GGUF is a new format introduced by the llama.cpp team on August 21st 2023. It is a replacement for GGML, which is no longer supported by llama.cpp.
Here is an incomplete list of clients and libraries that are known to support GGUF:
llama.cpp. The source project for GGUF. Offers a CLI and a server option.
text-generation-webui, the most widely used web UI, with many features and powerful extensions. Supports GPU acceleration.
KoboldCpp, a fully featured web UI, with GPU accel across all platforms and GPU architectures. Especially good for story telling.
GPT4All, a free and open source local running GUI, supporting Windows, Linux and macOS with full GPU accel.
LM Studio, an easy-to-use and powerful local GUI for Windows and macOS (Silicon), with GPU acceleration. Linux available, in beta as of 27/11/2023.
LoLLMS Web UI, a great web UI with many interesting and unique features, including a full model library for easy model selection.
Faraday.dev, an attractive and easy to use character-based chat GUI for Windows and macOS (both Silicon and Intel), with GPU acceleration.
llama-cpp-python, a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
candle, a Rust ML framework with a focus on performance, including GPU support, and ease of use.
ctransformers, a Python library with GPU accel, LangChain support, and OpenAI-compatible AI server. Note, as of time of writing (November 27th 2023), ctransformers has not been updated in a long time and does not support many recent models.
Below is an instruction that describes a task. Write a response that appropriately completes the request.
### Instruction:
{prompt}
### Response:
Compatibility
These quantised GGUFv2 files are compatible with llama.cpp from August 27th onwards, as of commit d0cee0d
They are also compatible with many third party UIs and libraries - please see the list at the top of this README.
Explanation of quantisation methods
Click to see details
The new methods available are:
GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
Refer to the Provided Files table below to see what files use which methods, and how.
Provided files
Name
Quant method
Bits
Size
Max RAM required
Use case
venus-120b-v1.1.Q2_K.gguf
Q2_K
2
50.71 GB
53.21 GB
smallest, significant quality loss - not recommended for most purposes
venus-120b-v1.1.Q3_K_S.gguf
Q3_K_S
3
51.81 GB
54.31 GB
very small, high quality loss
venus-120b-v1.1.Q3_K_M.gguf
Q3_K_M
3
57.64 GB
60.14 GB
very small, high quality loss
venus-120b-v1.1.Q3_K_L.gguf
Q3_K_L
3
63.01 GB
65.51 GB
small, substantial quality loss
venus-120b-v1.1.Q4_0.gguf
Q4_0
4
67.75 GB
70.25 GB
legacy; small, very high quality loss - prefer using Q3_K_M
venus-120b-v1.1.Q4_K_S.gguf
Q4_K_S
4
67.88 GB
70.38 GB
small, greater quality loss
venus-120b-v1.1.Q4_K_M.gguf
Q4_K_M
4
72.15 GB
74.65 GB
medium, balanced quality - recommended
venus-120b-v1.1.Q5_0.gguf
Q5_0
5
82.76 GB
85.26 GB
legacy; medium, balanced quality - prefer using Q4_K_M
venus-120b-v1.1.Q5_K_S.gguf
Q5_K_S
5
82.76 GB
85.26 GB
large, low quality loss - recommended
venus-120b-v1.1.Q5_K_M.gguf
Q5_K_M
5
85.02 GB
87.52 GB
large, very low quality loss - recommended
venus-120b-v1.1.Q6_K.gguf
Q6_K
6
98.71 GB
101.21 GB
very large, extremely low quality loss
venus-120b-v1.1.Q8_0.gguf
Q8_0
8
127.84 GB
130.34 GB
very large, extremely low quality loss - not recommended
Note: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
Q6_K and Q8_0 files are split and require joining
Note: HF does not support uploading files larger than 50GB. Therefore I have uploaded the Q6_K and Q8_0 files as split files.
Click for instructions regarding Q6_K and Q8_0 files
COPY /B venus-120b-v1.1.Q6_K.gguf-split-a + venus-120b-v1.1.Q6_K.gguf-split-b venus-120b-v1.1.Q6_K.gguf
del venus-120b-v1.1.Q6_K.gguf-split-a venus-120b-v1.1.Q6_K.gguf-split-b
COPY /B venus-120b-v1.1.Q8_0.gguf-split-a + venus-120b-v1.1.Q8_0.gguf-split-b venus-120b-v1.1.Q8_0.gguf
del venus-120b-v1.1.Q8_0.gguf-split-a venus-120b-v1.1.Q8_0.gguf-split-b
How to download GGUF files
Note for manual downloaders: You almost never want to clone the entire repo! Multiple different quantisation formats are provided, and most users only want to pick and download a single file.
The following clients/libraries will automatically download models for you, providing a list of available models to choose from:
LM Studio
LoLLMS Web UI
Faraday.dev
In text-generation-webui
Under Download Model, you can enter the model repo: TheBloke/Venus-120b-v1.1-GGUF and below it, a specific filename to download, such as: venus-120b-v1.1.Q4_K_M.gguf.
Then click Download.
On the command line, including multiple files at once
I recommend using the huggingface-hub Python library:
pip3 install huggingface-hub
Then you can download any individual model file to the current directory, at high speed, with a command like this:
Windows Command Line users: You can set the environment variable by running set HF_HUB_ENABLE_HF_TRANSFER=1 before the download command.
Example llama.cpp command
Make sure you are using llama.cpp from commit d0cee0d or later.
./main -ngl 35 -m venus-120b-v1.1.Q4_K_M.gguf --color -c 4096 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n### Instruction:\n{prompt}\n\n### Response:"
Change -ngl 32 to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
Change -c 4096 to the desired sequence length. For extended sequence models - eg 8K, 16K, 32K - the necessary RoPE scaling parameters are read from the GGUF file and set by llama.cpp automatically. Note that longer sequence lengths require much more resources, so you may need to reduce this value.
If you want to have a chat-style conversation, replace the -p <PROMPT> argument with -i -ins
You can use GGUF models from Python using the llama-cpp-python or ctransformers libraries. Note that at the time of writing (Nov 27th 2023), ctransformers has not been updated for some time and is not compatible with some recent models. Therefore I recommend you use llama-cpp-python.
How to load this model in Python code, using llama-cpp-python
Run one of the following commands, according to your system:
shell
1# Base ctransformers with no GPU acceleration2pip install llama-cpp-python
3# With NVidia CUDA acceleration4CMAKE_ARGS="-DLLAMA_CUBLAS=on" pip install llama-cpp-python
5# Or with OpenBLAS acceleration6CMAKE_ARGS="-DLLAMA_BLAS=ON -DLLAMA_BLAS_VENDOR=OpenBLAS" pip install llama-cpp-python
7# Or with CLBLast acceleration8CMAKE_ARGS="-DLLAMA_CLBLAST=on" pip install llama-cpp-python
9# Or with AMD ROCm GPU acceleration (Linux only)10CMAKE_ARGS="-DLLAMA_HIPBLAS=on" pip install llama-cpp-python
11# Or with Metal GPU acceleration for macOS systems only12CMAKE_ARGS="-DLLAMA_METAL=on" pip install llama-cpp-python
1314# In windows, to set the variables CMAKE_ARGS in PowerShell, follow this format; eg for NVidia CUDA:15$env:CMAKE_ARGS ="-DLLAMA_OPENBLAS=on"16pip install llama-cpp-python
Simple llama-cpp-python example code
python
1from llama_cpp import Llama
23# Set gpu_layers to the number of layers to offload to GPU. Set to 0 if no GPU acceleration is available on your system.4llm = Llama(5 model_path="./venus-120b-v1.1.Q4_K_M.gguf",# Download the model file first6 n_ctx=4096,# The max sequence length to use - note that longer sequence lengths require much more resources7 n_threads=8,# The number of CPU threads to use, tailor to your system and the resulting performance8 n_gpu_layers=35# The number of layers to offload to GPU, if you have GPU acceleration available9)1011# Simple inference example12output = llm(13"Below is an instruction that describes a task. Write a response that appropriately completes the request.\n\n### Instruction:\n{prompt}\n\n### Response:",# Prompt14 max_tokens=512,# Generate up to 512 tokens15 stop=["</s>"],# Example stop token - not necessarily correct for this specific model! Please check before using.16 echo=True# Whether to echo the prompt17)1819# Chat Completion API2021llm = Llama(model_path="./venus-120b-v1.1.Q4_K_M.gguf", chat_format="llama-2")# Set chat_format according to the model you are using22llm.create_chat_completion(23 messages =[24{"role":"system","content":"You are a story writing assistant."},25{26"role":"user",27"content":"Write a story about llamas."28}29]30)
How to use with LangChain
Here are guides on using llama-cpp-python and ctransformers with LangChain:
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
Patreon special mentions: Michael Levine, 阿明, Trailburnt, Nikolai Manek, John Detwiler, Randy H, Will Dee, Sebastain Graf, NimbleBox.ai, Eugene Pentland, Emad Mostaque, Ai Maven, Jim Angel, Jeff Scroggin, Michael Davis, Manuel Alberto Morcote, Stephen Murray, Robert, Justin Joy, Luke @flexchar, Brandon Frisco, Elijah Stavena, S_X, Dan Guido, Undi ., Komninos Chatzipapas, Shadi, theTransient, Lone Striker, Raven Klaugh, jjj, Cap'n Zoog, Michel-Marie MAUDET (LINAGORA), Matthew Berman, David, Fen Risland, Omer Bin Jawed, Luke Pendergrass, Kalila, OG, Erik Bjäreholt, Rooh Singh, Joseph William Delisle, Dan Lewis, TL, John Villwock, AzureBlack, Brad, Pedro Madruga, Caitlyn Gatomon, K, jinyuan sun, Mano Prime, Alex, Jeffrey Morgan, Alicia Loh, Illia Dulskyi, Chadd, transmissions 11, fincy, Rainer Wilmers, ReadyPlayerEmma, knownsqashed, Mandus, biorpg, Deo Leter, Brandon Phillips, SuperWojo, Sean Connelly, Iucharbius, Jack West, Harry Royden McLaughlin, Nicholas, terasurfer, Vitor Caleffi, Duane Dunston, Johann-Peter Hartmann, David Ziegler, Olakabola, Ken Nordquist, Trenton Dambrowitz, Tom X Nguyen, Vadim, Ajan Kanaga, Leonard Tan, Clay Pascal, Alexandros Triantafyllidis, JM33133, Xule, vamX, ya boyyy, subjectnull, Talal Aujan, Alps Aficionado, wassieverse, Ari Malik, James Bentley, Woland, Spencer Kim, Michael Dempsey, Fred von Graf, Elle, zynix, William Richards, Stanislav Ovsiannikov, Edmond Seymore, Jonathan Leane, Martin Kemka, usrbinkat, Enrico Ros
Thank you to all my generous patrons and donaters!
And thank you again to a16z for their generous grant.
The resulting model has 140 layers and approximately 122 billion parameters.
See mergekit-config.yml for details on the merge method used.
See the exl2-* branches for exllama2 quantizations. The 4.85 bpw quant should fit in 80GB VRAM, and the 3.0 bpw quant should (just barely) fit in 48GB VRAM with 4k context.
