Zishun Yu

icon indicating a website link www.zishun.me

icon location Chicago icon location CS PhD Student@UIC • Ex-TikTok • Kaggle Master • HUST Alumni

Results 3 comments of Zishun Yu

Error while training using notebook

Changing the 40th line in **base_layers.py** file in the **torchparse** package works for me. `- return torch.cat([channel, spatial])` `+ return torch.cat([channel, spatial.long()])`

Reproduce MOPO results

I agree with @IantheChan. To my knowledge, some model-based RL works uses this trick, which seems to be very critical for model-based RL. I find this makes sense for robot...

Problems in reproducing the RL fine-tuned results

Hi, folks, I also get pass@1 approximately 1% but pass@5 at 2.4% with CE loss fine-tuned model. After trying a bunch of temperature, 0.2 seems get me the best pass@1...