| { | |
| "kind": "bpe2048-factor", | |
| "recipe": "1L tanh-born factored one-hot BPE", | |
| "hid": 2048, | |
| "k": 32, | |
| "seq": 128, | |
| "bs": 32, | |
| "lr": 0.002, | |
| "seed": 7, | |
| "qr_seed": 11, | |
| "v": 2048, | |
| "vocab": 2048, | |
| "train_tokens": 1238546057, | |
| "val_tokens": 371069, | |
| "train_chars": 2449160526, | |
| "val_chars": 1142150, | |
| "device": "cpu", | |
| "n_born_theory": 4462656, | |
| "total": 300000, | |
| "corpora": [ | |
| "D:/z-system/_wikitext2", | |
| "D:/z-system/_wikitext103", | |
| "D:/z-system/_tinystories" | |
| ], | |
| "wiki_repeat": 3, | |
| "wiki_end": 181187803, | |
| "train_tokens_raw": 876170451, | |
| "step": 300000, | |
| "n_born": 4462656, | |
| "wall_s": 128507.80147910118, | |
| "last_loss": 3.0803170204162598, | |
| "nll_train": 3.0575421333312987, | |
| "nll_val": 3.9735021629486695, | |
| "ppl_val": 53.17041661157406, | |
| "nll_stories": 2.1479208919417907, | |
| "ppl_stories": 8.567028090895379 | |
| } |