| { |
| "step": 20, |
| "tokens": 1966080, |
| "parameters": { |
| "total": 998694784, |
| "active": 340154240, |
| "shared": 10883968, |
| "active_path": 329270272 |
| }, |
| "batch_size": 96, |
| "sequence_length": 1024, |
| "loss": 4.3125, |
| "auxiliary_loss": 4.9375, |
| "gradient_norm": 11.9375, |
| "route": 1, |
| "lr_scale": 0.02, |
| "tokens_per_second": 321014.64369120006, |
| "elapsed_seconds": 7.8329121970000415, |
| "average_patch_bytes": 8.42076409114271, |
| "patch_overflow_rate": 0.28125, |
| "gate": 0.50390625, |
| "peak_vram_gib": 31.73339605331421, |
| "documents": { |
| "HuggingFaceFW/fineweb-edu:sample-10BT": 694, |
| "HuggingFaceTB/finemath:finemath-4plus": 666, |
| "HuggingFaceTB/smollm-corpus:cosmopedia-v2": 526, |
| "open-r1/OpenR1-Math-220k:default": 32, |
| "open-r1/codeforces-cots:solutions_w_editorials_py_decontaminated": 0 |
| }, |
| "source_bytes": { |
| "HuggingFaceFW/fineweb-edu:sample-10BT": 3301362, |
| "HuggingFaceTB/finemath:finemath-4plus": 3189383, |
| "HuggingFaceTB/smollm-corpus:cosmopedia-v2": 1928354, |
| "open-r1/OpenR1-Math-220k:default": 43657, |
| "open-r1/codeforces-cots:solutions_w_editorials_py_decontaminated": 0 |
| } |
| } |
