You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:2 and cuda:0! (when checking argument for argument index in method wrapper_CUDA__index_select)
Running loglikelihood requests: 0%|
wenhuach21
changed the title
quantized qwen2.5-0.5B with 3 cards throw error at evaluation while 2 cards is ok.
quantizing qwen2.5-0.5B with 3 cards throw error at evaluation while 2 cards is ok.
Jan 15, 2025
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:2 and cuda:0! (when checking argument for argument index in method wrapper_CUDA__index_select)
Running loglikelihood requests: 0%|
CUDA_VISIBLE_DEVICES=1,2,3 python3 -m auto_round --model "/models/Qwen2.5-0.5B-Instruct" --format "auto_awq" --device "0,1,2" --tasks piqa --iters 1
The text was updated successfully, but these errors were encountered: