From 12a53037187342a3e455a56557e25e2f6bbadd19 Mon Sep 17 00:00:00 2001 From: Ashwin Mathur <97467100+awinml@users.noreply.github.com> Date: Mon, 22 Jan 2024 15:13:46 +0530 Subject: [PATCH] Add llama.cpp Integration (#118) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit * Add llama.cpp integration * Update import with new namespace Co-authored-by: Bilge Yücel * Change model_path to model Co-authored-by: ZanSara * Add a section on how to download models --------- Co-authored-by: Bilge Yücel Co-authored-by: ZanSara --- integrations/llama_cpp.md | 282 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 282 insertions(+) create mode 100644 integrations/llama_cpp.md diff --git a/integrations/llama_cpp.md b/integrations/llama_cpp.md new file mode 100644 index 00000000..5e1c98e3 --- /dev/null +++ b/integrations/llama_cpp.md @@ -0,0 +1,282 @@ +--- +layout: integration +name: Llama.cpp +description: Use Llama.cpp models with Haystack. +authors: + - name: Ashwin Mathur + socials: + github: awinml + twitter: awinml + linkedin: ashwin-mathur-ds +pypi: https://pypi.org/project/llama-cpp-haystack/ +repo: https://github.com/deepset-ai/haystack-core-integrations/tree/main/integrations/llama_cpp +type: Model Provider +report_issue: https://github.com/deepset-ai/haystack-core-integrations/issues +version: Haystack 2.0 +toc: true +--- + +### Table of Contents + +- [Introduction](#introduction) +- [Installation](#installation) + - [Using a different compute backend](#using-a-different-compute-backend) +- [Downloading Models](#downloading-models) +- [Usage](#usage) + - [Passing additional model parameters](#passing-additional-model-parameters) + - [Passing text generation parameters](#passing-text-generation-parameters) +- [Example: RAG Pipeline](#example-rag-pipeline) + +## Introduction + +[Llama.cpp](https://github.com/ggerganov/llama.cpp) is a library written in C/C++ for efficient inference of Large Language models. It uses the efficient quantized GGUF format, dramatically reducing memory requirements and accelerating inference. This means it is possible to run LLMs efficiently on standard machines (even without GPUs). + +## Installation + +Install the `llama-cpp-haystack` package: + +```bash +pip install llama-cpp-haystack +``` + +### Using a different compute backend + +The default installation behaviour is to build `llama.cpp` for CPU on Linux and Windows and use Metal on MacOS. To use other compute backends: + +1. Follow instructions on the [llama.cpp installation page](https://github.com/abetlen/llama-cpp-python#installation) to install [llama-cpp-python](https://github.com/abetlen/llama-cpp-python) for your preffered compute backend. +2. Install [llama-cpp-haystack](https://github.com/deepset-ai/haystack-core-integrations/tree/main/integrations/llama_cpp) using the command above. + +For example, to use `llama-cpp-haystack` with the **cuBLAS backend**, you have to run the following commands: + +```bash +export LLAMA_CUBLAS=1 +CMAKE_ARGS="-DLLAMA_CUBLAS=on" pip install llama-cpp-python +pip install llama-cpp-haystack +``` + +## Downloading Models + +Llama.cpp requires the quantized binary of the LLM in GGUF format. + +The GGUF versions of popular LLMs can be downloaded from [HuggingFace](https://huggingface.co/models?library=gguf). + +For example, to download the GGUF version of [OpenChat3.5](https://huggingface.co/openchat/openchat_3.5), we find the required [GGUF version on HuggingFace](https://huggingface.co/TheBloke/openchat-3.5-1210-GGUF) and then download the file to disk: + +```python +import os +import urllib.request + +def download_file(file_link, filename): + # Checks if the file already exists before downloading + if not os.path.isfile(filename): + urllib.request.urlretrieve(file_link, filename) + print("Model file downloaded successfully.") + else: + print("Model file already exists.") + +# Download GGUF model from HuggingFace +ggml_model_path = ( + "https://huggingface.co/TheBloke/openchat-3.5-1210-GGUF/resolve/main/openchat-3.5-1210.Q3_K_S.gguf" +) +filename = "openchat-3.5-1210.Q3_K_S.gguf" +download_file(ggml_model_path, filename) +``` + +You could also directly download the file from the command line using Curl: + +```bash +curl -L -O "https://huggingface.co/TheBloke/openchat-3.5-1210-GGUF/resolve/main/openchat-3.5-1210.Q3_K_S.gguf" +``` + +## Usage + +You can leverage Llama.cpp to run models by using the `LlamaCppGenerator` component. + +Initialize an `LlamaCppGenerator` with the the path to the GGUF file and also specify the required model and text generation parameters: + +```python +from haystack_integrations.components.generators.llama_cpp import LlamaCppGenerator + +generator = LlamaCppGenerator( + model="/content/openchat-3.5-1210.Q3_K_S.gguf", + n_ctx=512, + n_batch=128, + model_kwargs={"n_gpu_layers": -1}, + generation_kwargs={"max_tokens": 128, "temperature": 0.1}, +) +generator.warm_up() +prompt = f"Who is the best American actor?" +result = generator.run(prompt) +``` + +### Passing additional model parameters + +The `model_path`, `n_ctx`, `n_batch` arguments have been exposed for convenience and can be directly passed to the Generator during initialization as keyword arguments. + +The `model_kwargs` parameter can be used to pass additional arguments when initializing the model. In case of duplication, these parameters override the `model_path`, `n_ctx`, and `n_batch` initialization parameters. + +See [Llama.cpp's LLM documentation](https://llama-cpp-python.readthedocs.io/en/latest/api-reference/#llama_cpp.Llama.__init__) for more information on the available model arguments. + +For example, to offload the model to GPU during initialization: + +```python +from haystack_integrations.components.generators.llama_cpp import LlamaCppGenerator + +generator = LlamaCppGenerator( + model="/content/openchat-3.5-1210.Q3_K_S.gguf", + n_ctx=512, + n_batch=128, + model_kwargs={"n_gpu_layers": -1} +) +generator.warm_up() +prompt = f"Who is the best American actor?" +result = generator.run(prompt, generation_kwargs={"max_tokens": 128}) +generated_text = result["replies"][0] +print(generated_text) +``` + +### Passing text generation parameters + +The `generation_kwargs` parameter can be used to pass additional generation arguments like `max_tokens`, `temperature`, `top_k`, `top_p`, etc to the model during inference. + +See [Llama.cpp's Completion API documentation](https://llama-cpp-python.readthedocs.io/en/latest/api-reference/#llama_cpp.Llama.create_completion) for more information on the available generation arguments. + +For example, to set the `max_tokens` and `temperature`: + +```python +from haystack_integrations.components.generators.llama_cpp import LlamaCppGenerator + +generator = LlamaCppGenerator( + model="/content/openchat-3.5-1210.Q3_K_S.gguf", + n_ctx=512, + n_batch=128, + generation_kwargs={"max_tokens": 128, "temperature": 0.1}, +) +generator.warm_up() +prompt = f"Who is the best American actor?" +result = generator.run(prompt) +``` + +The `generation_kwargs` can also be passed to the `run` method of the generator directly: + +```python +from haystack_integrations.components.generators.llama_cpp import LlamaCppGenerator + +generator = LlamaCppGenerator( + model="/content/openchat-3.5-1210.Q3_K_S.gguf", + n_ctx=512, + n_batch=128, +) +generator.warm_up() +prompt = f"Who is the best American actor?" +result = generator.run( + prompt, + generation_kwargs={"max_tokens": 128, "temperature": 0.1}, +) +``` + +## Example: RAG Pipeline + +We use the `LlamaCppGenerator` in a Retrieval Augmented Generation pipeline on the [Simple Wikipedia](https://huggingface.co/datasets/pszemraj/simple_wikipedia) Dataset from HuggingFace and generate answers using the [OpenChat-3.5](https://huggingface.co/openchat/openchat-3.5-1210) LLM. + +**Load the dataset:** + +```python +# Install HuggingFace Datasets using "pip install datasets" +from datasets import load_dataset +from haystack import Document, Pipeline +from haystack.components.builders.answer_builder import AnswerBuilder +from haystack.components.builders.prompt_builder import PromptBuilder +from haystack.components.embedders import SentenceTransformersDocumentEmbedder, SentenceTransformersTextEmbedder +from haystack.components.retrievers import InMemoryEmbeddingRetriever +from haystack.components.writers import DocumentWriter +from haystack.document_stores import InMemoryDocumentStore + +# Import LlamaCppGenerator +from haystack_integrations.components.generators.llama_cpp import LlamaCppGenerator + +# Load first 100 rows of the Simple Wikipedia Dataset from HuggingFace +dataset = load_dataset("pszemraj/simple_wikipedia", split="validation[:100]") + +docs = [ + Document( + content=doc["text"], + meta={ + "title": doc["title"], + "url": doc["url"], + }, + ) + for doc in dataset +] +``` + +**Index the documents to the `InMemoryDocumentStore` using the `SentenceTransformersDocumentEmbedder` and `DocumentWriter`:** + +```python +doc_store = InMemoryDocumentStore(embedding_similarity_function="cosine") +doc_embedder = SentenceTransformersDocumentEmbedder(model_name_or_path="sentence-transformers/all-MiniLM-L6-v2") + +# Indexing Pipeline +indexing_pipeline = Pipeline() +indexing_pipeline.add_component(instance=doc_embedder, name="DocEmbedder") +indexing_pipeline.add_component(instance=DocumentWriter(document_store=doc_store), name="DocWriter") +indexing_pipeline.connect(connect_from="DocEmbedder", connect_to="DocWriter") + +indexing_pipeline.run({"DocEmbedder": {"documents": docs}}) +``` + +**Create the Retrieval Augmented Generation (RAG) pipeline and add the `LlamaCppGenerator` to it:** + +```python +# Prompt Template for the https://huggingface.co/openchat/openchat-3.5-1210 LLM +prompt_template = """GPT4 Correct User: Answer the question using the provided context. +Question: {{question}} +Context: +{% for doc in documents %} + {{ doc.content }} +{% endfor %} +<|end_of_turn|> +GPT4 Correct Assistant: +""" + +rag_pipeline = Pipeline() + +text_embedder = SentenceTransformersTextEmbedder(model_name_or_path="sentence-transformers/all-MiniLM-L6-v2") + +# Load the LLM using LlamaCppGenerator +model_path = "openchat-3.5-1210.Q3_K_S.gguf" +generator = LlamaCppGenerator(model=model_path, n_ctx=4096, n_batch=128) + +rag_pipeline.add_component( + instance=text_embedder, + name="text_embedder", +) +rag_pipeline.add_component(instance=InMemoryEmbeddingRetriever(document_store=doc_store, top_k=3), name="retriever") +rag_pipeline.add_component(instance=PromptBuilder(template=prompt_template), name="prompt_builder") +rag_pipeline.add_component(instance=generator, name="llm") +rag_pipeline.add_component(instance=AnswerBuilder(), name="answer_builder") + +rag_pipeline.connect("text_embedder", "retriever") +rag_pipeline.connect("retriever", "prompt_builder.documents") +rag_pipeline.connect("prompt_builder", "llm") +rag_pipeline.connect("llm.replies", "answer_builder.replies") +rag_pipeline.connect("retriever", "answer_builder.documents") +``` + +**Run the pipeline:** + +```python +question = "Which year did the Joker movie release?" +result = rag_pipeline.run( + { + "text_embedder": {"text": question}, + "prompt_builder": {"question": question}, + "llm": {"generation_kwargs": {"max_tokens": 128, "temperature": 0.1}}, + "answer_builder": {"query": question}, + } +) + +generated_answer = result["answer_builder"]["answers"][0] +print(generated_answer.data) +# The Joker movie was released on October 4, 2019. +```