Add Llama2 resources by wonhyeongseo · Pull Request #25531 · huggingface/transformers

wonhyeongseo · 2023-08-16T07:21:00Z

What does this PR do?

Adds resources of Llama2 according to this issue.
This PR serves as an example to our OSSCA mentees who will contribute more models.

Part of #20055

Before submitting

This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
Did you read the contributor guideline,
Pull Request section?
Was this discussed/approved via a Github issue or the forum? Please add a link
to it if that's the case.
Did you make sure to update the documentation with your changes? Here are the
documentation guidelines, and
here are tips on formatting docstrings.
Did you write any new necessary tests?

Who can review?

@stevhliu, may you please review this PR?

Co-authored-by: Woojun Jung <hello_984@naver.com>

HuggingFaceDocBuilderDev · 2023-08-16T07:52:16Z

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint.

wonhyeongseo · 2023-08-16T08:21:29Z

Hello everyone,

We've been working on organizing resources related to LLaMA2, and we've noticed that many of the resources intersect category-wise. We're trying to ensure that the documentation is intuitive and helpful for users. Could anyone provide suggestions or best practices on how to best arrange these resources?

Additionally, after conducting a full-text search on the repository, we couldn't locate any notebooks specifically related to LLaMA2. If anyone is aware of such notebooks or has worked on one, could you kindly point us in the right direction?

Thank you in advance for your assistance!

Best regards,
@wonhyeongseo and @jungnerd

ArthurZucker · 2023-08-16T08:27:14Z

There are quite a few ressources that are indeed not properly linked.
Here is one notebook: https://colab.research.google.com/drive/1PEQyJO1-f6j0S_XJ8DV50NkpzasXkrzd?usp=sharing%E2%80%A6 that has been shared + updated by a contributor!

* style: one liners for each resource Co-Authored-By: Woojun Jung <46880056+jungnerd@users.noreply.github.com> Co-Authored-By: Kihoon Son <75935546+kihoon71@users.noreply.github.com>

wonhyeongseo · 2023-08-16T23:20:52Z

Thank you for the pointer @ArthurZucker! With your guidance, we found googling site:https://colab.research.google.com transformers llama2 to be quite effective and useful.

We have a couple of questions:

Would it be okay to add Llama2 to the text-generation script and update the llm tutorial example to Llama2? (I feel that since llama2 is still gated, we cannot at this time)
How would you like us to treat spaces like this one for the resource section?
Would it be ok to add LLaMA resources on top of Llama2 as they are the same family like GPT? (cc. @stevhliu )

Nice, thank you for adding those. Now you just have a few more to go! 😁

Take a look at the OpenAI GPT2 resources page and feel free to add over whatever is missing here since the usage for OpenAI GPT is practically the same.

We will soon cover more models that are "mainstream" in Korea, translate the docs and contribute localized blogs and notebooks. Sorry for being so late on schedule.

Thank you so much for your friendly and honest guidance, hope we can have a call sometime.

Best regards,
Won

wonhyeongseo · 2023-08-21T01:40:44Z

May you please review this PR, @stevhliu ? Thanks a ton for your help!

stevhliu

Great job friends!

How would you like us to treat spaces like this one for the resource section?

This is a community model card so we shouldn't include it.

Would it be ok to add LLaMA resources on top of Llama2 as they are the #20084 (review)?

Yes absolutely! I would just make a note that the resource is for LLaMA, but it can also be applied to LLaMA2.

stevhliu · 2023-08-21T16:08:12Z

+- A notebook on how to [run the Llama 2 Chat Model with 4-bit quantization on a local computer or Google Colab](https://colab.research.google.com/drive/1X1z9Q6domMKl2CnEM0QGHNwidLfR4dW2?usp=sharing). 🌎
+
+🚀 Deploy
+- A blog on how to [use SageMaker with Llama2 and QLora for efficient model deployment](https://www.philschmid.de/sagemaker-llama2-qlora) by Phil Schmid.


Suggested change

- A blog on how to [use SageMaker with Llama2 and QLora for efficient model deployment](https://www.philschmid.de/sagemaker-llama2-qlora) by Phil Schmid.

- A blog on how to [use SageMaker with Llama2 and QLora for efficient model deployment](https://www.philschmid.de/sagemaker-llama2-qlora).

stevhliu · 2023-08-21T16:08:20Z

+
+🚀 Deploy
+- A blog on how to [use SageMaker with Llama2 and QLora for efficient model deployment](https://www.philschmid.de/sagemaker-llama2-qlora) by Phil Schmid.
+- A blog on how to [deploy Llama models using SageMaker for scalable applications](https://www.philschmid.de/sagemaker-llama-llm) by Phil Schmid.


Suggested change

- A blog on how to [deploy Llama models using SageMaker for scalable applications](https://www.philschmid.de/sagemaker-llama-llm) by Phil Schmid.

- A blog on how to [deploy Llama models using SageMaker for scalable applications](https://www.philschmid.de/sagemaker-llama-llm).

stevhliu · 2023-08-21T16:18:23Z

+⚗️ Optimization
+- A blog on how to [fine-tune Llama 2 using the Direct Preference Optimization (DPO) method](https://huggingface.co/blog/dpo-trl).
+- A blog on how to [instruction-tune Llama 2 for optimized performance](https://www.philschmid.de/instruction-tune-llama-2) by Phil Schmid.
+- A notebook on how to [fine-tune the Llama 2 model on a personal computer using QLoRa and TRL](https://colab.research.google.com/drive/1SYpgFpcmtIUzdE7pxqknrM4ArCASfkFQ?usp=sharing). 🌎


Put the link emphasis on "notebook" rather than the title of it (same comment for the rest of the notebooks)

wonhyeongseo

It's always fantastic to receive your feedback, @stevhliu ! 😊 Thank you so much for your incredible assistance! 🙌🌟

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

stevhliu

Awesome, once we fix the tiny typo we're ready to merge!

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

wonhyeongseo · 2023-08-22T23:14:24Z

Awesome, once we fix the tiny typo we're ready to merge!

Done and dusted. Thank you so much for your help @stevhliu ! Looking forward to our Q&A session!!

* docs: feat: model resources for llama2 Co-authored-by: Woojun Jung <hello_984@naver.com> * fix: add description for dpo and rearrange posts * docs: feat: add llama2 notebook resources * style: one liners for each resource Co-Authored-By: Woojun Jung <46880056+jungnerd@users.noreply.github.com> Co-Authored-By: Kihoon Son <75935546+kihoon71@users.noreply.github.com> * Apply suggestions from code review Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> * Fix typo Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com> --------- Co-authored-by: Woojun Jung <hello_984@naver.com> Co-authored-by: Woojun Jung <46880056+jungnerd@users.noreply.github.com> Co-authored-by: Kihoon Son <75935546+kihoon71@users.noreply.github.com> Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

docs: feat: model resources for llama2

299677f

Co-authored-by: Woojun Jung <hello_984@naver.com>

fix: add description for dpo and rearrange posts

92defe8

wonhyeongseo changed the title ~~Add LLaMA2 resources~~ Add Llama2 resources Aug 16, 2023

wonhyeongseo marked this pull request as draft August 16, 2023 08:12

docs: feat: add llama2 notebook resources

f9558b0

* style: one liners for each resource Co-Authored-By: Woojun Jung <46880056+jungnerd@users.noreply.github.com> Co-Authored-By: Kihoon Son <75935546+kihoon71@users.noreply.github.com>

wonhyeongseo marked this pull request as ready for review August 17, 2023 14:24

stevhliu approved these changes Aug 21, 2023

View reviewed changes

wonhyeongseo commented Aug 21, 2023

View reviewed changes

Apply suggestions from code review

0feb365

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

stevhliu approved these changes Aug 22, 2023

View reviewed changes

Comment thread docs/source/en/model_doc/llama2.md Outdated

Fix typo

479affa

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

stevhliu merged commit 5794363 into huggingface:main Aug 23, 2023

wonhyeongseo deleted the llama2-resources branch August 23, 2023 01:07

	- A blog on how to [use SageMaker with Llama2 and QLora for efficient model deployment](https://www.philschmid.de/sagemaker-llama2-qlora) by Phil Schmid.
	- A blog on how to [use SageMaker with Llama2 and QLora for efficient model deployment](https://www.philschmid.de/sagemaker-llama2-qlora).

	- A blog on how to [deploy Llama models using SageMaker for scalable applications](https://www.philschmid.de/sagemaker-llama-llm) by Phil Schmid.
	- A blog on how to [deploy Llama models using SageMaker for scalable applications](https://www.philschmid.de/sagemaker-llama-llm).

Uh oh!

Conversation

wonhyeongseo commented Aug 16, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What does this PR do?

Before submitting

Who can review?

Uh oh!

HuggingFaceDocBuilderDev commented Aug 16, 2023

Uh oh!

wonhyeongseo commented Aug 16, 2023

Uh oh!

ArthurZucker commented Aug 16, 2023

Uh oh!

wonhyeongseo commented Aug 16, 2023

Uh oh!

wonhyeongseo commented Aug 21, 2023

Uh oh!

stevhliu left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

stevhliu Aug 21, 2023

Choose a reason for hiding this comment

Uh oh!

stevhliu Aug 21, 2023

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

stevhliu Aug 21, 2023

Choose a reason for hiding this comment

Uh oh!

wonhyeongseo left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

stevhliu left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

wonhyeongseo commented Aug 22, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

wonhyeongseo commented Aug 16, 2023 •

edited

Loading

wonhyeongseo left a comment •

edited

Loading

wonhyeongseo commented Aug 22, 2023 •

edited

Loading