Update llama-3 PEFT notebook to download model from NGC

[shashank3959]

• Updates the llama-3 PEFT tutorial to download the model from NVIDIA NGC to skip conversion

[cuichenx]

This looks good to me

[jgerh]

I am providing additional copyedits to the file, ReadMe.rst, herein. I was not able to provide inline edits because part of the file was read-only.

Line 4 fix punctuation

Llama 3 <https://blogs.nvidia.com/blog/meta-llama3-inference-acceleration/>_ is an open-source large language model by Meta that delivers state-of-the-art performance on popular industry benchmarks.

Line 6 fix punctuation

Low-Rank Adaptation (LoRA) <https://arxiv.org/pdf/2106.09685>__ has emerged as a popular Parameter-Efficient Fine-Tuning (PEFT) technique that tunes a very small number of additional parameters as compared to full fine-tuning, thereby reducing the compute required.

Line 19 revise for clarity

NIM enables seamless deployment of multiple LoRA adapters (referred to as ‘multi-LoRA’) on the same base model. It dynamically loads adapter weights based on incoming requests at runtime. This flexibility allows handling inputs from various tasks or use cases without requiring a unique model for each individual scenario. For further details, consult the NIM documentation for LLMs.

Line 24 delete the following sentence, it is sufficient to retain the bullets

In order to proceed, ensure that you have met the following requirements:

Line 103/109 fix punctuation

2. Prepare the LoRA model store.

Line 110/116 fix grammar

To ensure the model store is organized as expected, create a folder named llama3-8b-pubmed-qa and move your .nemo checkpoint there.

Line 122/128 revise for clarity

Ensure that the LoRA model store directory follows this structure: the model name should be a sub-folder containing the .nemo file.

Line 138/144 fix punctuation

3. Set up NIM.

Line 170/176 fix grammar

The first time you run the command, it will download the model and cache it in $NIM_CACHE_PATH so subsequent deployments are even faster. There are several options to configure NIM other than the ones listed above. You can find a full list in the NIM configuration documentation.

Line 173/179 fix punctuation

4. Start the notebook.

Line 175/181 fix grammar

From another terminal, follow the same instructions as the previous notebook to launch Jupyter Lab, and then navigate to this notebook.

Line 178/185 revise for clarity

You can use the same NeMo Framework docker container in which Jupyter Lab was previously installed.

[cuichenx]

LGTM