openbmb

/

MiniCPM-Llama3-V-2_5

like 1.26k

Test post

https://github.com/OpenBMB/MiniCPM-V/tree/main/finetune
@deqiuqiuzhang Try this

Can i fine tune using qlora?

Hi,
Could you let me know if everyone has successfully fine-tuned the model? Additionally, I have a question about GPU requirements: is 31.2GB needed per GPU, or is it split between two GPUs? Also, I noticed that Kaggle offers 2 T4 GPUs—are these sufficient for fine-tuning my model with a custom dataset?

Hi,
Could you let me know if everyone has successfully fine-tuned the model? Additionally, I have a question about GPU requirements: is 31.2GB needed per GPU, or is it split between two GPUs? Also, I noticed that Kaggle offers 2 T4 GPUs—are these sufficient for fine-tuning my model with a custom dataset?

31.2GB per GPU was tested with two A100 GPU, as far as I know, you can use zero3 + offload to minimize the memory usage. And according to the deepspeed zero strategy, the more GPUs you have, the lower memory usage of each GPU. The final memory usage is also related to the max input length and the image resolution, if you have two T4 GPUs, you can try it by setting a suitable length and zero3 config.

Has the issue on cuda assertions solved?