Pipeline instantiation of model "facebook/nllb-200-distilled-600M" requires source and target language as mandatory

[drunkeninja42]

System Info

• transformers version: 4.35.2
• Platform: Linux-5.15.120+-x86_64-with-glibc2.35
• Python version: 3.10.12
• Huggingface_hub version: 0.19.4
• Safetensors version: 0.4.0
• Accelerate version: not installed
• Accelerate config: not found
• PyTorch version (GPU?): 2.1.0+cu118 (False)
• Tensorflow version (GPU?): 2.14.0 (False)
• Flax version (CPU?/GPU?/TPU?): 0.7.5 (cpu)
• Jax version: 0.4.20
• JaxLib version: 0.4.20
• Using GPU in script?: Yes
• Using distributed or parallel set-up in script?: No

Who can help?

@stevhliu @MKhalusova

Information

• The official example scripts
• My own modified scripts

Tasks

• An officially supported task in the examples folder (such as GLUE/SQuAD, ...)
• My own task or dataset (give details below)

Reproduction

Steps to Reproduce:

1. Go to https://huggingface.co/facebook/nllb-200-distilled-600M
2. Click Use in Transformers Tag
3. Copy the # Use a pipeline as a high-level helper code snippet
4. Run in your notebook
5. Error is displayed asking for src_lang and tgt_lang as mandatory kwags.
   

Expected behavior

I expect the code snippet in the Use in Transformers tag to mention these kwags in the example as they are mandatory to be used in this model and while the translation tasks.

I will love to contribute to this issue.
Thanks!

[stevhliu]

Hi, for translation pipelines you can do translation_xx_to_yy, for example:

from transformers import pipeline

pipe = pipeline("translation_en_to_fr", model="facebook/nllb-200-distilled-600M")
pipe("Let's go to france and see the eiffel tower")

Not every translation model requires this, but I think it'd be more inclusive to add translation_xx_to_yy for models that require a src_lang and tgt_lang. WDYT @ArthurZucker?

[ArthurZucker]

MMmm yes and not, the snippet is automatically generated and cannot account for every use-case. You are right @stevhliu, see the doc is very much lacking 😅

[drunkeninja42]

Hey @ArthurZucker @stevhliu !
If there are some improvements or work required in this issue, I would love to work and contribute to🤗

[github-actions]

This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.

Please note that issues that do not follow the contributing guidelines are likely to be ignored.

[ArthurZucker]

Hey @drunkeninja42 feel free to open a PR that increase the specific doc in this case if yo feel like it!

[drunkeninja42]

Sure @ArthurZucker , will raise the PR for this today, Thanks !

[st81]

@ArthurZucker
I'm eager to address this issue and I believe it might be beneficial to implement a code snippet similar to the one provided below.

# Use a pipeline as a high-level helper
from transformers import pipeline

# Change `xx` to the language of the input and `yy` to the language of the desired output.
pipe = pipeline("translation_xx_to_yy", model="facebook/nllb-200-distilled-600M")

However, I'm unsure about the exact location in the source code where I should make modifications to update the code snippet. Could you please provide guidance on this matter? Thank you!

[ArthurZucker]

This should go in the nllb.md IMO!