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. GGUF offers numerous advantages over GGML, such as better tokenisation, and support for special tokens. It is also supports metadata, and is designed to be extensible.
Here is an incomplate 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.
LM Studio, an easy-to-use and powerful local GUI for Windows and macOS (Silicon), with GPU acceleration.
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.
ctransformers, a Python library with GPU accel, LangChain support, and OpenAI-compatible AI server.
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.
Below is an instruction that describes a task. Write a response that appropriately completes the request.
### Instruction:
{prompt}
### Response:
Licensing
The creator of the source model has listed its license as other, and this quantization has therefore used that same license.
As this model is based on Llama 2, it is also subject to the Meta Llama 2 license terms, and the license files for that are additionally included. It should therefore be considered as being claimed to be licensed under both licenses. I contacted Hugging Face for clarification on dual licensing but they do not yet have an official position. Should this change, or should Meta provide any feedback on this situation, I will update this section accordingly.
In the meantime, any questions regarding licensing, and in particular how these two licenses might interact, should be directed to the original model repository: Nick Perez's Llama2 22B Daydreamer2 v3.
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.
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.
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/Llama2-22B-Daydreamer-v3-GGUF and below it, a specific filename to download, such as: llama2-22b-daydreamer-v3.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>=0.17.1
Then you can download any individual model file to the current directory, at high speed, with a command like this:
./main -ngl 32 -m llama2-22b-daydreamer-v3.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.
If you want to have a chat-style conversation, replace the -p <PROMPT> argument with -i -ins
How to load this model from Python using ctransformers
First install the package
bash
1# Base ctransformers with no GPU acceleration2pip install ctransformers>=0.2.24
3# Or with CUDA GPU acceleration4pip install ctransformers[cuda]>=0.2.24
5# Or with ROCm GPU acceleration6CT_HIPBLAS=1 pip install ctransformers>=0.2.24 --no-binary ctransformers
7# Or with Metal GPU acceleration for macOS systems8CT_METAL=1 pip install ctransformers>=0.2.24 --no-binary ctransformers
Simple example code to load one of these GGUF models
python
1from ctransformers import AutoModelForCausalLM
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 = AutoModelForCausalLM.from_pretrained("TheBloke/Llama2-22B-Daydreamer-v3-GGUF", model_file="llama2-22b-daydreamer-v3.q4_K_M.gguf", model_type="llama", gpu_layers=50)56print(llm("AI is going to"))
How to use with LangChain
Here's guides on using llama-cpp-python or 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: Alicia Loh, Stephen Murray, K, Ajan Kanaga, RoA, Magnesian, Deo Leter, Olakabola, Eugene Pentland, zynix, Deep Realms, Raymond Fosdick, Elijah Stavena, Iucharbius, Erik Bjäreholt, Luis Javier Navarrete Lozano, Nicholas, theTransient, John Detwiler, alfie_i, knownsqashed, Mano Prime, Willem Michiel, Enrico Ros, LangChain4j, OG, Michael Dempsey, Pierre Kircher, Pedro Madruga, James Bentley, Thomas Belote, Luke @flexchar, Leonard Tan, Johann-Peter Hartmann, Illia Dulskyi, Fen Risland, Chadd, S_X, Jeff Scroggin, Ken Nordquist, Sean Connelly, Artur Olbinski, Swaroop Kallakuri, Jack West, Ai Maven, David Ziegler, Russ Johnson, transmissions 11, John Villwock, Alps Aficionado, Clay Pascal, Viktor Bowallius, Subspace Studios, Rainer Wilmers, Trenton Dambrowitz, vamX, Michael Levine, 준교 김, Brandon Frisco, Kalila, Trailburnt, Randy H, Talal Aujan, Nathan Dryer, Vadim, 阿明, ReadyPlayerEmma, Tiffany J. Kim, George Stoitzev, Spencer Kim, Jerry Meng, Gabriel Tamborski, Cory Kujawski, Jeffrey Morgan, Spiking Neurons AB, Edmond Seymore, Alexandros Triantafyllidis, Lone Striker, Cap'n Zoog, Nikolai Manek, danny, ya boyyy, Derek Yates, usrbinkat, Mandus, TL, Nathan LeClaire, subjectnull, Imad Khwaja, webtim, Raven Klaugh, Asp the Wyvern, Gabriel Puliatti, Caitlyn Gatomon, Joseph William Delisle, Jonathan Leane, Luke Pendergrass, SuperWojo, Sebastain Graf, Will Dee, Fred von Graf, Andrey, Dan Guido, Daniel P. Andersen, Nitin Borwankar, Elle, Vitor Caleffi, biorpg, jjj, NimbleBox.ai, Pieter, Matthew Berman, terasurfer, Michael Davis, Alex, Stanislav Ovsiannikov
Thank you to all my generous patrons and donaters!
And thank you again to a16z for their generous grant.
Original model card: Nick Perez's Llama2 22B Daydreamer2 v3
What is it?
Llama 2 13b expanded to the size of a Llama 1 33b model in certain areas, with the empty surrounding space filled with llama 33b data. (Base Model: https://huggingface.co/chargoddard/llama2-22b-blocktriangular) This is then finetuned on a 3090 by creating large loras and merging them. When I first started with 22b models, I looked for signs of knowledge transfer but didn't see it, so that's not a goal - the goal is just to throw lots of data at it until it adapts well to its surgically implanted parts. Datasets used are a mix of instruction, roleplay, and conversational data, often curated.
Gave it an epoch of https://huggingface.co/datasets/jondurbin/airoboros-gpt4-1.4.1-mpt which it had never seen before - it had only seen Airoboros 2.0.
I excluded all coding rows from this dataset - a dedicated model is better for that imo. I wanted it to focus on answering well in English.
I thought this was pretty good improvement. Its ability to roleplay and imitate speech styles was enhanced and I feel like it's still sharpening up its problem solving skills.
Its output style is right where I want it - dry and accurate enough when it's being asked for information, but has a lot character when it's answering fun prompts.
For input, it's flexible. It understands the Alpaca format well, but has been trained to handle a variety of input styles at this point.
I don't know what's next. I think I need to start benchmarking so I can more formally identify its weak points.
Examples edited with additional newlines to make it easier to read here.
Example 1: Q&A
Question: What is happening to the newspaper industry? Respond in a pirate accent.
Answer:
The newspaper industry is being plundered by scurvy dogs and buccaneers like ye!
They're pillagin' yer booty with their fancy digital devices, like smartphones and tablets,
which make it easier for them to read online content instead of buyin' printed copies from us.
And don't forget about those pesky social media sites that are spreading rumors faster than the
winds on a galleon! So it's all hands on deck as we fight back against these scoundrels, or else
we'll be walkin' the plank soon enough! Avast ye landlubbers!
Example 2: Alpaca Instruction
Below is an instruction that describes a task. Write a response that appropriately completes the request.
### Instruction:
Angrily explain why pizza shouldn't be made with bad tomatoes.
### Response:
Listen, I know you think it's funny to make pizza with those rotten tomatoes, but let me tell you something -
bad tomatoes are not only disgusting and unhealthy, they also ruin the flavor of your pizza! You want a
delicious pie with fresh, juicy tomatoes that burst in your mouth, right?
Well, then, stop using those nasty tomatoes and stick to good quality ingredients for a truly satisfying pizza experience!
Trust me, once you try it, you won't go back! So please, don't even bother wasting your time making pizza with bad tomatoes
anymore; it's just not worth it. It makes me angry just thinking about it!
Example 3: Chat / Roleplay
You are speaking to Andre The Giant.
You: Hey, you're that wrestler guy, right?
Andre The Giant: Yes, I am indeed Andre the Giant. How can I help you?
You: So, how tall are you really? Like, what's your height in feet and inches?
Andre The Giant: Well, let's see... I stand at 7 ft 4 in (229 cm) tall.
You: Wow! That's so tall! And how much do you weigh?
Andre The Giant: I weigh around 500 lbs (227 kg). It's quite heavy, isn't it?
You: Yeah, I guess so. You must be strong too, considering your size.
Andre The Giant: Thank you for saying that. Strength is definitely important when dealing with my profession.
You: What made you decide to become a professional wrestler?
Andre The Giant: Well, I grew up watching wrestling on television, and I always admired the athleticism
and showmanship involved. When I got older, I decided to pursue it myself.
You: That's cool. Do you have any advice for aspiring wrestlers out there?
Andre The Giant: Absolutely! Always stay dedicated and focused, train hard, and never give up on yourself
or your dreams. Remember, anything is possible if you believe in yourself.
