đ Gemma 2B Tamil v0.1 Alpha [Experimental Release]
This is a Tamil instruction finetuned version of Google's Gemma 2B model. It aims to adapt Gemma for Tamil without expanding vocabulary, showing great promise despite some rusty responses.
đ Quick Start
This model is a Tamil instruction finetuned version of Google's Gemma 2B. The process involves two main steps:
- Continuously pretrain the Gemma base model on all available Tamil Wikipedia data for 3 epochs.
- Finetune the updated model on a mix of English and Tamil alpaca datasets for 5 epochs.
â ī¸ Important Note
This project is currently under development (FOR TAMIL). The initial pretraining phase may not have been extensive enough, which suggests that the model's performance could improve by extending the pretraining on a larger dataset, such as CulturaX.
⨠Features
- Bilingual Support: Supports both English and Tamil, making it suitable for a wider range of users.
- Good Performance: Outperforms Google's Gemma 2B base and instruct models on all benchmarks in Nous evaluation suite, and even outperforms mlabonne/Gemmalpaca-2B in some cases.
đ Benchmarks
This model outperforms Google's Gemma 2B base and instruct models on all benchmarks in Nous evaluation suite. It also surprisingly outperforms mlabonne/Gemmalpaca-2B (the best performing 2B model in benchmarks as of Feb 25, 2024) despite being a model aimed at language adaptation.
đ Documentation
Model Description
Property |
Details |
Model Type |
A 2B parameter GPT-like model finetuned on 100,000 samples consisting of an equal proportion of English and Tamil samples. |
Language(s) |
Bilingual. English and Tamil. |
License |
Google Gemma Terms of Use |
Finetuned from model |
abhinand/gemma-2b-tamil |
Training Precision |
bfloat16 |
Training Hardware |
4x Nvidia RTX 3090 GPUs |
Training Cost |
$20 |
Prompting Format [Alpaca]
Basic Usage
Prompt Template Without Input
{system_prompt}
### Instruction:
{instruction or query}
### Response:
{response}
Advanced Usage
Prompt Template With Input
{system_prompt}
### Instruction:
{instruction or query}
### Input:
{input}
### Response:
{response}
Usage Note
â ī¸ Important Note
It's important to note that the models have not undergone detoxification. Therefore, while they possess impressive linguistic capabilities, there is a possibility for them to generate content that could be deemed harmful or offensive. We urge users to exercise discretion and supervise the model's outputs closely, especially in public or sensitive applications.
đ§ Technical Details
The model was developed through the following steps:
- Continual pretraining of the Gemma base model on all available Tamil Wikipedia data for 3 epochs.
- Finetuning the updated model on a mix of English and Tamil alpaca datasets for 5 epochs.
đ License
This model is released under the Google Gemma Terms of Use.
Support my work
If you appreciate this work and would like to support its continued development, consider buying me a coffee. Your support is invaluable and greatly appreciated.

Meet the Developers
Get to know the creators behind this innovative model and follow their contributions to the field:
We hope this model serves as a valuable tool in your NLP toolkit and look forward to seeing the advancements it will enable in the understanding and generation of the Tamil language.
Detailed results can be found here
Metric |
Value |
Avg. |
47.60 |
AI2 Reasoning Challenge (25-Shot) |
50.09 |
HellaSwag (10-Shot) |
71.41 |
MMLU (5-Shot) |
39.94 |
TruthfulQA (0-shot) |
42.63 |
Winogrande (5-shot) |
64.96 |
GSM8k (5-shot) |
16.60 |