verigle
verigle
I hava similar problem， the GPU memory always increase， and then Out of Memory. ``` Evaluation [ 0/5000] eta: 1:57:37 time: 1.4116 data: 0.3283 max mem: 9518 Evaluation [ 10/5000]...
can you give some examples for how to modified code to finetuning a fraction of layers.
I just want to replace the domain of url "huggingface.co" to "hf-mirror.com" for regiested model "[TabbyML/DeepseekCoder-6.7B](https://huggingface.co/deepseek-ai/deepseek-coder-6.7b-base)" https://tabby.tabbyml.com/docs/models/
thank you for your reply, can i download it manually then put it on a directory？
is there any plan to release finetune code of pytorch ?
is there any plan to release samples code of covariates support and finetune support?
Watchdog timeout (self.watchdog_timeout=300) +1
I try it with gpustack with qwen2.5 7b, the same thing happens.