| { | |
| "model_type": "maurice", | |
| "architectures": [ | |
| "Maurice" | |
| ], | |
| "auto_map": { | |
| "AutoModel": "modeling_maurice.Maurice" | |
| }, | |
| "vocab_size": 17785, | |
| "d_model": 1280, | |
| "n_heads": 20, | |
| "n_kv_heads": 4, | |
| "n_layers": 16, | |
| "context_length": 4096, | |
| "rope_theta": 10000.0, | |
| "rms_norm_eps": 1e-06, | |
| "tie_word_embeddings": true, | |
| "pad_id": 2, | |
| "bos_id": 0, | |
| "eos_id": 1, | |
| "instr_prefix_id": 6, | |
| "torch_dtype": "float32", | |
| "train_step": 37000, | |
| "val_loss": 1.0853 | |
| } |