yeppp27 comments

Repositories
Issues
Comments

Results 4 comments of


                                            yeppp27

Nice work! Does the mantis has image seperator when sending to LLM?

Hi! Thanks for your great work~ I what to ask where is the interleaved text-image processing with the image seperator? Since I didn't find it in the data.py

Script about qlora.

Hi! When I use deepspeed zero3 with Qlora for sft, I get an error like "[rank2]: NotImplementedError: Cannot copy out of meta tensor; no data!"

Thanks! But I got Cuda out of memory for Qlora fine-tuning with zero2 on 72B model. My zero2 config is { "fp16": { "enabled": "auto", "loss_scale": 0, "loss_scale_window": 1000, "initial_scale_power":...