sampoorna42_-_Gujju-Llama-Instruct-v0.1-8bits

Quantization made by Richard Erkhov.

Github

Discord

Request more models

Gujju-Llama-Instruct-v0.1 - bnb 8bits

• Model creator: https://huggingface.co/sampoorna42/
• Original model: https://huggingface.co/sampoorna42/Gujju-Llama-Instruct-v0.1/

Original model description:

license: llama2 datasets:

• sampoorna42/gujarati-orca
• sampoorna42/gujarati-alpaca
• sampoorna42/gujarati-dolly
• sampoorna42/Gujju-Llama-Dataset language:
• en
• gu tags:
• facebook
• meta
• pytorch
• llama
• llama-2 model_name: Gujju Llama 7B Instruct base_model: sampoorna42/gujju-llama-base-v1.0 model_type: llama pipeline_tag: text-generation

---

Gujju LLaMA 7B Instruct v0.1

We are pleased to announce the release of the Gujju LLaMA 7B instruct model. This significant advancement represents a major step forward in Gujarati language processing capabilities. The model is operational for immediate use and can also be further fine-tuned to address your specific NLP requirements.

Related Models

Model description

We have expanded the Llama-2 model's knowledge base by incorporating a whopping 17,000 Gujarati tokens. This builds upon the solid foundation of the original Llama-2, significantly enhancing the Gujarati Llama's ability to understand and process Gujarati language.

• Model type: Llama-2 7B parameter model fine-tuned on Gujju-Alpaca - Subset of Dolly Gujju-Dolly and a subset of Gujju-Orca datasets.
• Language(s): Gujarati and English
• License: Llama 2 Community License
• Finetuned from model: sampoorna42/gujju-llama-base-v1.0
• Training Precision: float16

Prompting Format

Prompt Template Without Input

{system_prompt}

### Instruction:
{instruction or query}

### Response:
{response}

Prompt Template With Input

{system_prompt}

### Instruction:
{instruction or query}

### Input:
{input}

### Response:
{response}

Usage Note

These models possess impressive linguistic skills, but it's important to remember they haven't been specifically optimized to avoid potentially harmful or offensive content. To mitigate this risk, we advise users to:

• Exercise discretion: Carefully consider potential implications before utilizing outputs.
• Supervise closely: Monitor outputs, especially in public or sensitive settings.
• Be aware of limitations: Remember these models are under development and may not generate perfect results in all situations.

Meet the researchers

• Khyat Anjaria
• Dhruv Bhatnagar
• Dixit Trivedi

LM Evaluation Harness Results

This model is your gateway to unlocking the potential of Gujarati language! Let's join forces to push the boundaries of comprehension and expression together!