| { | |
| "architectures": [ | |
| "LlamaForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "auto_map": { | |
| "AutoConfig": "configuration_llama.LlamaConfig", | |
| "AutoModel": "modeling_llama.LlamaModel", | |
| "AutoModelForCausalLM": "modeling_llama.LlamaForCausalLM" | |
| }, | |
| "bos_token_id": 128000, | |
| "eos_token_id": 128001, | |
| "head_dim": 64, | |
| "hidden_act": "silu", | |
| "hidden_size": 1280, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 3584, | |
| "max_position_embeddings": 2048, | |
| "mlp_bias": false, | |
| "num_attention_heads": 20, | |
| "num_hidden_layers": 54, | |
| "num_query_heads": 5, | |
| "query_dim": 64, | |
| "num_key_heads": 1, | |
| "key_dim": 64, | |
| "num_value_heads": 1, | |
| "value_dim": 128, | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-05, | |
| "tie_word_embeddings": true, | |
| "torch_dtype": "bfloat16", | |
| "use_cache": true, | |
| "vocab_size": 128256 | |
| } | |