.vision.transformer.layers.34.mlp.fc2", |
| "transformer.vision.transformer.layers.35.attention.query_key_value", |
| "transformer.vision.transformer.layers.35.attention.dense", |
| "transformer.vision.transformer.layers.35.mlp.fc1", |
| "transformer.vision.transformer.layers.35.mlp.fc2", |
| "transformer.vision.transformer.layers.36.attention.query_key_value", |
| "transformer.vision.transformer.layers.36.attention.dense", |
| "transformer.vision.transformer.layers.36.mlp.fc1", |
| "transformer.vision.transformer.layers.36.mlp.fc2", |
| "transformer.vision.transformer.layers.37.attention.query_key_value", |
| "transformer.vision.transformer.layers.37.attention.dense", |
| "transformer.vision.transformer.layers.37.mlp.fc1", |
| "transformer.vision.transformer.layers.37.mlp.fc2", |
| "transformer.vision.transformer.layers.38.attention.query_key_value", |
| "transformer.vision.transformer.layers.38.attention.dense", |
| "transformer.vision.transformer.layers.38.mlp.fc1", |
| "transformer.vision.transformer.layers.38.mlp.fc2", |
| "transformer.vision.transformer.layers.39.attention.query_key_value", |
| "transformer.vision.transformer.layers.39.attention.dense", |
| "transformer.vision.transformer.layers.39.mlp.fc1", |
| "transformer.vision.transformer.layers.39.mlp.fc2", |
| "transformer.vision.transformer.layers.40.attention.query_key_value", |
| "transformer.vision.transformer.layers.40.attention.dense", |
| "transformer.vision.transformer.layers.40.mlp.fc1", |
| "transformer.vision.transformer.layers.40.mlp.fc2", |
| "transformer.vision.transformer.layers.41.attention.query_key_value", |
| "transformer.vision.transformer.layers.41.attention.dense", |
| "transformer.vision.transformer.layers.41.mlp.fc1", |
| "transformer.vision.transformer.layers.41.mlp.fc2", |
| "transformer.vision.transformer.layers.42.attention.query_key_value", |
| "transformer.vision.transformer.layers.42.attention.dense", |
| "transformer.vision.transformer.layers.42.mlp.fc1", |
| "transformer.vision.transformer.layers.42.mlp.fc2", |
| "transformer.vision.transformer.layers.43.attention.query_key_value", |
| "transformer.vision.transformer.layers.43.attention.dense", |
| "transformer.vision.transformer.layers.43.mlp.fc1", |
| "transformer.vision.transformer.layers.43.mlp.fc2", |
| "transformer.vision.transformer.layers.44.attention.query_key_value", |
| "transformer.vision.transformer.layers.44.attention.dense", |
| "transformer.vision.transformer.layers.44.mlp.fc1", |
| "transformer.vision.transformer.layers.44.mlp.fc2", |
| "transformer.vision.transformer.layers.45.attention.query_key_value", |
| "transformer.vision.transformer.layers.45.attention.dense", |
| "transformer.vision.transformer.layers.45.mlp.fc1", |
| "transformer.vision.transformer.layers.45.mlp.fc2", |
| "transformer.vision.transformer.layers.46.attention.query_key_value", |
| "transformer.vision.transformer.layers.46.attention.dense", |
| "transformer.vision.transformer.layers.46.mlp.fc1", |
| "transformer.vision.transformer.layers.46.mlp.fc2", |
| "transformer.vision.transformer.layers.47.attention.query_key_value", |
| "transformer.vision.transformer.layers.47.attention.dense", |
| "transformer.vision.transformer.layers.47.mlp.fc1", |
| "transformer.vision.transformer.layers.47.mlp.fc2", |
| "transformer.vision.transformer.layers.48.attention.query_key_value", |
| "transformer.vision.transformer.layers.48.attention.dense", |
| "transformer.vision.transformer.layers.48.mlp.fc1", |
| "transformer.vision.transformer.layers.48.mlp.fc2", |
| "transformer.vision.transformer.layers.49.attention.query_key_value", |
| "transformer.vision.transformer.layers.49.attention.dense", |
| "transformer.vision.transformer.layers.49.mlp.fc1", |
| "transformer.vision.transformer.layers.49.mlp.fc2", |
| "transformer.vision.transformer.layers.50.attention.query_key_value", |
| "transformer.vision.transformer.layers.50.attention.dense", |
| "transformer.vision.transformer.layers.50.mlp.fc1", |
| "transformer.vision.transformer.layers.50.mlp.fc2", |
| "transformer.vision.transformer.layers.51.attention.query_key_value", |
| "transformer.vision.transformer.layers.51.attention.dense", |
| "transformer.vision.transformer.layers.51.mlp.fc1", |
| "transformer.vision.transformer.layers.51.mlp.fc2", |
| "transformer.vision.transformer.layers.52.attention.query_key_value", |
| "transformer.vision.transformer.layers.52.attention.dense", |
| "transformer.vision.transformer.layers.52.mlp.fc1", |
| "transformer.vision.transformer.layers.52.mlp.fc2", |
| "transformer.vision.transformer.layers.53.attention.query_key_value", |
| "transformer.vision.transformer.layers.53.attention.dense", |
| "transformer.vision.transformer.layers.53.mlp.fc1", |
| "transformer.vision.transformer.layers.53.mlp.fc2", |
| "transformer.vision.transformer.layers.54.attention.query_key_value", |
| "transformer.vision.transformer.layers.54.attention.dense", |
| "transformer.vision.transformer.layers.54.mlp.fc1", |
| "transformer.vision.transformer.layers.54.mlp.fc2", |
| "transformer.vision.transformer.layers.55.attention.query_key_value", |
| "transformer.vision.transformer.layers.55.attention.dense", |
| "transformer.vision.transformer.layers.55.mlp.fc1", |
| "transformer.vision.transformer.layers.55.mlp.fc2", |
| "transformer.vision.transformer.layers.56.attention.query_key_value", |
| "transformer.vision.transformer.layers.56.attention.dense", |
| "transformer.vision.transformer.layers.56.mlp.fc1", |
| "transformer.vision.transformer.layers.56.mlp.fc2", |
| "transformer.vision.transformer.layers.57.attention.query_key_value", |
| "transformer.vision.transformer.layers.57.attention.dense", |
| "transformer.vision.transformer.layers.57.mlp.fc1", |
| "transformer.vision.transformer.layers.57.mlp.fc2", |
| "transformer.vision.transformer.layers.58.attention.query_key_value", |
| "transformer.vision.transformer.layers.58.attention.dense", |
| "transformer.vision.transformer.layers.58.mlp.fc1", |
| "transformer.vision.transformer.layers.58.mlp.fc2", |
| "transformer.vision.transformer.layers.59.attention.query_key_value", |
| "transformer.vision.transformer.layers.59.attention.dense", |
| "transformer.vision.transformer.layers.59.mlp.fc1", |
| "transformer.vision.transformer.layers.59.mlp.fc2", |
| "transformer.vision.transformer.layers.60.attention.query_key_value", |
| "transformer.vision.transformer.layers.60.attention.dense", |
| "transformer.vision.transformer.layers.60.mlp.fc1", |
| "transformer.vision.transformer.layers.60.mlp.fc2", |
| "transformer.vision.transformer.layers.61.attention.query_key_value", |
| "transformer.vision.transformer.layers.61.attention.dense", |
| "transformer.vision.transformer.layers.61.mlp.fc1", |
| "transformer.vision.transformer.layers.61.mlp.fc2", |
| "transformer.vision.transformer.layers.62.attention.query_key_value", |
| "transformer.vision.transformer.layers.62.attention.dense", |
| "transformer.vision.transformer.layers.62.mlp.fc1", |
| "transformer.vision.transformer.layers.62.mlp.fc2", |
| "transformer.vision.linear_proj.linear_proj", |
| "transformer.vision.linear_proj.dense_h_to_4h", |
| "transformer.vision.linear_proj.gate_proj", |
| "transformer.vision.linear_proj.dense_4h_to_h" |
| ], |
| "kv_cache_scheme": null, |
| "quant_method": "compressed-tensors", |
| "quantization_status": "compressed" |
| }, |
| "rmsnorm": true, |
| "rope_ratio": 1, |
| "seq_length": 8192, |
| "tie_word_embeddings": false, |
| "torch_dtype": "bfloat16", |
| "transformers_version": "4.47.1", |
| "use_cache": true, |
| "vision_config": { |
| "dropout_prob": 0.0, |
| "hidden_act": "gelu", |
| "hidden_size": 1792, |
| "image_size": 1120, |
| "in_channels": 3, |
| "intermediate_size": 15360, |
| "layer_norm_eps": 1e-06, |
| "num_heads": 16, |
| "num_hidden_layers": 63, |
| "num_positions": 6401, |
| "patch_size": 14, |
| "scaling_factor": 8 |
| }, |
| "vocab_size": 151552 |
| } |