| { | |
| "encoder": "answerdotai/ModernBERT-large", | |
| "head_layers": 2, | |
| "max_len": 1024, | |
| "head_max_len": 384, | |
| "amp_dtype": "bf16", | |
| "model_name": "decider", | |
| "temperature": [ | |
| 2.378414230005442, | |
| 0.8908987181403393, | |
| 0.9170040432046712 | |
| ], | |
| "training": { | |
| "updates": 7313, | |
| "epochs_completed": 1, | |
| "hours": 1.96, | |
| "world_size": 1, | |
| "fine_tuned_from_checkpoint": true | |
| }, | |
| "fine_tuned_from": "/nobackup/proj/disk/naiss2025-22-767/personal/caldasr/rag/hf_cache/hub/models--convaiinnovations--laya/snapshots/7b928d828b7b0e022f929d9bd2e44165aa270148", | |
| "temperature_by_options": { | |
| "choice:11+": 1.155352696872273, | |
| "choice:2": 3.267830906482199, | |
| "choice:3-5": 2.058604473286984, | |
| "choice:6-10": 1.5422108254079405, | |
| "noul:2": 0.9170040432046712, | |
| "score:3-5": 0.9438743126816933, | |
| "score:6-10": 0.7711054127039704 | |
| }, | |
| "temperature_fit": { | |
| "examples": 46583, | |
| "bucket_sizes": { | |
| "choice:11+": 806, | |
| "choice:2": 20281, | |
| "choice:3-5": 22172, | |
| "choice:6-10": 1099, | |
| "noul:2": 1404, | |
| "score:3-5": 565, | |
| "score:6-10": 256 | |
| }, | |
| "per_source_cap": 3000, | |
| "excluded_sources": [ | |
| "predev" | |
| ], | |
| "nll_T1": 0.5914, | |
| "nll_fitted": 0.409, | |
| "ece_T1": 0.0767, | |
| "ece_fitted": 0.0169 | |
| } | |
| } |