Any plan to release the post-training recipe?
#3
by
Yi30
- opened
This comment has been hidden (marked as Resolved)
你理解偏差啦，人家想表达的是支持MXFP4、MXFP6、MXFP8的量化...
Yi30
changed discussion title from
Was the model trained with UE8M0 FP8 scales but released with FP32 scales?
to Any plan to release the post-training recipe?