Original llama.cpp quant methods: q4_0, q4_1, q5_0, q5_1, q8_0
I have quantized these 'original' quantisation methods using an older version of llama.cpp so that they remain compatible with llama.cpp as of May 19th, commit 2d5db48.
They should be compatible with all current UIs and libraries that use llama.cpp, such as those listed at the top of this README.
These new quantisation methods are only compatible with llama.cpp as of June 6th, commit 2d43387.
They will NOT be compatible with koboldcpp, text-generation-ui, and other UIs and libraries yet. Support is expected to come over the next few days.
Explanation of the new k-quant methods
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
GGML_TYPE_Q8_K - "type-0" 8-bit quantization. Only used for quantizing intermediate results. The difference to the existing Q8_0 is that the block size is 256. All 2-6 bit dot products are implemented for this quantization type.
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
VicUnlocked-Alpaca-65B.ggmlv3.q4_0.bin
q4_0
4
36.73 GB
39.23 GB
Original llama.cpp quant method, 4-bit.
VicUnlocked-Alpaca-65B.ggmlv3.q4_1.bin
q4_1
4
40.81 GB
43.31 GB
Original llama.cpp quant method, 4-bit. Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models.
VicUnlocked-Alpaca-65B.ggmlv3.q5_0.bin
q5_0
5
44.89 GB
47.39 GB
Original llama.cpp quant method, 5-bit. Higher accuracy, higher resource usage and slower inference.
VicUnlocked-Alpaca-65B.ggmlv3.q5_1.bin
q5_1
5
48.97 GB
51.47 GB
Original llama.cpp quant method, 5-bit. Even higher accuracy, resource usage and slower inference.
VicUnlocked-Alpaca-65B.ggmlv3.q8_0.bin
q8_0
8
69.37 GB
71.87 GB
8-bit. Almost indistinguishable from float16. Huge resource use and slow. Not recommended for most use cases.
vicunlocked-65b.ggmlv3.q2_K.bin
q2_K
2
27.33 GB
29.83 GB
New k-quant method. Uses GGML_TYPE_Q4_K for the attention.vw and feed_forward.w2 tensors, GGML_TYPE_Q2_K for the other tensors.
vicunlocked-65b.ggmlv3.q3_K_L.bin
q3_K_L
3
34.55 GB
37.05 GB
New k-quant method. Uses GGML_TYPE_Q5_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K
vicunlocked-65b.ggmlv3.q3_K_M.bin
q3_K_M
3
31.40 GB
33.90 GB
New k-quant method. Uses GGML_TYPE_Q4_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K
vicunlocked-65b.ggmlv3.q3_K_S.bin
q3_K_S
3
28.06 GB
30.56 GB
New k-quant method. Uses GGML_TYPE_Q3_K for all tensors
vicunlocked-65b.ggmlv3.q4_K_M.bin
q4_K_M
4
39.28 GB
41.78 GB
New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q4_K
vicunlocked-65b.ggmlv3.q4_K_S.bin
q4_K_S
4
36.73 GB
39.23 GB
New k-quant method. Uses GGML_TYPE_Q4_K for all tensors
vicunlocked-65b.ggmlv3.q5_K_M.bin
q5_K_M
5
46.20 GB
48.70 GB
New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q5_K
vicunlocked-65b.ggmlv3.q5_K_S.bin
q5_K_S
5
44.89 GB
47.39 GB
New k-quant method. Uses GGML_TYPE_Q5_K for all tensors
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.
q8_0 file requires expansion from archive
Note: HF does not support uploading files larger than 50GB. Therefore I have uploaded the q8_0 file in a multi-part ZIP file. The ZIP is not compressed, it is just storing the .bin file in two parts.
To decompress it, please download
VicUnlocked-Alpaca-65B.ggmlv3.q8_0.zip
VicUnlocked-Alpaca-65B.ggmlv3.q8_0.z01
and extract the .zip archive. This will will expand both parts automatically. On Linux I found I had to use 7zip - the basic unzip tool did not work. Example:
sudo apt update -y && sudo apt install 7zip
7zz x VicUnlocked-Alpaca-65B.ggmlv3.q8_0.zip # Once the q8_0.bin is extracted you can delete the .zip and .z01
How to run in llama.cpp
I use the following command line; adjust for your tastes and needs:
./main -t 10 -ngl 32 -m vicunlocked-65b.ggmlv3.q5_0.bin --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "### Instruction: Write a story about llamas\n### Response:"
Change -t 10 to the number of physical CPU cores you have. For example if your system has 8 cores/16 threads, use -t 8.
Change -ngl 32 to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.
If you want to have a chat-style conversation, replace the -p <PROMPT> argument with -i -ins
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.
Special thanks to: Luke from CarbonQuill, Aemon Algiz, Dmitriy Samsonov.
Patreon special mentions: Ajan Kanaga, Kalila, Derek Yates, Sean Connelly, Luke, Nathan LeClaire, Trenton Dambrowitz, Mano Prime, David Flickinger, vamX, Nikolai Manek, senxiiz, Khalefa Al-Ahmad, Illia Dulskyi, trip7s trip, Jonathan Leane, Talal Aujan, Artur Olbinski, Cory Kujawski, Joseph William Delisle, Pyrater, Oscar Rangel, Lone Striker, Luke Pendergrass, Eugene Pentland, Johann-Peter Hartmann.
Thank you to all my generous patrons and donaters!
Original model card: Aeala's VicUnlocked Alpaca 65B QLoRA
LoRA Info:
Please note that this is a highly experimental LoRA model. It may do some good stuff, it might do some undesirable stuff. Training is paused for now. Feel free to try it!~
Important Note: While this is trained on a cleaned ShareGPT dataset like Vicuna used, this was trained in the Alpaca format, so prompting should be something like:
### Instruction:
<prompt> (without the <>)
### Response:
Current upload: checkpoint of a retrain at ~1000 steps with fixed QLoRA repo. (6/4/2023)
