accelerate
8159c98d - Models With Tied Weights Need Re-Tieing After FSDP Param Init (#3154)

Commit
1 year ago
Models With Tied Weights Need Re-Tieing After FSDP Param Init (#3154) * add fsdp_tool to retie after param init * make it handle generic param_init_fn * fix quality Signed-off-by: Yu Chin Fabian Lim <flim@sg.ibm.com> --------- Signed-off-by: Yu Chin Fabian Lim <flim@sg.ibm.com>
Author
Parents
Loading