| { | |
| "model_class": "llama2", | |
| "model_size": "7b", | |
| "architectures": [ | |
| "Llama2ForCausalLM" | |
| ], | |
| "dim": 4096, | |
| "n_layers": 32, | |
| "n_heads": 32, | |
| "n_kv_heads": null, | |
| "vocab_size": 32000, | |
| "multiple_of": 256, | |
| "ffn_dim_multiplier": null, | |
| "norm_eps": 1e-5, | |
| "max_batch_size": 32, | |
| "max_seq_len": 2048, | |
| "bos_token_id":1, | |
| "eos_token_id":2, | |
| "pad_token_id":-1, | |
| "torch_dtype": "float16", | |
| "pretraining_base": "llama2-7b-2t-tokens", | |
| "model_repo_folder_path": "llama2-7b-base" | |
| } | |