Author | SHA1 Message | Date |
---|---|---|
|
9def4fbafd Remove micro_batch_training parameter and replace with gradient_accumulation_steps | 2 years ago |
|
3faf005226 fix a bug in the config for use_fast_kernels | 2 years ago |
|
feaa344af3 resolve conflicts | 2 years ago |
|
44ef280d31 adding flash attention and xformer memory efficient through PT SDPA | 2 years ago |
|
895dfcea30 add nightly check for using low_cpu_fsdp mode | 2 years ago |
|
4767f09ecd Initial commit | 2 years ago |