OpenNMT / OpenNMT/CTranslate2

Anomalous T5 results using GPU inference on a 4090 graphics card

Open
#1,679 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
C++
Stars
4.7k
Forks
536
Avg merge
12h 12m
Merged PRs (30d)
4

Description

Thank you very much for your work. I'm using ctranslate2 accelerated https://huggingface.co/Maciel/T5Corrector-base-v2 reasoning and when using the cpu for inference the output is normal, but switching to using the GPU, the output is all: Response Text: {" translated_text":"..."} , where is the problem please?

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reproducing inference with the Hugging Face Maciel/T5Corrector-base-v2 model on the CPU and on a 4090 GPU, using the CTranslate2 paths described in the issue. Capture the relevant environment and conversion settings, then compare the outputs; done means GPU inference no longer produces repeated tokens and matches the normal CPU result.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp, machine-learning
Domain
machine-learning, performance
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.