LlamaCppGenerator

Generate text using LLMs through llama.cpp.

Key Features

• Runs LLMs locally using llama.cpp, a C/C++ library for efficient LLM inference.
• Uses the quantized GGUF format, suitable for running on standard machines without GPUs.
• Accepts string prompts and returns string replies.
• Compatible with any GGUF-format model available on Hugging Face.

Configuration

Before using this component, download the GGUF version of the model you want from Hugging Face and save it locally.

1. Drag the LlamaCppGenerator component onto the canvas from the Component Library.
2. Click on the component to open the configuration panel.
3. On the General tab:
   • Set the model parameter to the local path of the GGUF file (for example, /models/zephyr-7b-beta.Q4_0.gguf).
4. Go to the Advanced tab to configure n_ctx, n_batch, model_kwargs, and generation_kwargs.

Connections

LlamaCppGenerator accepts a prompt string as input. Connect its prompt input to the prompt output of PromptBuilder.

It outputs replies as a list of strings and meta as a list of metadata dictionaries. Connect its replies output to AnswerBuilder.

Source Code

To check this component's source code, open generator.py in the Haystack Core Integrations repository.

Usage Examples

Basic Configuration

  LlamaCppGenerator:
    type: haystack_integrations.components.generators.llama_cpp.generator.LlamaCppGenerator
    init_parameters:
      model: /models/zephyr-7b-beta.Q4_0.gguf
      n_ctx: 2048
      n_batch: 512
      generation_kwargs:
        max_tokens: 256
        temperature: 0.7
        top_p: 0.9

Connections

This example shows a simple question-answering pipeline using LlamaCppGenerator with a locally hosted GGUF model:

components:
  PromptBuilder:
    type: haystack.components.builders.prompt_builder.PromptBuilder
    init_parameters:
      template: |
        Answer the following question concisely and accurately.
        
        Question: {{ question }}
        Answer:
  LlamaCppGenerator:
    type: haystack_integrations.components.generators.llama_cpp.generator.LlamaCppGenerator
    init_parameters:
      model: /models/zephyr-7b-beta.Q4_0.gguf
      n_ctx: 2048
      n_batch: 512
      generation_kwargs:
        max_tokens: 256
        temperature: 0.7
        top_p: 0.9
  AnswerBuilder:
    type: haystack.components.builders.answer_builder.AnswerBuilder
    init_parameters:
      pattern:
      reference_pattern:

connections:
- sender: PromptBuilder.prompt
  receiver: LlamaCppGenerator.prompt
- sender: LlamaCppGenerator.replies
  receiver: AnswerBuilder.replies

max_runs_per_component: 100

metadata: {}

inputs:
  question:
  - PromptBuilder.question
  - AnswerBuilder.query

outputs:
  answers: AnswerBuilder.answers

Parameters

Inputs

Parameter	Type	Default	Description
prompt	str		The prompt to be sent to the generative model.
generation_kwargs	Optional[Dict[str, Any]]	None	A dictionary containing keyword arguments to customize text generation. These kwargs are merged with any generation_kwargs set during initialization. For more information on the available kwargs, see llama.cpp documentation.

Outputs

Parameter	Type	Default	Description
replies	List[str]		The list of string replies generated by the model.
meta	List[Dict[str, Any]]		Metadata about the request, including completion details from llama.cpp.

Init Parameters

These are the parameters you can configure in Pipeline Builder:

Parameter	Type	Default	Description
model	str		The path of a quantized model for text generation, for example, "zephyr-7b-beta.Q4_0.gguf". If the model path is also specified in the model_kwargs, this parameter is ignored.
n_ctx	Optional[int]	0	The number of tokens in the context. When set to 0, the context is taken from the model.
n_batch	Optional[int]	512	Prompt processing maximum batch size.
model_kwargs	Optional[Dict[str, Any]]	None	Dictionary containing keyword arguments used to initialize the LLM for text generation. These keyword arguments provide fine-grained control over the model loading. In case of duplication, these kwargs override model, n_ctx, and n_batch init parameters. For more information on the available kwargs, see llama.cpp documentation.
generation_kwargs	Optional[Dict[str, Any]]	None	A dictionary containing keyword arguments to customize text generation. For more information on the available kwargs, see llama.cpp documentation.

Run Method Parameters

These are the parameters you can configure for the component's run() method. This means you can pass these parameters at query time through the API, in Playground, or when running a job. For details, see Modify Pipeline Parameters at Query Time.

Parameter	Type	Default	Description
prompt	str		The prompt to be sent to the generative model.
generation_kwargs	Optional[Dict[str, Any]]	None	A dictionary containing keyword arguments to customize text generation. For more information on the available kwargs, see llama.cpp documentation.