TenyxChat-7B-v1-AWQ / README.md
TheBloke's picture
Upload README.md
f2195ad verified
metadata
base_model: tenyx/TenyxChat-7B-v1
inference: false
language:
  - en
library_name: transformers
license: apache-2.0
model_creator: Tenyx
model_name: TenyxChat 7B v1
model_type: mistral
prompt_template: |
  System: {system_message}
  User: {prompt}
  Assistant:
quantized_by: TheBloke
tags:
  - tenyx-fine-tuning
  - dpo
  - tenyxchat
TheBlokeAI

TheBloke's LLM work is generously supported by a grant from andreessen horowitz (a16z)


TenyxChat 7B v1 - AWQ

Description

This repo contains AWQ model files for Tenyx's TenyxChat 7B v1.

These files were quantised using hardware kindly provided by Massed Compute.

About AWQ

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:

Repositories available

Prompt template: System-User-Assistant-nohash

System: {system_message}
User: {prompt}
Assistant:

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.

  1. Click the Model tab.
  2. Under Download custom model or LoRA, enter TheBloke/TenyxChat-7B-v1-AWQ.
  3. Click Download.
  4. The model will start downloading. Once it's finished it will say "Done".
  5. In the top left, click the refresh icon next to Model.
  6. In the Model dropdown, choose the model you just downloaded: TenyxChat-7B-v1-AWQ
  7. Select Loader: AutoAWQ.
  8. Click Load, and the model will load and is now ready for use.
  9. 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.
  10. 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/TenyxChat-7B-v1-AWQ --quantization awq --dtype auto
  • When using vLLM from Python code, again set quantization=awq.

For example:

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'''System: {system_message}
User: {prompt}
Assistant:
'''

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

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

llm = LLM(model="TheBloke/TenyxChat-7B-v1-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/TenyxChat-7B-v1-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
from huggingface_hub import InferenceClient

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

prompt = "Tell me about AI"
prompt_template=f'''System: {system_message}
User: {prompt}
Assistant:
'''

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

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:

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)

from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer

model_name_or_path = "TheBloke/TenyxChat-7B-v1-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'''System: {system_message}
User: {prompt}
Assistant:
'''

# 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:

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.

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: Tenyx's TenyxChat 7B v1

TenyxChat: Language Model Alignment using Tenyx Fine-tuning

Introducing TenyxChat, a series of ChatGPT-like models trained to function as useful assistants through preference tuning, using Tenyx's recently released advanced fine-tuning technology (VentureBeat article). Our first chat model in the series, TenyxChat-7B-v1, is trained using the Direct Preference Optimization (DPO) framework on the open-source AI feedback dataset UltraFeedback.

We fine-tune Openchat-3.5 with our proprietary approach (blog, service), which shows an increase in MT-Bench, without a drop in performance of the model on other benchmarks. Our approach aims to mitigate forgetting in LLMs in a computationally efficient manner, thereby enabling continual fine-tuning capabilities without altering the pre-trained output distribution. TenyxChat-7B-v1 was trained using eight A100s (80GB) for two hours, with a training setup obtained from HuggingFaceH4 (GitHub).

Model details

Usage

Our model uses a simple chat template based on OpenChat 3.5. The chat template usage with a Hugging face generation example is shown below.

Chat Template (Jinja)

{{ bos_token }} 
{% for message in messages %}
    {% if message['role'] == 'user' %}
        {{ 'User:' + message['content'] + eos_token }}

    {% elif message['role'] == 'system' %}
        {{ 'System:' + message['content'] + eos_token }}

    {% elif message['role'] == 'assistant' %}
        {{ 'Assistant:'  + message['content'] + eos_token }}

    {% endif %}

{% if loop.last and add_generation_prompt %}\n{{ 'Assistant:' }}{% endif %}\n{% endfor %}

Hugging face Example

import torch
from transformers import pipeline

pipe = pipeline("text-generation", model="tenyx/TenyxChat-7B-v1", torch_dtype=torch.bfloat16, device_map="auto")

messages = [
    {"role": "system", "content": "You are a friendly chatbot who always responds in the style of a pirate."},
    {"role": "user", "content": "Hi. I would like to make a hotel booking."},
]

prompt = pipe.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
outputs = pipe(prompt, max_new_tokens=512, do_sample=False)

Output

<s> System:You are a friendly chatbot who always responds in the style of a pirate.<|end_of_turn|>
User:Hi. I would like to make a hotel booking.<|end_of_turn|>
Assistant: Ahoy there me hearty! Arr, ye be lookin' fer a place to rest yer weary bones, eh? 
Well then, let's set sail on this grand adventure and find ye a swell place to stay!

To begin, tell me the location ye be seekin' and the dates ye be lookin' to set sail. 
And don't ye worry, me matey, I'll be sure to find ye a place that'll make ye feel like a king or queen on land!

Performance

At the time of release (Jan 2024), TenyxChat-7B-v1 is the highest-ranked 7B chat model on the MT-Bench evaluation available for download and commercial use. We list here the benchmark results on several standard setups while comparing popular 7B models as baselines.

MT-Bench

MT-Bench is a benchmark made up of 80 high-quality multi-turn questions. These questions fall into eight categories: Writing, Roleplay, Reasoning, Math, Coding, Extraction, STEM, and Humanities. The chat models are rated using GPT-4 on a scale of 1 to 10, with higher values corresponding to better responses.

Model First Turn Second Turn Average
GPT-4* 8.95625 9.02500 8.990625
TenyxChat-7B-v1 8.45000 7.75625 8.103125
Starling-lm-7B-alpha 8.42500 7.68750 8.056250
OpenChat-3.5 8.18125 7.41250 7.796875
GPT-3.5-turbo* 8.07500 7.81250 7.943750
OpenLLM Leader-7B** 8.05000 7.61250 7.831250

*values reported on lmsys ChatBot Arena

**The OpenLLM Leader as of Jan 5, 2024 is the merge model available as samir-fama/SamirGPT-v1

hexplot.png

Comparison with additional Open LLM LeaderBoard models

Model First Turn Second Turn Average
TenyxChat-7B-v1 8.45000 7.756250 8.103125
SamirGPT-v1 8.05000 7.612500 7.831250
FernandoGPT-v1 8.08125 7.256250 7.668750
Go-Bruins-v2 8.13750 7.150000 7.643750
mistral_tv-neural-marconroni 7.76875 6.987500 7.378125
neuronovo-7B-v0.2 7.73750 6.662500 7.200000
neural-chat-7b-v3-3 7.39375 5.881250 6.637500

LM Evaluation - Open LLM Leaderboard

We assess models on 7 benchmarks using the Eleuther AI Language Model Evaluation Harness. This setup is based of that used for Open LLM Leaderboard.

  • AI2 Reasoning Challenge (25-shot) - grade-school science questions.
  • HellaSwag (10-shot) - commonsense inference test.
  • MMLU (5-shot) - multitask accuracy test covering 57 tasks.
  • TruthfulQA (0-shot) - test measuring model's propensity to reproduce online falsehoods.
  • Winogrande (5-shot) - Winograd benchmark for commonsense reasoning.
  • GSM8k (5-shot) - grade school math word problems test.

These benchmarks test reasoning and knowledge in various tasks in few-shot settings (higher scores are better).

Model MMLU Winogrande GSM8k ARC HellaSwag TruthfulQA Average
TenyxChat-7B-v1 63.6 72.3 69.0 62.7 66.6 46.7 63.48
Starling-7B-alpha 63.5 72.1 67.9 61.1 66.1 42.1 62.13
OpenChat-3.5 63.6 72.1 68.2 61.3 65.2 41.8 62.03
Mistral-7B 62.4 74.0 38.1 57.2 62.8 37.8 55.38
OpenLLM Leader-7B 64.3 78.7 73.3 66.6 68.4 58.5 68.3

Note: While the Open LLM Leaderboard indicates that these chat models perform less effectively compared to the leading 7B model, it's important to note that the leading model struggles in the multi-turn chat setting of MT-Bench (as demonstrated in our evaluation above). In contrast, TenyxChat-7B-v1 demonstrates robustness against common fine-tuning challenges, such as catastrophic forgetting. This unique feature enables TenyxChat-7B-v1 to excel not only in chat benchmarks like MT-Bench, but also in a wider range of general reasoning benchmarks on the Open LLM Leaderboard.

Limitations

TenyxChat-7B-v1, like other small-sized language models, has its own set of limitations. We haven’t fine-tuned the model explicitly to align with human safety preferences. Therefore, it is capable of producing undesirable outputs, particularly when adversarially prompted. From our observation, the model still tends to struggle with tasks that involve reasoning and math questions. In some instances, it might generate verbose or extraneous content.

License

TenyxChat-7B-v1, similar to OpenChat 3.5, is distributed under the Apache License 2.0.

Citation

If you use TenyxChat-7B for your research, cite us as

@misc{tenyxchat2024,
      title={TenyxChat: Language Model Alignment using Tenyx Fine-tuning}, 
      author={Tenyx},
      year={2024},
}