22dad1df73
The start route's precision preflight folded bf16/int8/fp8 into the CUDA requirement but omitted mxfp8, so an mxfp8 request on a GPU-less host (or an older CUDA GPU without Blackwell) passed the preflight, evicted resident image and chat models, then raised only in the spawned trainer child. Mirror _resolve_base_precision: require CUDA for mxfp8 and re-check the Blackwell (sm100+) capability up front, so a doomed run is rejected before teardown. |
||
|---|---|---|
| .. | ||
| data_recipe | ||
| export | ||
| inference | ||
| rag | ||
| training | ||
| __init__.py | ||
| _torchao_stub.py | ||
| import_guards.py | ||
| tool_healing.py | ||