Fix input_ids always on cpu issue #270
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: Akshay Sonawane <[email protected]>
apsonawane
force-pushed
the
asonawane/fix
branch
from
8a56bfb to
c632e32
Compare
| @@ -198,7 +200,9 @@ def _evaluate_model( | |||
| expected = dataset[task_id]["canonical_solution"] | |||
|
|
|||
| # Generate completion | |||
| input_ids = tokenizer(prompt, return_tensors="pt").input_ids | |||
| input_ids = tokenizer(prompt, return_tensors="pt").input_ids.to( | |||
There was a problem hiding this comment.
Is this change going to break the OGA tools? I don't know if OGA's input_ids support the .to() method or not.
We should add a test that loads an OGA model and passes it to humaneval (opened issue #271).
There was a problem hiding this comment.
Oh I am not sure, need to test
There was a problem hiding this comment.
I just opened PR #272 which adds a test for OGA that invokes humaneval: https://github.com/onnx/turnkeyml/actions/runs/12839092461/workflow?pr=272
You can try that if you want to make sure OGA will still work.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently when running this command:
lemonade -i microsoft/Phi-3.5-mini-instruct huggingface-load --device cuda accuracy-mmlu --ntrain 5 --tests managementit fails since the device we are setting to cuda and the input_ids are on cpu so both mmlu and humaneval test fails with error:
Error during model generation: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu! (when checking argument for argument index in method wrapper_CUDA__index_select)This PR adds the fix which puts the tensors based on the device specified