| { | |
| "model_type": "campgpt", | |
| "architectures": [ | |
| "CampGPT" | |
| ], | |
| "vocab_size": 50304, | |
| "n_embd": 768, | |
| "n_head": 12, | |
| "n_kv_head": 4, | |
| "n_layer": 12, | |
| "block_size": 1024, | |
| "norm_eps": 1e-06, | |
| "multiple_of": 64, | |
| "use_moe": true, | |
| "n_experts": 3, | |
| "n_experts_per_tok": 1, | |
| "n_shared_experts": 1, | |
| "total_params": 322680576, | |
| "training_stages": [ | |
| "pretrain_10B", | |
| "sft", | |
| "dpo" | |
| ], | |
| "val_loss": 0.0698780653377374 | |
| } |