| { | |
| "_name_or_path": "/home/huyiwen/miniyulan-ckpts/miniyulan-2B-s25d-decay80-1sqrt-long-28k-qkln-final-stage2-resume-27_20241213_051741/YuLan-Mini-2.4B-bfloat16", | |
| "architectures": [ | |
| "LlamaForCausalLM" | |
| ], | |
| "attention_bias": true, | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 1, | |
| "eos_token_id": 2, | |
| "head_dim": 64, | |
| "hidden_act": "silu", | |
| "hidden_size": 1920, | |
| "initializer_range": 5e-05, | |
| "intermediate_size": 4800, | |
| "max_position_embeddings": 28723, | |
| "mlp_bias": false, | |
| "model_type": "llama", | |
| "num_attention_heads": 30, | |
| "num_hidden_layers": 56, | |
| "num_key_value_heads": 6, | |
| "pad_token_id": 102, | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-06, | |
| "rope_scaling": null, | |
| "rope_theta": 490000.0, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.47.1", | |
| "use_cache": true, | |
| "vocab_size": 99000 | |
| } | |