| Commit message (Collapse) | Author | Age | Files | Lines | |
|---|---|---|---|---|---|
| * | mix.md: ClimbMix was not decontaminatedHEADmain | jdj | 13 hours | 1 | -2/+6 |
| | | | | | | | | The decontamination section said every source passes the 13-gram filter. ClimbMix was tokenized in bulk shards outside prepare.py and was not filtered, as the model card already states. Add the per-source dropped counts for the filtered sources. | ||||
| * | Add mix.md: the complete data mixture | jdj | 3 days | 2 | -0/+75 |
| | | | | | | Every source with its dataset, subsets, filters and license; the draw-weight schedule of every phase; batch, context and final stages. | ||||
| * | Boris-2 training code | jdj | 3 days | 50 | -0/+11892 |
| The code, configuration and data filters used to train Boris-2 (144M, 200B tokens, single V100): model, training loop, data preparation, n-gram table, V100 kernels, final stages and the exact run configs. | |||||
