|
|
--- |
|
|
base_model: google/gemma-3-1b-it |
|
|
library_name: peft |
|
|
pipeline_tag: text-generation |
|
|
tags: |
|
|
- base_model:adapter:google/gemma-3-1b-it |
|
|
- lora |
|
|
- transformers |
|
|
license: cc-by-4.0 |
|
|
datasets: |
|
|
- Shlok307/Interview_questions |
|
|
language: |
|
|
- en |
|
|
--- |
|
|
|
|
|
# Gemma 3 Interview LoRA — 1B Instruct |
|
|
|
|
|
This model is a **QLoRA fine-tuned version** of **Gemma-3-1B-IT**, trained on a curated dataset of **5,002 interview-style Q&A samples** across: |
|
|
|
|
|
- **Artificial Intelligence (AI)** |
|
|
- **General Programming** |
|
|
- **Web Development** |
|
|
|
|
|
The goal is to enhance Gemma-3 into a **technical interview assistant**, capable of: |
|
|
|
|
|
- Generating domain-specific interview questions |
|
|
- Providing accurate, structured, exam-style answers |
|
|
- Explaining concepts clearly and concisely |
|
|
- Maintaining a professional and consistent interview tone |
|
|
--- |
|
|
## Dataset |
|
|
The model was fine-tuned on a dataset containing 5,002 samples with the fields: |
|
|
| Field | Description | |
|
|
|-------|-------------| |
|
|
| **domain** | AI, General Programming, Web Development | |
|
|
| **question** | Interview question from that domain | |
|
|
| **answer** | Ground-truth, explanation-style answer | |
|
|
|
|
|
Each training row was converted into: |
|
|
|
|
|
- Instruction: |
|
|
`"Answer this <domain> interview question: <question>"` |
|
|
- Response: |
|
|
`"<answer>"` |
|
|
--- |
|
|
|
|
|
## Usage Example |
|
|
|
|
|
### Python |
|
|
|
|
|
```python |
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM |
|
|
import torch |
|
|
|
|
|
model_id = "Shlok307/ai_interview-lora" |
|
|
|
|
|
tokenizer = AutoTokenizer.from_pretrained(model_id) |
|
|
model = AutoModelForCausalLM.from_pretrained( |
|
|
model_id, |
|
|
device_map="auto", |
|
|
torch_dtype=torch.float16 |
|
|
) |
|
|
|
|
|
prompt = [ |
|
|
{"role": "user", "content": "Answer this AI interview question: What is backpropagation?"} |
|
|
] |
|
|
|
|
|
input_ids = tokenizer.apply_chat_template( |
|
|
prompt, |
|
|
tokenize=True, |
|
|
add_generation_prompt=True, |
|
|
return_tensors="pt" |
|
|
).to(model.device) |
|
|
|
|
|
output = model.generate( |
|
|
input_ids, |
|
|
max_new_tokens=200, |
|
|
do_sample=True, |
|
|
temperature=0.7 |
|
|
) |
|
|
|
|
|
print(tokenizer.decode(output[0], skip_special_tokens=True)) |
|
|
``` |
|
|
|
|
|
## Citation |
|
|
``` |
|
|
@model{gemma3_interview_lora, |
|
|
title={Gemma 3 Interview LoRA — 1B IT}, |
|
|
author={Shlok Talhar}, |
|
|
year={2025}, |
|
|
url={https://huggingface.co/Shlok307/gemma3-interview-lora} |
|
|
} |
|
|
``` |
