| license: apache-2.0 | |
| library_name: pytorch | |
| tags: | |
| - code | |
| - selection-mixing | |
| # selection-coder | |
| ~236M decoder, **only SelectionMixing (order-statistic routing), no attention**. Muon, bf16, trained on MBPP+Alpaca+Evol-Instruct. step=6000. | |
| license: apache-2.0 | |
| library_name: pytorch | |
| tags: | |
| - code | |
| - selection-mixing | |
| # selection-coder | |
| ~236M decoder, **only SelectionMixing (order-statistic routing), no attention**. Muon, bf16, trained on MBPP+Alpaca+Evol-Instruct. step=6000. | |