[Update Dec. 28, 2023] We added support for Llama Guard as a safety checker for our example inference script and also with standalone inference with an example script and prompt formatting. More details here. For details on formatting data for fine tuning Llama Guard, we provide a script and sample usage here.
[Update Dec 14, 2023] We recently released a series of Llama 2 demo apps here. These apps show how to run Llama (locally, in the cloud, or on-prem), how to use Azure Llama 2 API (Model-as-a-Service), how to ask Llama questions in general or about custom data (PDF, DB, or live), how to integrate Llama with WhatsApp and Messenger, and how to implement an end-to-end chatbot with RAG (Retrieval Augmented Generation).
The 'llama-recipes' repository is a companion to the Llama 2 model. The goal of this repository is to provide examples to quickly get started with fine-tuning for domain adaptation and how to run inference for the fine-tuned models. For ease of use, the examples use Hugging Face converted versions of the models. See steps for conversion of the model here.
In addition, we also provide a number of demo apps, to showcase the Llama 2 usage along with other ecosystem solutions to run Llama 2 locally, in the cloud, and on-prem.
Llama 2 is a new technology that carries potential risks with use. Testing conducted to date has not — and could not — cover all scenarios. In order to help developers address these risks, we have created the Responsible Use Guide. More details can be found in our research paper as well. For downloading the models, follow the instructions on Llama 2 repo.
- Quick start
- Model Conversion
- Fine-tuning
- Inference
- Demo Apps
- Repository Organization
- License and Acceptable Use Policy
Llama 2 Jupyter Notebook: This jupyter notebook steps you through how to finetune a Llama 2 model on the text summarization task using the samsum. The notebook uses parameter efficient finetuning (PEFT) and int8 quantization to finetune a 7B on a single GPU like an A10 with 24GB gpu memory.
Llama-recipes provides a pip distribution for easy install and usage in other projects. Alternatively, it can be installed from source.
pip install --extra-index-url https://download.pytorch.org/whl/test/cu118 llama-recipes
Llama-recipes offers the installation of optional packages. There are three optional dependency groups. To run the unit tests we can install the required dependencies with:
pip install --extra-index-url https://download.pytorch.org/whl/test/cu118 llama-recipes[tests]
For the vLLM example we need additional requirements that can be installed with:
pip install --extra-index-url https://download.pytorch.org/whl/test/cu118 llama-recipes[vllm]
To use the sensitive topics safety checker install with:
pip install --extra-index-url https://download.pytorch.org/whl/test/cu118 llama-recipes[auditnlg]
Optional dependencies can also be combines with [option1,option2].
To install from source e.g. for development use these commands. We're using hatchling as our build backend which requires an up-to-date pip as well as setuptools package.
git clone [email protected]:facebookresearch/llama-recipes.git
cd llama-recipes
pip install -U pip setuptools
pip install --extra-index-url https://download.pytorch.org/whl/test/cu118 -e .
For development and contributing to llama-recipes please install all optional dependencies:
git clone [email protected]:facebookresearch/llama-recipes.git
cd llama-recipes
pip install -U pip setuptools
pip install --extra-index-url https://download.pytorch.org/whl/test/cu118 -e .[tests,auditnlg,vllm]
Note All the setting defined in config files can be passed as args through CLI when running the script, there is no need to change from config files directly.
For more in depth information checkout the following:
- Single GPU Fine-tuning
- Multi-GPU Fine-tuning
- LLM Fine-tuning
- Adding custom datasets
- Inference
- Evaluation Harness
- FAQs
You can find Llama 2 models on Hugging Face hub here, where models with hf in the name are already converted to Hugging Face checkpoints so no further conversion is needed. The conversion step below is only for original model weights from Meta that are hosted on Hugging Face model hub as well.
The recipes and notebooks in this folder are using the Llama 2 model definition provided by Hugging Face's transformers library.
Given that the original checkpoint resides under models/7B you can install all requirements and convert the checkpoint with:
## Install Hugging Face Transformers from source
pip freeze | grep transformers ## verify it is version 4.31.0 or higher
git clone [email protected]:huggingface/transformers.git
cd transformers
pip install protobuf
python src/transformers/models/llama/convert_llama_weights_to_hf.py \
--input_dir /path/to/downloaded/llama/weights --model_size 7B --output_dir /output/pathFor fine-tuning Llama 2 models for your domain-specific use cases recipes for PEFT, FSDP, PEFT+FSDP have been included along with a few test datasets. For details see LLM Fine-tuning.
If you want to dive right into single or multi GPU fine-tuning, run the examples below on a single GPU like A10, T4, V100, A100 etc. All the parameters in the examples and recipes below need to be further tuned to have desired results based on the model, method, data and task at hand.
Note:
-
To change the dataset in the commands below pass the
datasetarg. Current options for integrated dataset aregrammar_dataset,alpaca_datasetandsamsum_dataset. Additionally, we integrate the OpenAssistant/oasst1 dataset as an example for a custom dataset. A description of how to use your own dataset and how to add custom datasets can be found in Dataset.md. Forgrammar_dataset,alpaca_datasetplease make sure you use the suggested instructions from here to set them up. -
Default dataset and other LORA config has been set to
samsum_dataset. -
Make sure to set the right path to the model in the training config.
-
To save the loss and perplexity metrics for evaluation, enable this by passing
--save_metricsto the finetuning script. The file can be plotted using the plot_metrics.py script,python examples/plot_metrics.py --file_path path/to/metrics.json
#if running on multi-gpu machine
export CUDA_VISIBLE_DEVICES=0
python -m llama_recipes.finetuning --use_peft --peft_method lora --quantization --model_name /path_of_model_folder/7B --output_dir path/to/save/PEFT/model
Here we make use of Parameter Efficient Methods (PEFT) as described in the next section. To run the command above make sure to pass the peft_method arg which can be set to lora, llama_adapter or prefix.
Note if you are running on a machine with multiple GPUs please make sure to only make one of them visible using export CUDA_VISIBLE_DEVICES=GPU:id
Make sure you set save_model parameter to save the model. Be sure to check the other training parameter in train config as well as others in the config folder as needed. All parameter can be passed as args to the training script. No need to alter the config files.
NOTE please make sure to use PyTorch Nightlies for using PEFT+FSDP. Also, note that int8 quantization from bit&bytes currently is not supported in FSDP.
torchrun --nnodes 1 --nproc_per_node 4 examples/finetuning.py --enable_fsdp --use_peft --peft_method lora --model_name /path_of_model_folder/7B --fsdp_config.pure_bf16 --output_dir path/to/save/PEFT/model
Here we use FSDP as discussed in the next section which can be used along with PEFT methods. To make use of PEFT methods with FSDP make sure to pass use_peft and peft_method args along with enable_fsdp. Here we are using BF16 for training.
Setting use_fast_kernels will enable using of Flash Attention or Xformer memory-efficient kernels based on the hardware being used. This would speed up the fine-tuning job. This has been enabled in optimum library from Hugging Face as a one-liner API, please read more here.
torchrun --nnodes 1 --nproc_per_node 4 examples/finetuning.py --enable_fsdp --use_peft --peft_method lora --model_name /path_of_model_folder/7B --fsdp_config.pure_bf16 --output_dir path/to/save/PEFT/model --use_fast_kernelsIf you are interested in running full parameter fine-tuning without making use of PEFT methods, please use the following command. Make sure to change the nproc_per_node to your available GPUs. This has been tested with BF16 on 8xA100, 40GB GPUs.
torchrun --nnodes 1 --nproc_per_node 8 examples/finetuning.py --enable_fsdp --model_name /path_of_model_folder/7B --dist_checkpoint_root_folder model_checkpoints --dist_checkpoint_folder fine-tuned --use_fast_kernels
If you are interested in running full parameter fine-tuning on the 70B model, you can enable low_cpu_fsdp mode as the following command. This option will load model on rank0 only before moving model to devices to construct FSDP. This can dramatically save cpu memory when loading large models like 70B (on a 8-gpu node, this reduces cpu memory from 2+T to 280G for 70B model). This has been tested with BF16 on 16xA100, 80GB GPUs.
torchrun --nnodes 1 --nproc_per_node 8 examples/finetuning.py --enable_fsdp --low_cpu_fsdp --fsdp_config.pure_bf16 --model_name /path_of_model_folder/70B --batch_size_training 1 --dist_checkpoint_root_folder model_checkpoints --dist_checkpoint_folder fine-tuned
sbatch multi_node.slurm
# Change the num nodes and GPU per nodes in the script before running.
You can read more about our fine-tuning strategies here.
Here, we make use lm-evaluation-harness from EleutherAI for evaluation of fine-tuned Llama 2 models. This also can extend to evaluate other optimizations for inference of Llama 2 model such as quantization. Please use this get started doc.
This folder contains a series of Llama2-powered apps:
- Quickstart Llama deployments and basic interactions with Llama
- Llama on your Mac and ask Llama general questions
- Llama on Google Colab
- Llama on Cloud and ask Llama questions about unstructured data in a PDF
- Llama on-prem with vLLM and TGI
- Llama chatbot with RAG (Retrieval Augmented Generation)
- Azure Llama 2 API (Model-as-a-Service)
- Specialized Llama use cases:
- Ask Llama to summarize a video content
- Ask Llama questions about structured data in a DB
- Ask Llama questions about live data on the web
- Build a Llama-enabled WhatsApp chatbot
This folder contains a series of benchmark scripts for Llama 2 models inference on various backends:
- On-prem - Popular serving frameworks and containers (i.e. vLLM)
- (WIP) Cloud API - Popular API services (i.e. Azure Model-as-a-Service)
- (WIP) On-device - Popular on-device inference solutions on Android and iOS (i.e. mlc-llm, QNN)
- (WIP) Optimization - Popular optimization solutions for faster inference and quantization (i.e. AutoAWQ)
This repository is organized in the following way: benchmarks: Contains a series of benchmark scripts for Llama 2 models inference on various backends.
configs: Contains the configuration files for PEFT methods, FSDP, Datasets.
docs: Example recipes for single and multi-gpu fine-tuning recipes.
datasets: Contains individual scripts for each dataset to download and process. Note: Use of any of the datasets should be in compliance with the dataset's underlying licenses (including but not limited to non-commercial uses)
demo_apps: Contains a series of Llama2-powered apps, from quickstart deployments to how to ask Llama questions about unstructured data, structured data, live data, and video summary.
examples: Contains examples script for finetuning and inference of the Llama 2 model as well as how to use them safely.
inference: Includes modules for inference for the fine-tuned models.
model_checkpointing: Contains FSDP checkpoint handlers.
policies: Contains FSDP scripts to provide different policies, such as mixed precision, transformer wrapping policy and activation checkpointing along with any precision optimizer (used for running FSDP with pure bf16 mode).
utils: Utility files for:
-
train_utils.pyprovides training/eval loop and more train utils. -
dataset_utils.pyto get preprocessed datasets. -
config_utils.pyto override the configs received from CLI. -
fsdp_utils.pyprovides FSDP wrapping policy for PEFT methods. -
memory_utils.pycontext manager to track different memory stats in train loop.