SkillAgentSearch skills...

Inferflow

Inferflow is an efficient and highly configurable inference engine for large language models (LLMs).

GenerateConvertImprove

Install / Use

/learn @inferflow/Inferflow
About this skill

Quality Score

0/100

Supported Platforms

Universal

README

Inferflow

<h4>

<img src="https://img.shields.io/badge/Version-1.0-blue.svg" alt="Version"> <img src="https://img.shields.io/github/stars/inferflow/inferflow?color=yellow" alt="Stars"> <img src="https://img.shields.io/github/issues/inferflow/inferflow?color=red" alt="Issues">

</h4>

Inferflow is an efficient and highly configurable inference engine for large language models (LLMs). With Inferflow, users can serve most of the common transformer models by simply modifying some lines in corresponding configuration files, without writing a single line of source code. Further details can be found in our technical report.

Quick Links

  1. Getting started (on Windows | on Linux, Mac, and Windows Subsystem for Linux (WSL))
  2. Serving 34B or 40B models on a single 24GB-VRAM GPU (e.g., RTX 3090 and 4090)

Milestones

  • 2024-2-18: Added support for mixture-of-experts (MoE) models.
  • 2024-1-17: Version 0.1.0 was formally released.

Main Features

  1. Extensible and highly configurable: A typical way of using Inferflow to serve a new model is editing a model specification file, but not adding/editing source codes. We implement in Inferflow a modular framework of atomic building-blocks and technologies, making it compositionally generalizable to new models. A new model can be served by Inferflow if the atomic building-blocks and technologies in this model have been "known" (to Inferflow).
  2. 3.5-bit quantization: Inferflow implements 2-bit, 3-bit, 3.5-bit, 4-bit, 5-bit, 6-bit and 8-bit quantization. Among the quantization schemes, 3.5-bit quantization is a new one introduced by Inferflow.
  3. Hybrid model partition for multi-GPU inference: Inferflow supports multi-GPU inference with three model partitioning strategies to choose from: partition-by-layer (pipeline parallelism), partition-by-tensor (tensor parallelism), and hybrid partitioning (hybrid parallelism). Hybrid partitioning is seldom supported by other inference engines.
  4. Wide file format support (and safely loading pickle data): Inferflow supports loading models of multiple file formats directly, without reliance on an external converter. Supported formats include pickle, safetensors, llama.cpp gguf, etc. It is known that there are security issues to read pickle files using Python codes. By implementing a simplified pickle parser in C++, Inferflow supports safely loading models from pickle data.
  5. Wide network type support: Supporting three types transformer models: decoder-only models, encoder-only models, and encoder-decoder models.
  6. GPU/CPU hybrid inference: Supporting GPU-only, CPU-only, and GPU/CPU hybrid inference.

Below is a comparison between Inferflow and some other inference engines:

| Inference Engine | New Model Support | Supported File Formats | Network Structures | Quantization Bits | Hybrid Parallelism for Multi-GPU Inference | Programming Languages | |--------------------------------------------------------------|----------------------|--------------------------|--------------------|-------------------|:------------------------------------------:|-----------------------| | Huggingface Transformers | Adding/editing source codes | pickle (unsafe), safetensors | decoder-only, encoder-decoder, encoder-only | 4b, 8b | ✘ | Python | | vLLM | Adding/editing source codes | pickle (unsafe), safetensors | decoder-only | 4b, 8b | ✘ | Python | | TensorRT-LLM | Adding/editing source codes | | decoder-only, encoder-decoder, encoder-only | 4b, 8b | ✘ | C++, Python | | DeepSpeed-MII | Adding/editing source codes | pickle (unsafe), safetensors | decoder-only | - | ✘ | Python | | llama.cpp | Adding/editing source codes | gguf | decoder-only | 2b, 3b, 4b, 5b, 6b, 8b | ✘ | C/C++ | | llama2.c | Adding/editing source codes | llama2.c | decoder-only                        | - | ✘ | C | | LMDeploy | Adding/editing source codes | pickle (unsafe), TurboMind | decoder-only | 4b, 8b | ✘ | C++, Python | | Inferflow | Editing configuration files | pickle (safe), safetensors, gguf, llama2.c | decoder-only, encoder-decoder, encoder-only | 2b, 3b, 3.5b, 4b, 5b, 6b, 8b |      ✔      | C++ |

Support Matrix

Supported Model File Formats

  • [X] Pickle (Inferflow reduces the security issue of most other inference engines in loading pickle-format files).
  • [X] Safetensors
  • [X] llama.cpp gguf
  • [X] llama2.c

Supported Technologies, Modules, and Options

  • Supported modules and technologies related to model definition:

    • Normalization functions: STD, RMS
    • Activation functions: RELU, GELU, SILU
    • Position embeddings: ALIBI, RoPE, Sinusoidal
    • Grouped-query attention
    • Parallel attention

  • Supported technologies and options related to serving:

    • Linear quantization of weights and KV cache elements: 2-bit, 3b, 3.5b, 4b, 5b, 6b, 8b
    • The option of moving part of all of the KV cache from VRAM to regular RAM
    • The option of placing the input embedding tensor(s) to regular RAM
    • Model partitioning strategies for multi-GPU inference: partition-by-layer, partition-by-tensor, hybrid partitioning
    • Dynamic batching
    • Decoding strategies: Greedy, top-k, top-p, FSD, typical, mirostat...

Supported Transformer Models

  • Decoder-only: Inferflow supports many types of decoder-only transformer models.
  • Encoder-decoder: Some types of encoder-decoder models are supported.
  • Encoder-only: Some types of encoder-only models are supported.

Models with Predefined Specification Files

Users can serve a model with Inferflow by editing a model specification file. We have built predefined specification files for some popular or representative models. Below is a list of such models.

  • [X] Aquila (aquila_chat2_34b)
  • [X] Baichuan (baichuan2_7b_chat, baichuan2_13b_chat)
  • [X] BERT (bert-base-multilingual-cased)
  • [X] Bloom (bloomz_3b)
  • [X] ChatGLM (chatglm2_6b)
  • [X] Deepseek (deepseek_moe_16b_base)
  • [X] Facebook m2m100 (facebook_m2m100_418m)
  • [X] Falcon (falcon_7b_instruct, falcon_40b_instruct)
  • [X] FuseLLM (fusellm_7b)
  • [X] Gemma (gemma_2b_it)
  • [X] Internlm (internlm-chat-20b)
  • [X] LLAMA2 (llama2_7b, llama2_7b_chat, llama2_13b_chat)
  • [X] MiniCPM (minicpm_2b_dpo_bf16)
  • [X] Mistral (mistral_7b_instruct)
  • [X] Mixtral (mixtral_8x7b_instruct_v0.1)
  • [X] Open LLAMA (open_llama_3b)
  • [X] OPT (opt_350m, opt_13b, opt_iml_max_30b)
  • [X] Orion (orion_14b_chat)
  • [X] Phi-2 (phi_2)
  • [X] Qwen (qwen1.5_7b_chat)
  • [X] XVERSE (xverse_13b_chat)
  • [X] YI (yi_6b, yi_34b_chat)

Getting Started

Windows users: Please refer to docs/getting_started.win.md for the instructions about building and running the Inferflow tools and service on Windows.

The following instructions are for Linux, Mac, and WSL (Windows Subsystem for Linux).

Get the Code

git clone https://github.com/inferflow/inferflow
cd inferflow

Build

  • Build the GPU version (that supports GPU/CPU hybrid inference):

    mkdir build/gpu
cd build/gpu
cmake ../.. -DUSE_CUDA=1 -DCMAKE_CUDA_ARCHITECTURES=75
make install -j 8

  • Build the CPU-only version:

    mkdir build/cpu
cd build/cpu
cmake ../.. -DUSE_CUDA=0
make install -j 8

Upon a successful build, executables are generated and copied to bin/release/

Run the LLM Inferencing Tool (bin/llm_inference)

  • Example-1: Load a tiny model and perform inference

    • Step-1: Download the model

      #> cd {inferflow-root-dir}/data/models/llama2.c/
#> bash download.sh

      Instead of running the above batch script, you can also manually download the model files and copy them to the above folder. The source URL and file names can be found from download.sh.

    • Step-2: Run the llm_inference tool:

      #> cd {inferflow-root-dir}/bin/
#> release/llm_inference llm_inference.tiny.ini

      Please note that it is okay for llm_inference and llm_inference.tiny.ini not being in the same folder (llm_inference.tiny.ini is in bin/ and llm_inference is in bin/release/).

  • Example-2: Run the llm_inference tool to load a larger model for inference

    • Step-1: Edit configuration file bin/inferflow_service.ini to choose a model.

      In the "transformer_engine" section of bin/inferflow_service.ini, there are multiple lines starting with "models = " or ";models = ". The lines starting with the ";" character are comments. To choose a model for inference, please uncomment the line corresponding to this model, and comment the lines of other models. By default, the phi-2 model is selected. Please refer to docs/model_serving_config.md for more information about editing the configuration of inferflow_service.

    • Step-2: Download the selected model

      #> cd {inferflow-root-dir}/data/models/{model-name}/
#> bash download.sh

    • Step-3: E

inferflow

View profile

View on GitHub
GitHub Stars250
CategoryDevelopment
Updated12d ago
Forks23
inferflow/inferflow

Languages

C++

Security Score

100/100

Audited on Mar 15, 2026

No findings