| { | |
| "architectures": [ | |
| "YoutuForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "auto_map": { | |
| "AutoConfig": "configuration_youtu.YoutuConfig", | |
| "AutoModel": "modeling_youtu.YoutuModel", | |
| "AutoModelForCausalLM": "modeling_youtu.YoutuForCausalLM" | |
| }, | |
| "bos_token_id": 128000, | |
| "dtype": "bfloat16", | |
| "embedding_initializer_range": 0.02795084971874737, | |
| "eos_token_id": 128001, | |
| "head_dim": 64, | |
| "hidden_act": "silu", | |
| "hidden_size": 2048, | |
| "initializer_range": 0.013975424859373685, | |
| "intermediate_size": 6144, | |
| "kv_lora_rank": 512, | |
| "max_position_embeddings": 131072, | |
| "mlp_bias": false, | |
| "model_type": "youtu_llm", | |
| "num_attention_heads": 16, | |
| "num_hidden_layers": 32, | |
| "num_key_value_heads": 16, | |
| "q_lora_rank": 1536, | |
| "qk_head_dim": 192, | |
| "qk_nope_head_dim": 128, | |
| "qk_rope_head_dim": 64, | |
| "quantization_config": { | |
| "autoround_version": "0.9.5", | |
| "bits": 4, | |
| "data_type": "int", | |
| "group_size": 128, | |
| "packing_format": "auto_round:auto_gptq", | |
| "quant_method": "auto-round", | |
| "sym": true | |
| }, | |
| "rms_norm_eps": 1e-06, | |
| "rope_interleave": true, | |
| "rope_scaling": null, | |
| "rope_theta": 1600000, | |
| "transformers_version": "4.57.1", | |
| "use_cache": true, | |
| "v_head_dim": 128, | |
| "vocab_size": 128256 | |
| } | |