| { | |
| "architectures": [ | |
| "RMSN" | |
| ], | |
| "model_type": "rmsn", | |
| "vocab_size": 49152, | |
| "dim": 960, | |
| "depth": 16, | |
| "slots": 64, | |
| "seq_len": 512, | |
| "num_parameters": 220655088, | |
| "tokenizer_class": "PreTrainedTokenizerFast", | |
| "tokenizer_name": "HuggingFaceTB/cosmo2-tokenizer", | |
| "val_ppl": 72.7, | |
| "val_lyap": -0.0015 | |
| } |