Visual Question Answering
Transformers
TensorBoard
Safetensors
internvl_chat
feature-extraction
custom_code

multi-gpu support?

#9
by matatonic - opened

I'm seeing errors like this when using device_map=auto

File "modules/transformers_modules/OpenGVLab/InternVL-Chat-V1-5-Int8/0d0a0d47a718aaef231cb00c6cf9ce3fb905269b/modeling_internvl_chat.py", line 348, in generate
   input_embeds[selected] = vit_embeds.reshape(-1, C)
   ~~~~~~~~~~~~^^^^^^^^^^
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cuda:1!
  Gracefully stopping... (press Ctrl+C again to force)

Is this known? Is there a fix?

I have the same problem..

This comment has been hidden

็›ธๅŒ็š„้—ฎ้ข˜

Sign up or log in to comment