always getting (error_code: 1)

OpenGVLab / Multi-Modality-Arena

Chatbot Arena meets multi-modality! Multi-Modality Arena allows you to benchmark vision-language models side-by-side while providing images as inputs. Supports MiniGPT-4, LLaMA-Adapter V2, LLaVA, BLIP-2, and many more!

419 stars 30 forks source link

always getting (error_code: 1) #7

Open emanuelevivoli opened 1 year ago

emanuelevivoli commented 1 year ago

Hello and thank you for your amazing work!

However, I have a problem: the models are loaded well but I continue getting

NETWORK ERROR DUE TO HIGH TRAFFIC. PLEASE REGENERATE OR REFRESH THIS PAGE. (error_code: 1)

The models (blip2 and minigpt4) are both on GPUs (I'm using them with --keep-in-device otherwise they were not even loaded) but every try I do, I always get the same error.

Any advice?

BellXP commented 11 months ago

Apologies for the delayed response. Due to GPU resource limitations, we initially employed the "keep-in-device" option to determine whether to transfer the model between computer memory and GPU memory during inference generation. This process incurred significant time overhead as the model needed to be moved back and forth. However, in our latest code update, we have optimized this by maintaining a constant placement of the model in GPU memory.

Regarding the occurrence of "Network error," this issue could be attributed to unstable connections between the web server, controller, and model worker.

iraj465 commented 4 months ago

hey i'm still getting this error, what could be the issue here? I have stable connection between controller and the model workers