MyGit

h2oai/h2ogpt

Fork: 1168 Star: 10560 (更新于 2024-05-13 01:47:29)

license: Apache-2.0

Language: Python .

Private chat with local GPT with document, images, video, etc. 100% private, Apache 2.0. Supports oLLaMa, Mixtral, llama.cpp, and more. Demo: https://gpt.h2o.ai/ https://codellama.h2o.ai/

最后发布版本: client-nightly-2024-03-14 ( 2024-03-14 20:04:05)

官方网址 GitHub网址

h2oGPT

Turn ★ into ⭐ (top-right corner) if you like the project!

Query and summarize your documents or just chat with local private GPT LLMs using h2oGPT, an Apache V2 open-source project.

  • Private offline database of any documents (PDFs, Excel, Word, Images, Video Frames, Youtube, Audio, Code, Text, MarkDown, etc.)
    • Persistent database (Chroma, Weaviate, or in-memory FAISS) using accurate embeddings (instructor-large, all-MiniLM-L6-v2, etc.)
    • Efficient use of context using instruct-tuned LLMs (no need for LangChain's few-shot approach)
    • Parallel summarization and extraction, reaching an output of 80 tokens per second with the 13B LLaMa2 model
    • HYDE (Hypothetical Document Embeddings) for enhanced retrieval based upon LLM responses
  • Variety of models supported (LLaMa2, Mistral, Falcon, Vicuna, WizardLM. With AutoGPTQ, 4-bit/8-bit, LORA, etc.)
    • GPU support from HF and LLaMa.cpp GGML models, and CPU support using HF, LLaMa.cpp, and GPT4ALL models
    • Attention Sinks for arbitrarily long generation (LLaMa-2, Mistral, MPT, Pythia, Falcon, etc.)
  • UI or CLI with streaming of all models
    • Upload and View documents through the UI (control multiple collaborative or personal collections)
    • Vision Models LLaVa, Claude-3, Gemini-Pro-Vision, GPT-4-Vision
    • Image Generation Stable Diffusion (sdxl-turbo, sdxl) and PlaygroundAI (playv2)
    • Voice STT using Whisper with streaming audio conversion
    • Voice TTS using MIT-Licensed Microsoft Speech T5 with multiple voices and Streaming audio conversion
    • Voice TTS using MPL2-Licensed TTS including Voice Cloning and Streaming audio conversion
    • AI Assistant Voice Control Mode for hands-free control of h2oGPT chat
    • Bake-off UI mode against many models at the same time
    • Easy Download of model artifacts and control over models like LLaMa.cpp through the UI
    • Authentication in the UI by user/password via Native or Google OAuth
    • State Preservation in the UI by user/password
  • Linux, Docker, macOS, and Windows support
  • Inference Servers support (oLLaMa, HF TGI server, vLLM, Gradio, ExLLaMa, Replicate, OpenAI, Azure OpenAI, Anthropic)
  • OpenAI-compliant
    • Server Proxy API (h2oGPT acts as drop-in-replacement to OpenAI server)
    • Python client API (to talk to Gradio server)
  • JSON Mode with any model via code block extraction. Also supports MistralAI JSON mode, Claude-3 via function calling with strict Schema, OpenAI via JSON mode, and vLLM via guided_json with strict Schema
  • Web-Search integration with Chat and Document Q/A
  • Agents for Search, Document Q/A, Python Code, CSV frames (Experimental, best with OpenAI currently)
  • Evaluate performance using reward models
  • Quality maintained with over 1000 unit and integration tests taking over 4 GPU-hours

Get Started

GitHub license Linux macOS Windows Docker

To quickly try out h2oGPT with limited document Q/A capability, create a fresh Python 3.10 environment and run:

  • CPU or MAC (M1/M2):
    # for windows/mac use "set" or relevant environment setting mechanism
    export PIP_EXTRA_INDEX_URL="https://download.pytorch.org/whl/cpu"
    
  • Linux/Windows CPU/CUDA/ROC:
    # for windows/mac use "set" or relevant environment setting mechanism
    export PIP_EXTRA_INDEX_URL="https://download.pytorch.org/whl/cu121 https://huggingface.github.io/autogptq-index/whl/cu121"
    # for cu118 use export PIP_EXTRA_INDEX_URL="https://download.pytorch.org/whl/cu118 https://huggingface.github.io/autogptq-index/whl/cu118"
    

Then choose your llama_cpp_python options, by changing CMAKE_ARGS to whichever system you have according to llama_cpp_python backend documentation. E.g. CUDA on Linux:

export LLAMA_CUBLAS=1
export CMAKE_ARGS="-DLLAMA_CUBLAS=on -DCMAKE_CUDA_ARCHITECTURES=all"
export FORCE_CMAKE=1

Note for some reason things will fail with llama_cpp_python if don't add all cuda arches, and building with all those arches does take some time. Windows CUDA:

set CMAKE_ARGS=-DLLAMA_CUBLAS=on -DCMAKE_CUDA_ARCHITECTURES=all
set LLAMA_CUBLAS=1
set FORCE_CMAKE=1

Note for some reason things will fail with llama_cpp_python if don't add all cuda arches, and building with all those arches does take some time. Metal M1/M2:

export CMAKE_ARGS="-DLLAMA_METAL=on"
export FORCE_CMAKE=1

Then run the following commands on any system:

git clone https://github.com/h2oai/h2ogpt.git
cd h2ogpt
pip install -r requirements.txt
pip install -r reqs_optional/requirements_optional_langchain.txt

pip uninstall llama_cpp_python llama_cpp_python_cuda -y
pip install -r reqs_optional/requirements_optional_llamacpp_gpt4all.txt --no-cache-dir

pip install -r reqs_optional/requirements_optional_langchain.urls.txt
# GPL, only run next line if that is ok:
pip install -r reqs_optional/requirements_optional_langchain.gpllike.txt

# choose up to 32768 if have enough GPU memory:
python generate.py --base_model=TheBloke/Mistral-7B-Instruct-v0.2-GGUF --prompt_type=mistral --max_seq_len=4096

Next, go to your browser by visiting http://127.0.0.1:7860 or http://localhost:7860. Choose 13B for a better model than 7B.

We recommend quantized models for most small-GPU systems, e.g. LLaMa-2-7B-Chat-GGUF for 9GB+ GPU memory or larger models like LLaMa-2-13B-Chat-GGUF if you have 16GB+ GPU memory.

See Offline for how to run h2oGPT offline.


Note that for all platforms, some packages such as DocTR, Unstructured, BLIP, Stable Diffusion, etc. download models at runtime that appear to delay operations in the UI. The progress appears in the console logs.

Windows 10/11 64-bit with full document Q/A capability

  • One-Click Installer

    • CPU or GPU: Download h2oGPT Windows Installer (1.3GB file)
      • Once installed, feel free to change start directory for icon from %HOMEDRIVE%\%HOMEPATH% to (e.g.) %HOMEDRIVE%\%HOMEPATH%\h2ogpt_data so all created files (like database) go there. All paths saved are relative to this path.
    • CPU: Click the h2oGPT icon in the Start menu. Give it about 15 seconds to open in a browser if many optional packages are included. By default, the browser will launch with the actual local IP address, not localhost.
    • GPU: Before starting, run the following commands (replace pseud with your user):
      C:\Users\pseud\AppData\Local\Programs\h2oGPT\Python\python.exe -m pip uninstall -y torch
      C:\Users\pseud\AppData\Local\Programs\h2oGPT\Python\python.exe -m pip install https://h2o-release.s3.amazonaws.com/h2ogpt/torch-2.1.2%2Bcu118-cp310-cp310-win_amd64.whl
      
      Now click the h2oGPT icon in the Start menu. Give it about 20 seconds to open in a browser if many optional packages are included. By default, the browser will launch with the actual local IP address, not localhost.
      • Some other users may have python located here: C:\Program Files (x86)\h2oGPT\Python\python.exe.
    • To debug any issues, run the following (replace pseud with your user):
      C:\Users\pseud\AppData\Local\Programs\h2oGPT\Python\python.exe "C:\Users\pseud\AppData\Local\Programs\h2oGPT\h2oGPT.launch.pyw"
      
      Any start-up exceptions are appended to log, e.g. C:\Users\pseud\h2ogpt_exception.log.
  • To control startup, tweak the python startup file, e.g. for user pseud: C:\Users\pseud\AppData\Local\Programs\h2oGPT\pkgs\win_run_app.py

    • In this Python code, set ENVs anywhere before main_h2ogpt() is called
      • E.g. os.environ['name'] = 'value', e.g. os.environ['n_jobs'] = '10' (must be always a string).
    • Environment variables can be changed, e.g.:
      • n_jobs: number of cores for various tasks
      • OMP_NUM_THREADS thread count for LLaMa
      • CUDA_VISIBLE_DEVICES which GPUs are used. Recommend set to single fast GPU, e.g. CUDA_VISIBLE_DEVICES=0 if have multiple GPUs. Note that UI cannot control which GPUs (or CPU mode) for LLaMa models.
      • Any CLI argument from python generate.py --help with environment variable set as h2ogpt_x, e.g. h2ogpt_h2ocolors to False.
      • Set env h2ogpt_server_name to actual IP address for LAN to see app, e.g. h2ogpt_server_name to 192.168.1.172 and allow access through firewall if have Windows Defender activated.
  • One can tweak installed h2oGPT code at, e.g. C:\Users\pseud\AppData\Local\Programs\h2oGPT.

  • To terminate the app, go to System Tab and click Admin and click Shutdown h2oGPT.

    • If startup fails, run as console and check for errors, e.g. and kill any old Python processes.
  • Full Windows 10/11 Manual Installation Script

    • Single .bat file for installation (if you do not skip any optional packages, takes about 9GB filled on disk).
    • Recommend base Conda env, which allows for DocTR that requires pygobject that has otherwise no support (except mysys2 that cannot be used by h2oGPT).
    • Also allows for the TTS package by Coqui, which is otherwise not currently enabled in the one-click installer.

Linux (CPU/CUDA) with full document Q/A capability


macOS (CPU/M1/M2) with full document Q/A capability

  • One-click Installers (Experimental and subject to changes, we haven't tested each and every feature with these installers, we encourage the community to try them and report any issues)

    Mar 07, 2024

    Nov 08, 2023

    Download the runnable file and open it from the Finder. It will take a few minutes to unpack and run the application. These one-click installers are experimental. Report any issues with steps to reproduce at https://github.com/h2oai/h2ogpt/issues.

    Note: The app bundle is unsigned. If you experience any issues with running the app, run the following commands:

    $ xattr -dr com.apple.quarantine {file-path}/h2ogpt-osx-m1-gpu
    $ chmod +x {file-path}/h2ogpt-osx-m1-gpu
    
  • macOS Manual Install and Run Docs


Example Models

GPU mode requires CUDA support via torch and transformers. A 7B/13B model in 16-bit uses 14GB/26GB of GPU memory to store the weights (2 bytes per weight). Compression such as 4-bit precision (bitsandbytes, AWQ, GPTQ, etc.) can further reduce memory requirements down to less than 6GB when asking a question about your documents. (For more information, see low-memory mode.)

CPU mode uses GPT4ALL and LLaMa.cpp, e.g. gpt4all-j, requiring about 14GB of system RAM in typical use.


Live Demos

Inference Benchmarks for Summarization & Generation

Resources

Partners

Video Demo

https://github.com/h2oai/h2ogpt/assets/2249614/2f805035-2c85-42fb-807f-fd0bca79abc6

YouTube 4K version: https://www.youtube.com/watch?v=_iktbj4obAI

Docs Guide

Experimental features

These are not part of normal installation instructions and are experimental.

  • Agents -- in Alpha testing. Optimal for OpenAI, but that also fails sometimes.

Roadmap

  • Integration of code and resulting LLMs with downstream applications and low/no-code platforms
  • Complement h2oGPT chatbot with other APIs like ToolBench
  • Enhance the model's code completion, reasoning, and mathematical capabilities, ensure factual correctness, minimize hallucinations, and avoid repetitive output
  • Add better agents for SQL and CSV question/answer

Development

  • To create a development environment for training and generation, follow the installation instructions.
  • To fine-tune any LLM models on your data, follow the fine-tuning instructions.
  • To run h2oGPT tests:
    pip install requirements-parser pytest-instafail pytest-random-order playsound==1.3.0
    conda install -c conda-forge gst-python -y
    sudo apt-get install gstreamer-1.0
    pip install pygame
    pytest --instafail -s -v tests
    # for openai server test on already-running local server
    pytest -s -v -n 4 openai_server/test_openai_server.py::test_openai_client
    
    or tweak/run tests/test4gpus.sh to run tests in parallel.

Help

Acknowledgements

Why H2O.ai?

Our Makers at H2O.ai have built several world-class Machine Learning, Deep Learning and AI platforms:

We also built platforms for deployment and monitoring, and for data wrangling and governance:

  • H2O MLOps to deploy and monitor models at scale
  • H2O Feature Store in collaboration with AT&T
  • Open-source Low-Code AI App Development Frameworks Wave and Nitro
  • Open-source Python datatable (the engine for H2O Driverless AI feature engineering)

Many of our customers are creating models and deploying them enterprise-wide and at scale in the H2O AI Cloud:

We are proud to have over 25 (of the world's 280) Kaggle Grandmasters call H2O home, including three Kaggle Grandmasters who have made it to world #1.

Disclaimer

Please read this disclaimer carefully before using the large language model provided in this repository. Your use of the model signifies your agreement to the following terms and conditions.

  • Biases and Offensiveness: The large language model is trained on a diverse range of internet text data, which may contain biased, racist, offensive, or otherwise inappropriate content. By using this model, you acknowledge and accept that the generated content may sometimes exhibit biases or produce content that is offensive or inappropriate. The developers of this repository do not endorse, support, or promote any such content or viewpoints.
  • Limitations: The large language model is an AI-based tool and not a human. It may produce incorrect, nonsensical, or irrelevant responses. It is the user's responsibility to critically evaluate the generated content and use it at their discretion.
  • Use at Your Own Risk: Users of this large language model must assume full responsibility for any consequences that may arise from their use of the tool. The developers and contributors of this repository shall not be held liable for any damages, losses, or harm resulting from the use or misuse of the provided model.
  • Ethical Considerations: Users are encouraged to use the large language model responsibly and ethically. By using this model, you agree not to use it for purposes that promote hate speech, discrimination, harassment, or any form of illegal or harmful activities.
  • Reporting Issues: If you encounter any biased, offensive, or otherwise inappropriate content generated by the large language model, please report it to the repository maintainers through the provided channels. Your feedback will help improve the model and mitigate potential issues.
  • Changes to this Disclaimer: The developers of this repository reserve the right to modify or update this disclaimer at any time without prior notice. It is the user's responsibility to periodically review the disclaimer to stay informed about any changes.

By using the large language model provided in this repository, you agree to accept and comply with the terms and conditions outlined in this disclaimer. If you do not agree with any part of this disclaimer, you should refrain from using the model and any content generated by it.

Star History

Star History Chart

最近版本更新:(数据更新于 2024-05-20 16:04:11)

2024-03-14 20:04:05 client-nightly-2024-03-14

2024-03-12 21:20:23 client-nightly-2024-03-12

2024-03-11 22:33:24 client-nightly-2024-03-11

2024-03-10 20:03:31 client-nightly-2024-03-10

2024-03-09 21:01:43 client-nightly-2024-03-09

2024-03-08 21:03:45 client-nightly-2024-03-08

2024-03-07 21:02:19 client-nightly-2024-03-07

2024-03-06 21:02:26 client-nightly-2024-03-06

2024-03-05 21:02:18 client-nightly-2024-03-05

2024-03-04 21:36:36 client-nightly-2024-03-04

主题(topics):

ai, chatgpt, embeddings, generative, gpt, gpt4all, llama2, llm, mixtral, pdf, private, privategpt, vectorstore

h2oai/h2ogpt同语言 Python最近更新仓库

2024-06-21 18:02:49 argilla-io/argilla

2024-06-18 16:26:58 hpcaitech/Open-Sora

2024-06-18 10:45:23 vanna-ai/vanna

2024-06-15 05:16:32 NVIDIA/warp

2024-06-14 10:42:40 1Panel-dev/MaxKB

2024-06-13 22:59:46 mikumifa/biliTickerBuy