TheBloke's LLM work is generously supported by a grant from andreessen horowitz (a16z)
Nanbeige 16B Chat - GPTQ
- Model creator: Nanbeige LLM Lab
- Original model: Nanbeige 16B Chat
Description
This repo contains GPTQ model files for Nanbeige LLM Lab's Nanbeige 16B Chat.
Multiple GPTQ parameter permutations are provided; see Provided Files below for details of the options provided, their parameters, and the software used to create them.
These files were quantised using hardware kindly provided by Massed Compute.
Repositories available
- GPTQ models for GPU inference, with multiple quantisation parameter options.
- 2, 3, 4, 5, 6 and 8-bit GGUF models for CPU+GPU inference
- Nanbeige LLM Lab's original unquantised fp16 model in pytorch format, for GPU inference and for further conversions
Prompt template: Unknown
{prompt}
Known compatible clients / servers
These GPTQ models are known to work in the following inference servers/webuis.
This may not be a complete list; if you know of others, please let me know!
Provided files, and GPTQ parameters
Multiple quantisation parameters are provided, to allow you to choose the best one for your hardware and requirements.
Each separate quant is in a different branch. See below for instructions on fetching from different branches.
Most GPTQ files are made with AutoGPTQ. Mistral models are currently made with Transformers.
Explanation of GPTQ parameters
- Bits: The bit size of the quantised model.
- GS: GPTQ group size. Higher numbers use less VRAM, but have lower quantisation accuracy. "None" is the lowest possible value.
- Act Order: True or False. Also known as
desc_act. True results in better quantisation accuracy. Some GPTQ clients have had issues with models that use Act Order plus Group Size, but this is generally resolved now. - Damp %: A GPTQ parameter that affects how samples are processed for quantisation. 0.01 is default, but 0.1 results in slightly better accuracy.
- GPTQ dataset: The calibration dataset used during quantisation. Using a dataset more appropriate to the model's training can improve quantisation accuracy. Note that the GPTQ calibration dataset is not the same as the dataset used to train the model - please refer to the original model repo for details of the training dataset(s).
- Sequence Length: The length of the dataset sequences used for quantisation. Ideally this is the same as the model sequence length. For some very long sequence models (16+K), a lower sequence length may have to be used. Note that a lower sequence length does not limit the sequence length of the quantised model. It only impacts the quantisation accuracy on longer inference sequences.
- ExLlama Compatibility: Whether this file can be loaded with ExLlama, which currently only supports Llama and Mistral models in 4-bit.
| Branch | Bits | GS | Act Order | Damp % | GPTQ Dataset | Seq Len | Size | ExLlama | Desc |
|---|---|---|---|---|---|---|---|---|---|
| main | 4 | 128 | Yes | 0.1 | wikitext | 4096 | 4.99 GB | No | 4-bit, with Act Order and group size 128g. Uses even less VRAM than 64g, but with slightly lower accuracy. |
| gptq-4bit-32g-actorder_True | 4 | 32 | Yes | 0.1 | wikitext | 4096 | 4.97 GB | No | 4-bit, with Act Order and group size 32g. Gives highest possible inference quality, with maximum VRAM usage. |
| gptq-8bit--1g-actorder_True | 8 | None | Yes | 0.1 | wikitext | 4096 | 4.98 GB | No | 8-bit, with Act Order. No group size, to lower VRAM requirements. |
| gptq-8bit-128g-actorder_True | 8 | 128 | Yes | 0.1 | wikitext | 4096 | 4.94 GB | No | 8-bit, with group size 128g for higher inference quality and with Act Order for even higher accuracy. |
| gptq-8bit-32g-actorder_True | 8 | 32 | Yes | 0.1 | wikitext | 4096 | 4.96 GB | No | 8-bit, with group size 32g and Act Order for maximum inference quality. |
| gptq-4bit-64g-actorder_True | 4 | 64 | Yes | 0.1 | wikitext | 4096 | 4.98 GB | No | 4-bit, with Act Order and group size 64g. Uses less VRAM than 32g, but with slightly lower accuracy. |
How to download, including from branches
In text-generation-webui
To download from the main branch, enter TheBloke/Nanbeige-16B-Chat-GPTQ in the "Download model" box.
To download from another branch, add :branchname to the end of the download name, eg TheBloke/Nanbeige-16B-Chat-GPTQ:gptq-4bit-32g-actorder_True
From the command line
I recommend using the huggingface-hub Python library:
pip3 install huggingface-hub
To download the main branch to a folder called Nanbeige-16B-Chat-GPTQ:
mkdir Nanbeige-16B-Chat-GPTQ
huggingface-cli download TheBloke/Nanbeige-16B-Chat-GPTQ --local-dir Nanbeige-16B-Chat-GPTQ --local-dir-use-symlinks False
To download from a different branch, add the --revision parameter:
mkdir Nanbeige-16B-Chat-GPTQ
huggingface-cli download TheBloke/Nanbeige-16B-Chat-GPTQ --revision gptq-4bit-32g-actorder_True --local-dir Nanbeige-16B-Chat-GPTQ --local-dir-use-symlinks False
More advanced huggingface-cli download usage
If you remove the --local-dir-use-symlinks False parameter, the files will instead be stored in the central Hugging Face cache directory (default location on Linux is: ~/.cache/huggingface), and symlinks will be added to the specified --local-dir, pointing to their real location in the cache. This allows for interrupted downloads to be resumed, and allows you to quickly clone the repo to multiple places on disk without triggering a download again. The downside, and the reason why I don't list that as the default option, is that the files are then hidden away in a cache folder and it's harder to know where your disk space is being used, and to clear it up if/when you want to remove a download model.
The cache location can be changed with the HF_HOME environment variable, and/or the --cache-dir parameter to huggingface-cli.
For more documentation on downloading with huggingface-cli, please see: HF -> Hub Python Library -> Download files -> Download from the CLI.
To accelerate downloads on fast connections (1Gbit/s or higher), install hf_transfer:
pip3 install hf_transfer
And set environment variable HF_HUB_ENABLE_HF_TRANSFER to 1:
mkdir Nanbeige-16B-Chat-GPTQ
HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download TheBloke/Nanbeige-16B-Chat-GPTQ --local-dir Nanbeige-16B-Chat-GPTQ --local-dir-use-symlinks False
Windows Command Line users: You can set the environment variable by running set HF_HUB_ENABLE_HF_TRANSFER=1 before the download command.
With git (not recommended)
To clone a specific branch with git, use a command like this:
git clone --single-branch --branch gptq-4bit-32g-actorder_True https://huggingface.co/TheBloke/Nanbeige-16B-Chat-GPTQ
Note that using Git with HF repos is strongly discouraged. It will be much slower than using huggingface-hub, and will use twice as much disk space as it has to store the model files twice (it stores every byte both in the intended target folder, and again in the .git folder as a blob.)
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/Nanbeige-16B-Chat-GPTQ.- To download from a specific branch, enter for example
TheBloke/Nanbeige-16B-Chat-GPTQ:gptq-4bit-32g-actorder_True - see Provided Files above for the list of branches for each option.
- To download from a specific branch, enter for example
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:
Nanbeige-16B-Chat-GPTQThe model will automatically 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.
- Note that you do not need to and should not set manual GPTQ parameters any more. These are set automatically from the file
quantize_config.json.
- Note that you do not need to and should not set manual GPTQ parameters any more. These are set automatically from the file
Once you're ready, click the Text Generation tab and enter a prompt to get started!
Serving this model from Text Generation Inference (TGI)
It's recommended to 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/Nanbeige-16B-Chat-GPTQ --port 3000 --quantize gptq --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'''{prompt}
'''
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}")
Python code example: inference from this GPTQ model
Install the necessary packages
Requires: Transformers 4.33.0 or later, Optimum 1.12.0 or later, and AutoGPTQ 0.4.2 or later.
pip3 install --upgrade transformers optimum
# If using PyTorch 2.1 + CUDA 12.x:
pip3 install --upgrade auto-gptq
# or, if using PyTorch 2.1 + CUDA 11.x:
pip3 install --upgrade auto-gptq --extra-index-url https://huggingface.github.io/autogptq-index/whl/cu118/
If you are using PyTorch 2.0, you will need to install AutoGPTQ from source. Likewise if you have problems with the pre-built wheels, you should try building from source:
pip3 uninstall -y auto-gptq
git clone https://github.com/PanQiWei/AutoGPTQ
cd AutoGPTQ
git checkout v0.5.1
pip3 install .
Example Python code
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
model_name_or_path = "TheBloke/Nanbeige-16B-Chat-GPTQ"
# To use a different branch, change revision
# For example: revision="gptq-4bit-32g-actorder_True"
model = AutoModelForCausalLM.from_pretrained(model_name_or_path,
device_map="auto",
trust_remote_code=True,
revision="main")
tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True)
prompt = "Tell me about AI"
prompt_template=f'''{prompt}
'''
print("\n\n*** Generate:")
input_ids = tokenizer(prompt_template, return_tensors='pt').input_ids.cuda()
output = model.generate(inputs=input_ids, temperature=0.7, do_sample=True, top_p=0.95, top_k=40, max_new_tokens=512)
print(tokenizer.decode(output[0]))
# Inference can also be done using transformers' 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 Transformers. For non-Mistral models, AutoGPTQ can also be used directly.
ExLlama is compatible with Llama and Mistral models in 4-bit. Please see the Provided Files table above for per-file compatibility.
For a list of clients/servers, please see "Known compatible clients / servers", above.
Discord
For further support, and discussions on these models and AI in general, join us at:
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: Brandon Frisco, LangChain4j, Spiking Neurons AB, transmissions 11, Joseph William Delisle, Nitin Borwankar, Willem Michiel, Michael Dempsey, vamX, Jeffrey Morgan, zynix, jjj, Omer Bin Jawed, Sean Connelly, jinyuan sun, Jeromy Smith, Shadi, Pawan Osman, Chadd, Elijah Stavena, Illia Dulskyi, Sebastain Graf, Stephen Murray, terasurfer, Edmond Seymore, Celu Ramasamy, Mandus, Alex, biorpg, Ajan Kanaga, Clay Pascal, Raven Klaugh, 阿明, K, ya boyyy, usrbinkat, Alicia Loh, John Villwock, ReadyPlayerEmma, Chris Smitley, Cap'n Zoog, fincy, GodLy, S_X, sidney chen, Cory Kujawski, OG, Mano Prime, AzureBlack, Pieter, Kalila, Spencer Kim, Tom X Nguyen, Stanislav Ovsiannikov, Michael Levine, Andrey, Trailburnt, Vadim, Enrico Ros, Talal Aujan, Brandon Phillips, Jack West, Eugene Pentland, Michael Davis, Will Dee, webtim, Jonathan Leane, Alps Aficionado, Rooh Singh, Tiffany J. Kim, theTransient, Luke @flexchar, Elle, Caitlyn Gatomon, Ari Malik, subjectnull, Johann-Peter Hartmann, Trenton Dambrowitz, Imad Khwaja, Asp the Wyvern, Emad Mostaque, Rainer Wilmers, Alexandros Triantafyllidis, Nicholas, Pedro Madruga, SuperWojo, Harry Royden McLaughlin, James Bentley, Olakabola, David Ziegler, Ai Maven, Jeff Scroggin, Nikolai Manek, Deo Leter, Matthew Berman, Fen Risland, Ken Nordquist, Manuel Alberto Morcote, Luke Pendergrass, TL, Fred von Graf, Randy H, Dan Guido, NimbleBox.ai, Vitor Caleffi, Gabriel Tamborski, knownsqashed, Lone Striker, Erik Bjäreholt, John Detwiler, Leonard Tan, Iucharbius
Thank you to all my generous patrons and donaters!
And thank you again to a16z for their generous grant.
Original model card: Nanbeige LLM Lab's Nanbeige 16B Chat
Nanbeige-16B-Chat
模型介绍（Introduction）
Nanbeige-16B（南北阁-16B）是南北阁大模型实验室研发的160亿参数规模的大语言模型，采用了2.5T Tokens进行预训练，数据包含大量互联网高质量语料、各类书籍、代码等领域脱敏文本，在各个权威测评数据集上都取得了不错的效果。本次发布包含有 Base、Chat 以及扩展上下文长度的 Base-32k、Chat-32k 版本。
Base-32k 版本基于Nanbeige-16B-Base模型，采用YaRN插值方法对位置编码进行修改，并以32k为最大长度进行了20B Tokens的中文、英文、代码语料的全参数增量预训练。
Chat 版本和 Chat-32k 版本分别基于Nanbeige-16B-Base模型和Nanbeige-16B-Base-32k模型，经过了大量人类对齐训练，能够更好、更安全地回复用户的问题。
如果您需要处理更长的上下文，我们推荐您使用Nanbeige-16B-Base-32k和Nanbeige-16B-Chat-32k。
本仓库为 Nanbeige-16B-Chat 的模型仓库。
Nanbeige-16B is a 16 billion parameter language model developed by Nanbeige LLM Lab. It uses 2.5T Tokens for pre-training. The training data includes a large amount of high-quality internet corpus, various books, code, etc. It has achieved good results on various authoritative evaluation data sets. This release includes the Base, Chat, Base-32k and Chat-32k.
The Base-32k version is based on the Nanbeige-16B-Base model, which uses YaRN interpolation method to modify the position encoding, and performs full parameter incremental pre-training with 20 billion tokens of Chinese, English, and code corpora, under 32k max length.
The Chat version and Chat-32k version are based on the Nanbeige-16B-Base model and Nanbeige-16B-Base-32k model respectively. They have undergone extensive human-aligned training.
If you need to deal with longer contexts, we recommend using Nanbeige-16B-Base-32k and Nanbeige-16B-Chat-32k.
This repository is the one for Nanbeige-16B-Chat model.
| Base Model | Base-32k Model | Chat Model | Chat-32k Model | |
|---|---|---|---|---|
| 16B | 🤗 Nanbeige-16B-Base | 🤗 Nanbeige-16B-Base-32k | 🤗 Nanbeige-16B-Chat | 🤗 Nanbeige-16B-Chat-32k |
模型推理 (Inference)
相关依赖
python 3.8及以上版本
transformers 4.33.3
pytorch 2.0及以上版本
python 3.8 and above
transformers 4.33.3
pytorch 2.0及以上版本
可以通过以下pip命令安装相关依赖库
You can install the dependent libraries with the following pip command
pip install transformers==4.33.3 transformers_stream_generator deepspeed einops==0.3.2 datasets==2.10.0
推理代码
通过以下代码可以调用模型进行聊天对话：
The following code can be used to invoke the model for chat dialogue:
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained("Nanbeige/Nanbeige-16B-Chat", use_fast=False, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("Nanbeige/Nanbeige-16B-Chat", device_map="auto", torch_dtype=torch.bfloat16, trust_remote_code=True)
question = "你可以给我一些具体的SEO优化技巧吗？"
output, messages = model.chat(tokenizer, question)
print(output)
性能测评（Performance Evaluation）
LLMEval-3
LLMEval-3 ( Github / 主页 ) 聚焦于专业知识能力评测，涵盖哲学、经济学、法学、教育学、文学、历史学、理学、工学、农学、医学、军事学、管理学、艺术学等教育部划定的13个学科门类、50余个二级学科，共计约20W道标准生成式问答题目。防止作弊是LLMEval-3考虑的重要因素。现有公开评测基准存在测试题库泄露的问题，因此可能出现“刷榜”、“刷分”等不公平现象，在LLMEval-3中，每个参与评测的系统需要完成从总题库中随机抽样的1000题，针对同一机构的模型，确保每次评测题目不重复。
我们基于 LLMEval-3 对 Nanbeige-16B-Chat 模型进行了全面测评，测评结果如下：
We conducted a comprehensive evaluation of Nanbeige-16B-Chat model based on LLMEval-3 ( Github / Homepage ), and the results are as follows:
| 模型名称 | 相对分数-GPT4 | 相对分数-GPT3.5 | 绝对分数 | 工学 | 经济学 | 教育学 | 法学 | 文学 | 管理学 | 理学 | 历史学 | 医学 | 军事学 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Baidu3.5 | 104.21 | 121.39 | 77.53 | 8.13 | 8.00 | 8.63 | 7.97 | 6.23 | 7.63 | 7.33 | 8.77 | 7.47 | 7.37 |
| ChatGLM-pro | 103.45 | 120.51 | 76.97 | 6.97 | 8.47 | 7.97 | 8.93 | 7.23 | 7.70 | 6.33 | 8.37 | 7.13 | 7.87 |
| GPT-4 | 100.00 | 116.49 | 74.40 | 7.23 | 7.80 | 7.73 | 8.40 | 6.73 | 7.67 | 7.73 | 7.07 | 6.20 | 7.83 |
| Nanbeige-16B-Chat | 94.26 | 109.80 | 70.13 | 6.00 | 7.87 | 8.20 | 8.33 | 6.07 | 6.83 | 6.00 | 7.80 | 5.80 | 7.23 |
| minimax-abab5 | 93.28 | 108.66 | 69.40 | 5.83 | 7.50 | 7.77 | 8.37 | 6.40 | 6.33 | 5.07 | 8.33 | 5.93 | 7.87 |
| Baichuan2-13B-Chat | 92.91 | 108.23 | 69.13 | 6.00 | 7.53 | 8.63 | 8.13 | 6.23 | 6.33 | 5.63 | 8.20 | 5.43 | 7.00 |
| Qwen-14B-Chat | 86.33 | 100.56 | 64.23 | 5.77 | 7.07 | 7.07 | 7.37 | 5.70 | 6.20 | 5.93 | 6.97 | 5.40 | 6.77 |
| GPT-3.5-turbo | 85.84 | 100.00 | 63.87 | 6.27 | 6.87 | 7.23 | 7.40 | 5.40 | 6.30 | 6.37 | 6.00 | 5.17 | 6.87 |
| ChatGLM2-6B | 75.71 | 88.19 | 56.33 | 4.03 | 6.33 | 7.00 | 7.57 | 4.77 | 5.93 | 4.23 | 5.87 | 5.07 | 5.53 |
| ziya_v1.1-13b | 70.20 | 81.78 | 52.23 | 4.67 | 5.77 | 6.07 | 6.53 | 4.53 | 5.33 | 3.70 | 5.00 | 4.63 | 6.00 |
| BELLE-Llama2-13B-chat-0.4M | 68.82 | 80.16 | 51.20 | 4.47 | 5.93 | 6.20 | 6.77 | 4.33 | 4.97 | 4.10 | 5.07 | 3.77 | 5.60 |
| Linly-Chinese-LLaMA2-13B | 67.82 | 79.00 | 50.46 | 3.87 | 5.80 | 5.83 | 6.57 | 3.93 | 5.37 | 4.07 | 5.43 | 3.93 | 5.67 |
| InternLM-Chat-7B | 61.06 | 71.13 | 45.43 | 3.83 | 5.13 | 5.27 | 6.57 | 3.90 | 4.83 | 3.10 | 4.87 | 3.67 | 4.27 |
| Llama-2-7b-chat-hf | 51.11 | 59.54 | 38.03 | 3.33 | 4.77 | 3.77 | 5.03 | 3.07 | 3.77 | 3.93 | 4.00 | 2.40 | 3.97 |
局限性（Limitations）
虽然我们在训练过程中非常注重模型的安全性，力求确保其输出符合伦理和法律要求的文本，但由于模型大小和概率生成范式的限制，无法完全避免产生各种不符合预期的输出情况。这些输出可能包含偏见、歧视等有害内容，请勿传播这些内容。我们不承担因传播不良信息而导致的任何后果。
While we place great emphasis on the safety of the model during the training process, striving to ensure that its outputs align with ethical and legal requirements, it may not completely avoid generating unexpected outputs due to the model's size and probabilistic nature. These outputs may include harmful content such as bias or discrimination. Please don't propagate such content. We do not assume any responsibility for the consequences resulting from the dissemination of inappropriate information.
协议（License）
使用 Nanbeige 模型时，您必须遵守 Apache 2.0 许可证和《南北阁大语言模型许可协议》。如果您打算将 Nanbeige 模型或其衍生产品用于商业目的，请通过以下联系邮箱 nanbeige@126.com 提交申请材料，以满足《南北阁大语言模型许可协议》的要求。经过审核后，我们将授予您非排他性、全球范围内、不可转让、不可再许可、可撤销的商业版权许可。
When using the Nanbeige models, you must comply with the Apache 2.0 License and the License Agreement for Large Language Models Nanbeige. If you intend to use the Nanbeige Models or its derivatives for commercial purposes, please submit application materials to meet the requirements of the Nanbeige Models Community License Agreement by contacting nanbeige@126.com. After review, We will grant you a non-exclusive, worldwide, non-transferable, non-sublicensable and revocable commercial copyright license.
- Downloads last month
- 14