NEURAL-CHAT-7B-V3-1 - GGUF

Model creator: Intel
Original model: Intel/neural-chat-7b-v3-1

Finetuning on habana HPU

This model is a fine-tuned model based on mistralai/Mistral-7B-v0.1 on the open source dataset Open-Orca/SlimOrca. Then we align it with DPO algorithm. For more details, you can refer our blog: NeuralChat: Simplifying Supervised Instruction Fine-Tuning and Reinforcement Aligning and The Practice of Supervised Fine-tuning and Direct Preference Optimization on Habana Gaudi2.

Model date

neural-chat-7b-v3-1 was trained between September and October, 2023.

HOW TO USE

What is GGUF

GGUF is a new format introduced by the llama.cpp team on August 21st 2023. It is a replacement for GGML, which is no longer supported by llama.cpp.

What is the software that support GGUF

Here is an incomplate list of clients and libraries that are known to support GGUF:

llama.cpp. The source project for GGUF. Offers a CLI and a server option.
text-generation-webui, the most widely used web UI, with many features and powerful extensions. Supports GPU acceleration.
KoboldCpp, a fully featured web UI, with GPU accel across all platforms and GPU architectures. Especially good for story telling.
LM Studio, an easy-to-use and powerful local GUI for Windows and macOS (Silicon), with GPU acceleration.
LoLLMS Web UI, a great web UI with many interesting and unique features, including a full model library for easy model selection.
Faraday.dev, an attractive and easy to use character-based chat GUI for Windows and macOS (both Silicon and Intel), with GPU acceleration.
ctransformers, a Python library with GPU accel, LangChain support, and OpenAI-compatible AI server.
llama-cpp-python, a Python library with GPU accel, LangChain support, and OpenAI-compatible API server.
candle, a Rust ML framework with a focus on performance, including GPU support, and ease of use.

Compatibility

These quantised GGUFv2 files are compatible with llama.cpp from August 27th onwards, as of commit d0cee0d

They are also compatible with many third party UIs and libraries - please see the list at the top of this README.

Provided files

Name	Quant method	Bits	Size	Use case
neural-chat-7b-v3-1-model-Q2_K.gguf	Q2_K	2	3.08 GB	smallest, significant quality loss - not recommended for most purposes
neural-chat-7b-v3-1-model-Q3_K_M.gguf	Q3_K_M	3	3.52 GB	very small, high quality loss
neural-chat-7b-v3-1-model-Q4_0.gguf	Q4_0	4	4.11 GB	legacy; small, very high quality loss - prefer using Q3_K_M
neural-chat-7b-v3-1-model-Q4_K_M.gguf	Q4_K_M	4	4.37 GB	medium, balanced quality - recommended
neural-chat-7b-v3-1-model-Q5_0.gguf	Q5_0	5	5 GB	legacy; medium, balanced quality - prefer using Q4_K_M
neural-chat-7b-v3-1-model-Q5_K_M.gguf	Q5_K_M	5	5.13 GB	large, very low quality loss - recommended
neural-chat-7b-v3-1-model-Q6_K.gguf	Q6_K	6	5.94 GB	very large, extremely low quality loss
neural-chat-7b-v3-1-model-Q8_0.gguf	Q8_0	8	7.7 GB	very large, extremely low quality loss - not recommended

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.

Explanation of quantisation methods

Click to see details

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

How to download GGUF files

Note for manual downloaders: You almost never want to clone the entire repo! Multiple different quantisation formats are provided, and most users only want to pick and download a single file.

The following clients/libraries will automatically download models for you, providing a list of available models to choose from:

LM Studio
LoLLMS Web UI
Faraday.dev

In `text-generation-webui`

Under Download Model, you can enter the model repo: Ichsan2895/neural-chat-7b-v3-1-GGUF and below it, a specific filename to download, such as: neural-chat-7b-v3-1-model-q5_k_m.gguf.

Then click Download.

On the command line, including multiple files at once

I recommend using the huggingface-hub Python library:

pip3 install huggingface-hub

Then you can download any individual model file to the current directory, at high speed, with a command like this:

huggingface-cli download Ichsan2895/neural-chat-7b-v3-1-GGUF neural-chat-7b-v3-1-model-q5_k_m.gguf --local-dir . --local-dir-use-symlinks False

More advanced huggingface-cli download usage

You can also download multiple files at once with a pattern:

huggingface-cli download Ichsan2895/neural-chat-7b-v3-1-GGUF --local-dir . --local-dir-use-symlinks False --include='*Q4_K*gguf'

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:

HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download Ichsan2895/neural-chat-7b-v3-1-GGUF neural-chat-7b-v3-1-model-q5_k_m.gguf --local-dir . --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.

Example `llama.cpp` command

Make sure you are using llama.cpp from commit d0cee0d or later.

./main -ngl 32 -m neural-chat-7b-v3-1-model-q5_k_m.gguf --color -c 2048 --temp 0.7 --repeat_penalty 1.1 -n -1 -p "<|im_start|>system\n{system_message}<|im_end|>\n<|im_start|>user\n{prompt}<|im_end|>\n<|im_start|>assistant"

Change -ngl 32 to the number of layers to offload to GPU. Remove it if you don't have GPU acceleration.

Change -c 2048 to the desired sequence length. For extended sequence models - eg 8K, 16K, 32K - the necessary RoPE scaling parameters are read from the GGUF file and set by llama.cpp automatically.

If you want to have a chat-style conversation, replace the -p <PROMPT> argument with -i -ins

For other parameters and how to use them, please refer to the llama.cpp documentation

How to run in `text-generation-webui`

Further instructions here: text-generation-webui/docs/llama.cpp.md.

How to run from Python code

You can use GGUF models from Python using the llama-cpp-python or ctransformers libraries.

How to load this model in Python code, using ctransformers

First install the package

Run one of the following commands, according to your system:

# Base ctransformers with no GPU acceleration
pip install ctransformers
# Or with CUDA GPU acceleration
pip install ctransformers[cuda]
# Or with AMD ROCm GPU acceleration (Linux only)
CT_HIPBLAS=1 pip install ctransformers --no-binary ctransformers
# Or with Metal GPU acceleration for macOS systems only
CT_METAL=1 pip install ctransformers --no-binary ctransformers

Simple ctransformers example code

from ctransformers import AutoModelForCausalLM

# Set gpu_layers to the number of layers to offload to GPU. Set to 0 if no GPU acceleration is available on your system.
llm = AutoModelForCausalLM.from_pretrained("Ichsan2895/neural-chat-7b-v3-1-GGUF", model_file="neural-chat-7b-v3-1-model-q5_k_m.gguf", model_type="mistral", gpu_layers=50)

print(llm("AI is going to"))

How to use with LangChain

Here are guides on using llama-cpp-python and ctransformers with LangChain:

Evaluation

We submit our model to open_llm_leaderboard, and the model performance has been improved significantly as we see from the average metric of 7 tasks from the leaderboard.

Model	Average ⬆️	ARC (25-s) ⬆️	HellaSwag (10-s) ⬆️	MMLU (5-s) ⬆️	TruthfulQA (MC) (0-s) ⬆️	Winogrande (5-s)	GSM8K (5-s)	DROP (3-s)
mistralai/Mistral-7B-v0.1	50.32	59.58	83.31	64.16	42.15	78.37	18.12	6.14
Intel/neural-chat-7b-v3	57.31	67.15	83.29	62.26	58.77	78.06	1.21	50.43
Intel/neural-chat-7b-v3-1	59.06	66.21	83.64	62.37	59.65	78.14	19.56	43.84

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 1e-04
train_batch_size: 1
eval_batch_size: 2
seed: 42
distributed_type: multi-HPU
num_devices: 8
gradient_accumulation_steps: 8
total_train_batch_size: 64
total_eval_batch_size: 8
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.02
num_epochs: 2.0

Inference with transformers

import transformers
model = transformers.AutoModelForCausalLM.from_pretrained(
  'Intel/neural-chat-7b-v3-1'
)

Ethical Considerations and Limitations

neural-chat-7b-v3-1 can produce factually incorrect output, and should not be relied on to produce factually accurate information. neural-chat-7b-v3-1 was trained on Open-Orca/SlimOrca based on mistralai/Mistral-7B-v0.1. Because of the limitations of the pretrained model and the finetuning datasets, it is possible that this model could generate lewd, biased or otherwise offensive outputs.

Therefore, before deploying any applications of neural-chat-7b-v3-1, developers should perform safety testing.

Disclaimer

The license on this model does not constitute legal advice. We are not responsible for the actions of third parties who use this model. Please cosult an attorney before using this model for commercial purposes.

Organizations developing the model

The NeuralChat team with members from Intel/SATG/AIA/AIPT. Core team members: Kaokao Lv, Liang Lv, Chang Wang, Wenxin Zhang, Xuhui Ren, and Haihao Shen.

Useful links

Intel Neural Compressor link
Intel Extension for Transformers link
Intel Extension for PyTorch link