AWQ is an efficient, accurate and blazing-fast low-bit weight quantization method, currently supporting 4-bit quantization. Compared to GPTQ, it offers faster Transformers-based inference with equivalent or better quality compared to the most commonly used GPTQ settings.

AWQ models are currently supported on Linux and Windows, with NVidia GPUs only. macOS users: please use GGUF models instead.

It is supported by:

Text Generation Webui - using Loader: AutoAWQ
vLLM - version 0.2.2 or later for support for all model types.
Hugging Face Text Generation Inference (TGI)
Transformers version 4.35.0 and later, from any code or client that supports Transformers
AutoAWQ - for use from Python code

Repositories available

Prompt template: Alpaca

Below is an instruction that describes a task. Write a response that appropriately completes the request.

### Instruction:
{prompt}

### Response:

Provided files, and AWQ parameters

I currently release 128g GEMM models only. The addition of group_size 32 models, and GEMV kernel models, is being actively considered.

Models are released as sharded safetensors files.

Branch	Bits	GS	AWQ Dataset	Seq Len	Size
main	4	128	VMware Open Instruct	4096	4.15 GB

How to easily download and use this model in text-generation-webui

Please make sure you're using the latest version of text-generation-webui.

It is strongly recommended to use the text-generation-webui one-click-installers unless you're sure you know how to make a manual install.

Click the Model tab.
Under Download custom model or LoRA, enter TheBloke/Sonya-7B-AWQ.
Click Download.
The model will start downloading. Once it's finished it will say "Done".
In the top left, click the refresh icon next to Model.
In the Model dropdown, choose the model you just downloaded: Sonya-7B-AWQ
Select Loader: AutoAWQ.
Click Load, and the model will load and is now ready for use.
If you want any custom settings, set them and then click Save settings for this model followed by Reload the Model in the top right.
Once you're ready, click the Text Generation tab and enter a prompt to get started!

Multi-user inference server: vLLM

Documentation on installing and using vLLM can be found here.

Please ensure you are using vLLM version 0.2 or later.
When using vLLM as a server, pass the --quantization awq parameter.

For example:

python3 -m vllm.entrypoints.api_server --model TheBloke/Sonya-7B-AWQ --quantization awq --dtype auto

When using vLLM from Python code, again set quantization=awq.

For example:

python

from vllm import LLM, SamplingParams

prompts = [
    "Tell me about AI",
    "Write a story about llamas",
    "What is 291 - 150?",
    "How much wood would a woodchuck chuck if a woodchuck could chuck wood?",
]
prompt_template=f'''Below is an instruction that describes a task. Write a response that appropriately completes the request.

### Instruction:
{prompt}

### Response:
'''

prompts = [prompt_template.format(prompt=prompt) for prompt in prompts]

sampling_params = SamplingParams(temperature=0.8, top_p=0.95)

llm = LLM(model="TheBloke/Sonya-7B-AWQ", quantization="awq", dtype="auto")

outputs = llm.generate(prompts, sampling_params)

# Print the outputs.
for output in outputs:
    prompt = output.prompt
    generated_text = output.outputs[0].text
    print(f"Prompt: {prompt!r}, Generated text: {generated_text!r}")

Multi-user inference server: Hugging Face Text Generation Inference (TGI)

Use TGI version 1.1.0 or later. The official Docker container is: ghcr.io/huggingface/text-generation-inference:1.1.0

Example Docker parameters:

--model-id TheBloke/Sonya-7B-AWQ --port 3000 --quantize awq --max-input-length 3696 --max-total-tokens 4096 --max-batch-prefill-tokens 4096

Example Python code for interfacing with TGI (requires huggingface-hub 0.17.0 or later):

pip3 install huggingface-hub

python

from huggingface_hub import InferenceClient

endpoint_url = "https://your-endpoint-url-here"

prompt = "Tell me about AI"
prompt_template=f'''Below is an instruction that describes a task. Write a response that appropriately completes the request.

### Instruction:
{prompt}

### Response:
'''

client = InferenceClient(endpoint_url)
response = client.text_generation(prompt,
                                  max_new_tokens=128,
                                  do_sample=True,
                                  temperature=0.7,
                                  top_p=0.95,
                                  top_k=40,
                                  repetition_penalty=1.1)

print(f"Model output: ", response)

Inference from Python code using Transformers

Install the necessary packages

Requires: Transformers 4.35.0 or later.
Requires: AutoAWQ 0.1.6 or later.

pip3 install --upgrade "autoawq>=0.1.6" "transformers>=4.35.0"

Note that if you are using PyTorch 2.0.1, the above AutoAWQ command will automatically upgrade you to PyTorch 2.1.0.

If you are using CUDA 11.8 and wish to continue using PyTorch 2.0.1, instead run this command:

pip3 install https://github.com/casper-hansen/AutoAWQ/releases/download/v0.1.6/autoawq-0.1.6+cu118-cp310-cp310-linux_x86_64.whl

If you have problems installing AutoAWQ using the pre-built wheels, install it from source instead:

shell

pip3 uninstall -y autoawq
git clone https://github.com/casper-hansen/AutoAWQ
cd AutoAWQ
pip3 install .

Transformers example code (requires Transformers 4.35.0 and later)

python

from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer

model_name_or_path = "TheBloke/Sonya-7B-AWQ"

tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)
model = AutoModelForCausalLM.from_pretrained(
    model_name_or_path,
    low_cpu_mem_usage=True,
    device_map="cuda:0"
)

# Using the text streamer to stream output one token at a time
streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)

prompt = "Tell me about AI"
prompt_template=f'''Below is an instruction that describes a task. Write a response that appropriately completes the request.

### Instruction:
{prompt}

### Response:
'''

# Convert prompt to tokens
tokens = tokenizer(
    prompt_template,
    return_tensors='pt'
).input_ids.cuda()

generation_params = {
    "do_sample": True,
    "temperature": 0.7,
    "top_p": 0.95,
    "top_k": 40,
    "max_new_tokens": 512,
    "repetition_penalty": 1.1
}

# Generate streamed output, visible one token at a time
generation_output = model.generate(
    tokens,
    streamer=streamer,
    **generation_params
)

# Generation without a streamer, which will include the prompt in the output
generation_output = model.generate(
    tokens,
    **generation_params
)

# Get the tokens from the output, decode them, print them
token_output = generation_output[0]
text_output = tokenizer.decode(token_output)
print("model.generate output: ", text_output)

# Inference is also possible via Transformers' pipeline
from transformers import pipeline

pipe = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
    **generation_params
)

pipe_output = pipe(prompt_template)[0]['generated_text']
print("pipeline output: ", pipe_output)

Compatibility

The files provided are tested to work with:

text-generation-webui using Loader: AutoAWQ.
vLLM version 0.2.0 and later.
Hugging Face Text Generation Inference (TGI) version 1.1.0 and later.
Transformers version 4.35.0 and later.
AutoAWQ version 0.1.1 and later.

Discord

For further support, and discussions on these models and AI in general, join us at:

TheBloke AI's Discord server

Thanks, and how to contribute

Thanks to the chirper.ai team!

Thanks to Clay from gpus.llm-utils.org!

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: https://patreon.com/TheBlokeAI
Ko-Fi: https://ko-fi.com/TheBlokeAI

Special thanks to: Aemon Algiz.

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.

Original model card: Sanji Watsuki's Sonya 7B

Top 1 Performer MT-bench 🤪

WTF is This?

Sonya-7B is, at the time of writing, the #1 performing model in MT-Bench first turn, ahead of GPT-4, and overall the #2 model in MT-Bench, to the best of my knowledge. Sonya-7B should be a good all-purpose model for all tasks including assistant, RP, etc.

Sonya-7B has a similar structure to my previous model, Silicon-Maid-7B, and uses a very similar merge. It's a merge of xDAN-AI/xDAN-L1-Chat-RL-v1, Jan-Ai's Stealth v1.2, chargoddard/piano-medley-7b, NeverSleep/Noromaid-7B-v0.2, and athirdpath/NSFW_DPO_vmgb-7b. Sauce is below. Somehow, by combining these pieces, it substantially outscores any of its parents on MT-Bench.

I picked these models because:

MT-Bench normally correlates well with real world model quality and xDAN performs well on it.
Almost all models in the mix were Alpaca prompt formatted which gives prompt consistency.
Stealth v1.2 has been a magic sprinkle that seems to increase my MT-Bench scores.
I added RP models because it boosted the Writing and Roleplay benchmarks 👀

Based on the parent models, I expect this model to be used with an 8192 context window. Please use NTK scaling alpha of 2.6 to experimentally try out 16384 context.

Let me be candid: Despite the test scores, this model is NOT is a GPT killer. I think it's a very sharp model for a 7B, it probably punches way above its weight for a 7B, but it's still a 7B model. Even for a 7B model, I think it's quirky and has some weird outputs, probably due to how Frankenstein this merge is. Keep your expectations in check 😉

MT-Bench Average Turn

model	score	size
gpt-4	8.99	-
Sonya-7B	8.52	7b
xDAN-L1-Chat-RL-v1	8.34	7b
Starling-7B	8.09	7b
Claude-2	8.06	-
Silicon-Maid	7.96	7b
Loyal-Macaroni-Maid	7.95	7b
gpt-3.5-turbo	7.94	20b?
Claude-1	7.90	-
OpenChat-3.5	7.81	-
vicuna-33b-v1.3	7.12	33b
wizardlm-30b	7.01	30b
Llama-2-70b-chat	6.86	70b

The Sauce

models:
  - model: xDAN-AI/xDAN-L1-Chat-RL-v1
    parameters:
      weight: 1
      density: 1
  - model: chargoddard/piano-medley-7b
    parameters:
      weight: 0.3
  - model: jan-hq/stealth-v1.2
    parameters:
      weight: 0.2
  - model: NeverSleep/Noromaid-7b-v0.2
    parameters:
      weight: 0.2
  - model: athirdpath/NSFW_DPO_vmgb-7b
    parameters:
      weight: 0.2
merge_method: ties
base_model: mistralai/Mistral-7B-v0.1
parameters:
  density: 0.4
  int8_mask: true
  normalize: true
dtype: bfloat16

There was no additional training, finetuning, or DPO. This is a straight merger.

Prompt Template (Alpaca)

Below is an instruction that describes a task. Write a response that appropriately completes the request.

### Instruction:
{prompt}

### Response:

I found that this model performed worse with the xDAN prompt format so, despite the heavy weight of xDAN in this merger, I recommeend against its use.

Other Benchmark Stuff

########## First turn ##########

model	turn	score	size
Sonya-7B	1	9.06875	7b
gpt-4	1	8.95625	-
xDAN-L1-Chat-RL-v1	1	8.87500	7b
xDAN-L2-Chat-RL-v2	1	8.78750	30b
claude-v1	1	8.15000	-
gpt-3.5-turbo	1	8.07500	20b
vicuna-33b-v1.3	1	7.45625	33b
wizardlm-30b	1	7.13125	30b
oasst-sft-7-llama-30b	1	7.10625	30b
Llama-2-70b-chat	1	6.98750	70b

########## Second turn ##########

model	turn	score	size
gpt-4	2	9.025000	-
xDAN-L2-Chat-RL-v2	2	8.087500	30b
Sonya-7B	2	7.962500	7b
xDAN-L1-Chat-RL-v1	2	7.825000	7b
gpt-3.5-turbo	2	7.812500	20b
claude-v1	2	7.650000	-
wizardlm-30b	2	6.887500	30b
vicuna-33b-v1.3	2	6.787500	33b
Llama-2-70b-chat	2	6.725000	70b

If you'd like to replicate the MT-Bench run, please ensure that the Alpaca prompt template is applied to the model. I did this by putting "alpaca" in the model path to trigger the AlpacaAdapter.