Helpingai Dhanishtha 2.0 Preview GGUF

🚀 Llamacpp imatrix Quantizations of Dhanishtha-2.0-preview by HelpingAI

This project focuses on the Llamacpp imatrix quantizations of the Dhanishtha-2.0-preview model by HelpingAI. It provides various quantized models with different configurations, enabling users to choose according to their specific needs.

✨ Features

• Multilingual Support: The base model supports multiple languages, including English, Hindi, Chinese, Spanish, French, German, Japanese, Korean, Arabic, Portuguese, Russian, Italian, Dutch, Turkish, Polish, Swedish, Danish, Norwegian, Finnish, Hebrew, Thai, Vietnamese, Indonesian, Malay, Tagalog, Swahili, Yoruba, Zulu, Amharic, Bengali, Gujarati, Kannada, Malayalam, Marathi, Nepali, Odia, Punjabi, Tamil, Telugu, Urdu, and multilingual.
• Rich Functionality: It has capabilities in reasoning, intermediate thinking, and conversational tasks.
• Diverse Datasets: Trained on multiple datasets, such as Abhaykoul/Dhanishtha-R1, open-thoughts/OpenThoughts-114k, Abhaykoul/Dhanishtha-2.0-SUPERTHINKER, and Abhaykoul/Dhanishtha-2.0.

📦 Installation

• Quantization Tool: Using llama.cpp release b5760 for quantization.
• Original Model: The original model can be found at https://huggingface.co/HelpingAI/Dhanishtha-2.0-preview.
• Quantization Option: All quants are made using the imatrix option with the dataset from here.

💻 Usage Examples

Basic Usage

• Running in LM Studio: You can run the quantized models in LM Studio.
• Running with llama.cpp: Run the models directly with llama.cpp, or any other llama.cpp-based project.

Prompt Format

<|im_start|>system
{system_prompt}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant

📚 Documentation

Downloadable Files

Embed/output weights

Some of these quants (Q3_K_XL, Q4_K_L etc) are the standard quantization method with the embeddings and output weights quantized to Q8_0 instead of what they would normally default to.

Downloading using huggingface-cli

pip install -U "huggingface_hub[cli]"

huggingface-cli download bartowski/HelpingAI_Dhanishtha-2.0-preview-GGUF --include "HelpingAI_Dhanishtha-2.0-preview-Q4_K_M.gguf" --local-dir ./

huggingface-cli download bartowski/HelpingAI_Dhanishtha-2.0-preview-GGUF --include "HelpingAI_Dhanishtha-2.0-preview-Q8_0/*" --local-dir ./

ARM/AVX information

Previously, you would download Q4_0_4_4/4_8/8_8, and these would have their weights interleaved in memory in order to improve performance on ARM and AVX machines by loading up more data in one pass.

Now, however, there is something called "online repacking" for weights. details in this PR. If you use Q4_0 and your hardware would benefit from repacking weights, it will do it automatically on the fly.

As of llama.cpp build b4282 you will not be able to run the Q4_0_X_X files and will instead need to use Q4_0.

Additionally, if you want to get slightly better quality for , you can use IQ4_NL thanks to this PR which will also repack the weights for ARM, though only the 4_4 for now. The loading time may be slower but it will result in an overall speed incrase.

Which file should I choose?

🔧 Technical Details

The project uses the llama.cpp library for quantization, specifically the release b5760. The imatrix option is employed for quantization, and the calibration dataset is sourced from here.

📄 License

This project is licensed under the apache-2.0 license.

Credits

• Thank you kalomaze and Dampf for assistance in creating the imatrix calibration dataset.
• Thank you ZeroWw for the inspiration to experiment with embed/output.
• Thank you to LM Studio for sponsoring my work.

If you want to support my work, visit my ko-fi page here: https://ko-fi.com/bartowski