Two local torch training runs for attestation and dataset-filter checks.
tape_validation/torch_20260906_171316/next_token_mse vs. estimated ISO FLOPs
2 matching runs · 6ND; missing tokens estimated at 20× parameters
All 2 filtered runs
| Run | Scale (B) | ISO FLOPs | Training data | tape_validation/torch_20260906_171316/next_token_mse |
|---|---|---|---|---|
| tape_validation/torch_20260906_171316_baking_window8 No description yet | <0.001 | ~3.2e4 | 0.02 | |
| tape_validation/torch_20260906_171316_astronomy_window8 No description yet | <0.001 | ~1.4e4 | 0.03 |