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.

It is supported by:

Text Generation Webui - using Loader: AutoAWQ
vLLM - Llama and Mistral models only
Hugging Face Text Generation Inference (TGI)
AutoAWQ - for use from Python code

Repositories available

Prompt template: Inkbot

<#meta#>
- Date: [DATE]
- Task: [TASK TYPE]
<#system#>
{system_message}
<#chat#>
<#user#>
{prompt}
<#bot#>

Provided files, and AWQ parameters

For my first release of AWQ models, I am releasing 128g models only. I will consider adding 32g as well if there is interest, and once I have done perplexity and evaluation comparisons, but at this time 32g models are still not fully tested with AutoAWQ and vLLM.

Models are released as sharded safetensors files.

Branch	Bits	GS	AWQ Dataset	Seq Len	Size
main	4	128	wikitext	4096	7.25 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/Inkbot-13B-8k-0.2-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: Inkbot-13B-8k-0.2-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 python -m vllm.entrypoints.api_server --model TheBloke/Inkbot-13B-8k-0.2-AWQ --quantization awq

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'''<#meta#>
- Date: [DATE]
- Task: [TASK TYPE]
<#system#>
{system_message}
<#chat#>
<#user#>
{prompt}
<#bot#>
'''

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

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

llm = LLM(model="TheBloke/Inkbot-13B-8k-0.2-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/Inkbot-13B-8k-0.2-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'''<#meta#>
- Date: [DATE]
- Task: [TASK TYPE]
<#system#>
{system_message}
<#chat#>
<#user#>
{prompt}
<#bot#>
'''

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 AutoAWQ

Install the AutoAWQ package

Requires: AutoAWQ 0.1.1 or later.

pip3 install autoawq

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 .

AutoAWQ example code

python

from awq import AutoAWQForCausalLM
from transformers import AutoTokenizer

model_name_or_path = "TheBloke/Inkbot-13B-8k-0.2-AWQ"

# Load tokenizer
tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, trust_remote_code=False)
# Load model
model = AutoAWQForCausalLM.from_quantized(model_name_or_path, fuse_layers=True,
                                          trust_remote_code=False, safetensors=True)

prompt = "Tell me about AI"
prompt_template=f'''<#meta#>
- Date: [DATE]
- Task: [TASK TYPE]
<#system#>
{system_message}
<#chat#>
<#user#>
{prompt}
<#bot#>
'''

print("*** Running model.generate:")

token_input = tokenizer(
    prompt_template,
    return_tensors='pt'
).input_ids.cuda()

# Generate output
generation_output = model.generate(
    token_input,
    do_sample=True,
    temperature=0.7,
    top_p=0.95,
    top_k=40,
    max_new_tokens=512
)

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

"""
# Inference should be possible with transformers pipeline as well in future
# But currently this is not yet supported by AutoAWQ (correct as of September 25th 2023)
from transformers import pipeline

print("*** Pipeline:")
pipe = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
    max_new_tokens=512,
    do_sample=True,
    temperature=0.7,
    top_p=0.95,
    top_k=40,
    repetition_penalty=1.1
)

print(pipe(prompt_template)[0]['generated_text'])
"""

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.
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: Pierre Kircher, Stanislav Ovsiannikov, Michael Levine, Eugene Pentland, Andrey, 준교 김, Randy H, Fred von Graf, Artur Olbinski, Caitlyn Gatomon, terasurfer, Jeff Scroggin, James Bentley, Vadim, Gabriel Puliatti, Harry Royden McLaughlin, Sean Connelly, Dan Guido, Edmond Seymore, Alicia Loh, subjectnull, AzureBlack, Manuel Alberto Morcote, Thomas Belote, Lone Striker, Chris Smitley, Vitor Caleffi, Johann-Peter Hartmann, Clay Pascal, biorpg, Brandon Frisco, sidney chen, transmissions 11, Pedro Madruga, jinyuan sun, Ajan Kanaga, Emad Mostaque, Trenton Dambrowitz, Jonathan Leane, Iucharbius, usrbinkat, vamX, George Stoitzev, Luke Pendergrass, theTransient, Olakabola, Swaroop Kallakuri, Cap'n Zoog, Brandon Phillips, Michael Dempsey, Nikolai Manek, danny, Matthew Berman, Gabriel Tamborski, alfie_i, Raymond Fosdick, Tom X Nguyen, Raven Klaugh, LangChain4j, Magnesian, Illia Dulskyi, David Ziegler, Mano Prime, Luis Javier Navarrete Lozano, Erik Bjäreholt, 阿明, Nathan Dryer, Alex, Rainer Wilmers, zynix, TL, Joseph William Delisle, John Villwock, Nathan LeClaire, Willem Michiel, Joguhyik, GodLy, OG, Alps Aficionado, Jeffrey Morgan, ReadyPlayerEmma, Tiffany J. Kim, Sebastain Graf, Spencer Kim, Michael Davis, webtim, Talal Aujan, knownsqashed, John Detwiler, Imad Khwaja, Deo Leter, Jerry Meng, Elijah Stavena, Rooh Singh, Pieter, SuperWojo, Alexandros Triantafyllidis, Stephen Murray, Ai Maven, ya boyyy, Enrico Ros, Ken Nordquist, Deep Realms, Nicholas, Spiking Neurons AB, Elle, Will Dee, Jack West, RoA, Luke @flexchar, Viktor Bowallius, Derek Yates, Subspace Studios, jjj, Toran Billups, Asp the Wyvern, Fen Risland, Ilya, NimbleBox.ai, Chadd, Nitin Borwankar, Emre, Mandus, Leonard Tan, Kalila, K, Trailburnt, S_X, Cory Kujawski

Thank you to all my generous patrons and donaters!

And thank you again to a16z for their generous grant.

Original model card: Adam Brusselback's Inkbot 13B 8K 0.2

Model Card for Inkbot

Model Details

Inkbot is a conversational AI model designed to interpret and respond to structured prompts with or without contextual information. Its unique feature lies in the structured prompt system, allowing users to engage in dynamic dialogues that can evolve based on context, metadata, and user input.

Performance

The model excels in RAG type queries, answering from context, and overriding memory when necessary.
It can handle very large contexts, but may sometimes enter a repeating text loop, especially during complex tasks.
The model is intended to be more functional and less chatty, avoiding the waste of tokens on superfluous language.

How to Use

Inkbot uses a structured prompt template system.

Prompt Template Structure

With Context:

<#meta#>
- Date: {current_date}
- Task: {task_name}
<#system#>
{system_prompt}
<#chat#>
<#user#>
{user}
<#user_context#>
{user_context}
<#bot#>
{bot}

Without Context:

<#meta#>
- Date: {current_date}
- Task: {task_name}
<#system#>
{system_prompt}
<#chat#>
<#user#>
{user}
<#bot#>
{bot}

Examples

With Context:

<#meta#>
- Date: 2023-07-20
- Task: general
<#system#>
A dialogue focused on determining dates based on context clues.
<#chat#>
<#user#>
What is today's date?
<#bot#>
The date is: 2023-07-20
<#user#>
When did the event in the provided context take place?
<#user_context#>
It's the day after new years eve, and I'm starting 2018 with a headache.
<#bot#>
The event in the context took place on: 2018-01-01

Without Context:

<#meta#>
- Date: 2023-07-20
- Task: general
<#system#>
A simple dialogue about daily queries.
<#chat#>
<#user#>
Will it rain today?
<#bot#>
I am not sure, without knowing your location I am unable to determine the weather.
<#user#>
Can you compose a haiku for me?
<#bot#>
Whispers of the breeze,
Nature's melodies unfold,
Dawn dispels the night.

Task Options

The model has been trained on a variety of tasks. Below is a breakdown of these tasks, along with example prompts to help guide your usage.

2. Content Generation

Tasks in this category involve creating or expanding content.

kg_writer (800 examples)

Example Prompts:

"Using the provided knowledge graph, write an article about the topics and entities in the graph, incorporating the linked ideas. Use idea tags while writing to help focus."
"Construct a story based on the information in the knowledge graph."

summary (1,600 examples)

Example Prompts:

"Generate an extensive summary of the given document."
"Please read the provided document to understand the context and content. Use this understanding to generate a summary. Separate the article into chunks, and sequentially create a summary for each chunk. Give me a final summary in the end."

paraphrase (1,100 examples)

Example Prompts:

"Rephrase the following sentence while retaining its original meaning."
"Can you provide an alternative wording for the paragraph below?"

3. Content Analysis

Tasks in this category evaluate, grade, or filter content.

grading (400 examples)

Example Prompts:

"Based on the provided document, please rate the usefulness as training data on a scale from 0-5."

sponsorblock (5,200 examples)

Example Prompts:

"Read the document and extract any sentences or phrases that contain explicit mentions of sponsorship, promotional partnerships, or any form of paid content."

4. Information Structuring

Tasks in this category involve the structured representation or extraction of information.

kg (3,600 examples)

Example Prompts:

"Create a Knowledge Graph from the document provided."
"Extract key concepts and relationships from the conversation to form a knowledge graph."

5. General Interaction

Tasks in this category are designed for general questions and interactions.

general (1,600 examples)

Example Prompts:

"What is the capital of France?"
"Explain particle physics to a 5 years old."

Limitations

Adhere to the prompt structure for best results.
When providing contextual details, clarity is essential for Inkbot to derive accurate and meaningful responses.
The overriding memory from user_context property generally only works for the next prompt or two, after which the model reverts to its original behavior.
On complex tasks, like creating a coherent story based on a set of facts from context, there's a potential for a repeating text loop as context fills.
Sometimes the model doesn't know when to end a knowledge graph, which can result in adding nodes and edges until it runs out of context.

Additional Notes

Use rope-freq-scale=0.5 or compress_pos_emb=2 for 8k ctx
The 'date', 'task', and 'system' are crucial metadata components that need to be provided outside the core dialogue.
Use the 'user_context' when you want to offer supplementary context that guides Inkbot's response. You can interleave it in the chat log as necessary. It comes after the users instruction.
The specific tag format, such as <#word#>, is used to because there are filters in a lot of APIs for <|word|> and this makes interactions easier.

TheBloke/Inkbot-13B-8k-0.2-AWQ

Inkbot 13B 8K 0.2 - AWQ

Description

About AWQ

Repositories available

Prompt template: Inkbot

Provided files, and AWQ parameters

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

Multi-user inference server: vLLM

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

Inference from Python code using AutoAWQ

Install the AutoAWQ package

AutoAWQ example code

Compatibility

Discord

Thanks, and how to contribute

Original model card: Adam Brusselback's Inkbot 13B 8K 0.2

Model Card for Inkbot

Model Details

Performance

How to Use

Prompt Template Structure

With Context:

Without Context:

Examples

With Context:

Without Context:

Task Options

2. Content Generation

kg_writer (800 examples)

summary (1,600 examples)

paraphrase (1,100 examples)

3. Content Analysis

grading (400 examples)

sponsorblock (5,200 examples)

4. Information Structuring

kg (3,600 examples)

5. General Interaction

general (1,600 examples)

Limitations

Additional Notes

license: llama2

TheBloke/
Inkbot-13B-8k-0.2-AWQ