| { |
| "step": 11100, |
| "tokens": 1091174400, |
| "parameters": { |
| "total": 998694784, |
| "active": 340154240, |
| "shared": 10883968, |
| "active_path": 329270272 |
| }, |
| "batch_size": 96, |
| "sequence_length": 1024, |
| "loss": 0.9765625, |
| "auxiliary_loss": 1.3515625, |
| "gradient_norm": 0.138671875, |
| "route": 2, |
| "lr_scale": 0.9770887670336095, |
| "tokens_per_second": 456579.6963227891, |
| "elapsed_seconds": 1844.6724838319988, |
| "average_patch_bytes": 8.908382419574082, |
| "patch_coalesced_rate": 0.07291666666666667, |
| "gate": 0.53125, |
| "peak_vram_gib": 24.91581106185913, |
| "documents": { |
| "HuggingFaceFW/fineweb-edu:sample-10BT": 86801, |
| "HuggingFaceTB/finemath:finemath-4plus": 30911, |
| "HuggingFaceTB/smollm-corpus:cosmopedia-v2": 18665, |
| "open-r1/OpenR1-Math-220k:default": 38052, |
| "open-r1/codeforces-cots:solutions_w_editorials_py_decontaminated": 538 |
| }, |
| "source_bytes": { |
| "HuggingFaceFW/fineweb-edu:sample-10BT": 405420184, |
| "HuggingFaceTB/finemath:finemath-4plus": 153781212, |
| "HuggingFaceTB/smollm-corpus:cosmopedia-v2": 69899623, |
| "open-r1/OpenR1-Math-220k:default": 48932159, |
| "open-r1/codeforces-cots:solutions_w_editorials_py_decontaminated": 21002254 |
| } |
| } |
