Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

summarization using fine-tuned flan-t5 model in TGI outputs "generated text" instead of "summary_text" and outputs are completely different #2889

Open
2 of 4 tasks
maiiabocharova opened this issue Jan 7, 2025 · 0 comments

Comments

@maiiabocharova
Copy link

maiiabocharova commented Jan 7, 2025

System Info

ghcr.io/huggingface/text-generation-inference:3.0.1

Information

  • Docker
  • The CLI directly

Tasks

  • An officially supported command
  • My own modifications

Reproduction

Equivalent curl command:

curl https://api.endpoints.huggingface.cloud/v2/endpoint/...	\
-X POST 	-d '{"compute":{"accelerator":"gpu","instanceSize":"x1","instanceType":"nvidia-t4","scaling":{"maxReplica":1,"measure":{"hardwareUsage":80},"minReplica":0,"scaleToZeroTimeout":15,"metric":"hardwareUsage"}},"model":{"env":{},"framework":"pytorch","image":{"tgi":{"disableCustomKernels":false,"healthRoute":"/health","url":"ghcr.io/huggingface/text-generation-inference:3.0.1"}},"repository":"Daxtra/t5-base-summarization","secrets":{},"task":"summarization"},"name":"t5-base-summarization-yfg","provider":{"region":"us-east-1","vendor":"aws"},"type":"protected"}' 	\
-H "Content-Type: application/json" 	\
-H "Authorization: Bearer XXXXX"

Expected behavior

I expect outputs to be same as when using "pipeline" and contain "summary_text" keyword. Outputs are fully rubbish - as though only decoder is used without encoder in TGI.

pipe = pipeline(
    "summarization",
    model="my-t5-base-summarization",
    tokenizer="my-t5-base-summarization",
    truncation=True, 
    max_length=1_024,
)
@maiiabocharova maiiabocharova changed the title summarization using fine-tuned T5 model in TGI outputs "generated text" instead of "summary_text" and are incorrect summarization using fine-tuned T5 model in TGI outputs "generated text" instead of "summary_text" and outputs are completely different Jan 7, 2025
@maiiabocharova maiiabocharova changed the title summarization using fine-tuned T5 model in TGI outputs "generated text" instead of "summary_text" and outputs are completely different summarization using fine-tuned flan-t5 model in TGI outputs "generated text" instead of "summary_text" and outputs are completely different Jan 7, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant