aitingliu comments

Repositories
Issues
Comments

Results 4 comments of


                                            aitingliu

[BUG]running step3 use bloomz + lora + zero3, raise RuntimeError(f"{param.ds_summary()} already in registry")

@cmikeh2 @jeffra @lekurile @awan-10

[BUG]running step3 use bloomz + lora + zero3, raise RuntimeError(f"{param.ds_summary()} already in registry")

> Hello @liuaiting. Thank you for reporting this issue to us. One of our recent fixes #3462 may have already fixed this error. Could you update your deepspeed and give...

DeepSpeed-Chat: prefetch of layers during reward model forward pass leads to error during sample generation

`raise RuntimeError(f"still have inflight params "f"{[p.ds_summary for p in self.__inflight_param_registry.keys()]}")` this error is still reported when I running step3 use bloomz + zero3. I encountered this bug after the previous...

gpt ppo training error

same issue!!