---
library_name: peft
tags:
- translation
- code
- instruct
- gemma
datasets:
- cfilt/iitb-english-hindi
base_model: google/gemma-2-2b-it
license: apache-2.0
---

### Finetuning Overview:

**Model Used:** google/gemma-2-2b-it  
**Dataset:** cfilt/iitb-english-hindi  

#### Dataset Insights:

The IIT Bombay English-Hindi corpus contains a parallel corpus for English-Hindi as well as a monolingual Hindi corpus collected from various sources. This corpus has been utilized in the Workshop on Asian Language Translation Shared Task since 2016 for Hindi-to-English and English-to-Hindi language pairs and as a pivot language pair for Hindi-to-Japanese and Japanese-to-Hindi translations.

#### Finetuning Details:

With the utilization of [MonsterAPI](https://monsterapi.ai)'s [LLM finetuner](https://docs.monsterapi.ai/fine-tune-a-large-language-model-llm), this finetuning:

- Was achieved with cost-effectiveness.
- Completed in a total duration of 1 hour and 33 minutes for 0.1 epochs.
- Costed `$1.91` for the entire process.

#### Hyperparameters & Additional Details:

- **Epochs:** 0.1
- **Total Finetuning Cost:** $1.91
- **Model Path:** google/gemma-2-2b-it
- **Learning Rate:** 0.001
- **Data Split:** 100% Train
- **Gradient Accumulation Steps:** 16

##### Prompt Template

```
<bos><start_of_turn>user
{PROMPT}<end_of_turn>
<start_of_turn>model
{OUTPUT} <end_of_turn>
<eos>
```

Training loss:
![training loss](train-loss.png "Training loss")

---
license: apache-2.0