| { | |
| "architectures": [ | |
| "YoutuForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 128000, | |
| "dtype": "bfloat16", | |
| "embedding_initializer_range": null, | |
| "eos_token_id": 128001, | |
| "hidden_act": "silu", | |
| "hidden_size": 2048, | |
| "initializer_range": null, | |
| "intermediate_size": 6144, | |
| "kv_lora_rank": 512, | |
| "max_position_embeddings": 131072, | |
| "mlp_bias": false, | |
| "model_type": "youtu", | |
| "num_attention_heads": 16, | |
| "num_hidden_layers": 32, | |
| "num_key_value_heads": 16, | |
| "pad_token_id": null, | |
| "q_lora_rank": 1536, | |
| "qk_head_dim": 192, | |
| "qk_nope_head_dim": 128, | |
| "qk_rope_head_dim": 64, | |
| "rms_norm_eps": 1e-06, | |
| "rope_interleave": true, | |
| "rope_parameters": { | |
| "rope_theta": 1600000, | |
| "rope_type": "default" | |
| }, | |
| "tie_word_embeddings": true, | |
| "transformers_version": "5.0.0.dev0", | |
| "use_cache": true, | |
| "v_head_dim": 128, | |
| "vocab_size": 128256 | |
| } | |