Skip to content

multiple GPU BUG when scoring logit #44

Description

@Irisicy4

Given:

    --gpu_per_parallel 2 \
    --parallel_per_task 1 \
    --circular_eval False \
    --score_target

BUG:

Shard 0 (4329 samples): 0%| | 0/4329 [00:00<?, ?it/s]You shouldn't move a model that is dispatched using accelerate hooks.
[Shard 0] ❌ Error: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cuda:1! (when checking argument for argument mat2 in method wrapper_CUDA_bmm)
[Shard 0] ⚠️ Max retries reached, skipping sample
Shard 0 (4329 samples): 0%| | 1/4329 [00:04<4:53:38, 4.07s/it]You shouldn't move a model that is dispatched using accelerate hooks.
[Shard 0] ❌ Error: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cuda:1! (when checking argument for argument mat2 in method wrapper_CUDA_bmm)

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Fields

    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions