
Most common position in a pipeline	After a `PromptBuilder`
Mandatory init variables	"model": The path of the model to use
Mandatory run variables	“prompt”: A string containing the prompt for the LLM
Output variables	“replies”: A list of strings with all the replies generated by the LLM ”meta”: A list of dictionaries with the metadata associated with each reply, such as token count and others
API reference	Llama.cpp
GitHub link	https://github.com/deepset-ai/haystack-core-integrations/tree/main/integrations/llama_cpp

Overview

Llama.cpp is a library written in C/C++ for efficient inference of Large Language Models. It leverages 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).

Llama.cpp uses the quantized binary file of the LLM in GGUF format that can be downloaded from Hugging Face. LlamaCppGenerator supports models running on Llama.cpp by taking the path to the locally saved GGUF file as model parameter at initialization.

Installation

Install the llama-cpp-haystack package:

pip install llama-cpp-haystack

Using a different compute backend

The default installation behavior is to build llama.cpp for CPU on Linux and Windows and use Metal on MacOS. To use other compute backends:

Follow instructions on the llama.cpp installation page to install llama-cpp-python for your preferred compute backend.
Install llama-cpp-haystack using the command above.

For example, to use llama-cpp-haystack with the cuBLAS backend, you have to run the following commands:

export LLAMA_CUBLAS=1
CMAKE_ARGS="-DLLAMA_CUBLAS=on" pip install llama-cpp-python
pip install llama-cpp-haystack

Usage

You need to download the GGUF version of the desired LLM. The GGUF versions of popular models can be downloaded from Hugging Face.
Initialize a LlamaCppGenerator with the path to the GGUF file and also specify the required model and text generation parameters:

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, n_ctx, n_batch arguments have been exposed for convenience and can be directly passed to the Generator during initialization as keyword arguments. Note that model translates to llama.cpp's model_path parameter.

The model_kwargs parameter can pass additional arguments when initializing the model. In case of duplication, these parameters override the model, n_ctx, and n_batch initialization parameters.

See Llama.cpp's LLM documentation for more information on the available model arguments.

For example, to offload the model to GPU during initialization:

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 pass additional generation arguments like max_tokens, temperature, top_k, top_p, and others to the model during inference.

See Llama.cpp's Completion API documentation for more information on the available generation arguments.

For example, to set the max_tokens and temperature:

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:

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},
)

Using in a Pipeline

We use the LlamaCppGenerator in a Retrieval Augmented Generation pipeline on the Simple Wikipedia Dataset from HuggingFace and generate answers using the OpenChat-3.5 LLM.

Load the dataset:

# 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.in_memory import InMemoryEmbeddingRetriever
from haystack.components.writers import DocumentWriter
from haystack.document_stores.in_memory 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:

doc_store = InMemoryDocumentStore(embedding_similarity_function="cosine")
doc_embedder = SentenceTransformersDocumentEmbedder(model="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:

# 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="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:

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.