| { | |
| "step": 1400, | |
| "stage": "sft", | |
| "val_loss": 0.5056432619690895, | |
| "version": 3, | |
| "params": 40844416, | |
| "dtype": "bf16", | |
| "config": { | |
| "vocab_size": 8192, | |
| "block_size": 1024, | |
| "n_layer": 13, | |
| "n_head": 8, | |
| "n_embd": 512, | |
| "dropout": 0.0, | |
| "n_kv_head": 2, | |
| "qk_norm": true, | |
| "arch_version": 3 | |
| } | |
| } |