| { | |
| "step": 2400, | |
| "stage": "sft", | |
| "val_loss": 0.9569745600223541, | |
| "version": 2, | |
| "params": 7738112, | |
| "dtype": "bf16", | |
| "config": { | |
| "vocab_size": 8192, | |
| "block_size": 512, | |
| "n_layer": 8, | |
| "n_head": 8, | |
| "n_embd": 256, | |
| "dropout": 0.0, | |
| "n_kv_head": 2, | |
| "qk_norm": true, | |
| "arch_version": 2 | |
| } | |
| } |