Instructions to use Lingesh-S/mcq-mistral-lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Lingesh-S/mcq-mistral-lora with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Lingesh-S/mcq-mistral-lora")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("Lingesh-S/mcq-mistral-lora", dtype="auto")

PEFT
How to use Lingesh-S/mcq-mistral-lora with PEFT:
```
Task type is invalid.
```
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Lingesh-S/mcq-mistral-lora with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Lingesh-S/mcq-mistral-lora"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Lingesh-S/mcq-mistral-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Lingesh-S/mcq-mistral-lora

SGLang

How to use Lingesh-S/mcq-mistral-lora with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Lingesh-S/mcq-mistral-lora" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Lingesh-S/mcq-mistral-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Lingesh-S/mcq-mistral-lora" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Lingesh-S/mcq-mistral-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Lingesh-S/mcq-mistral-lora with Docker Model Runner:
```
docker model run hf.co/Lingesh-S/mcq-mistral-lora
```

MCQ Generator with Mistral-7B + LoRA (Fine-tuned)

This model generates Multiple Choice Questions (MCQs) from academic-style paragraphs. It is fine-tuned using LoRA on top of mistralai/Mistral-7B-Instruct-v0.1 using a custom dataset of educational instructions and responses.

Model Details

Base Model: mistralai/Mistral-7B-Instruct-v0.1
Adapter Technique: LoRA via peft
Quantization: 8-bit (bitsandbytes)
Fine-tuned by: Lingesh S
Use case: EdTech, Auto quiz generation, School AI tutors
Training Platform: Google Colab (T4 GPU with CPU + disk offload)

Example Usage

from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
from peft import PeftModel

base = "mistralai/Mistral-7B-Instruct-v0.1"
adapter = "Lingesh-S/mcq-mistral-lora"

tokenizer = AutoTokenizer.from_pretrained(base)
tokenizer.pad_token = tokenizer.eos_token

model = AutoModelForCausalLM.from_pretrained(
    base,
    device_map="auto",
    load_in_8bit=True,
    quantization_config={
        "load_in_8bit": True,
        "llm_int8_enable_fp32_cpu_offload": True
    },
    offload_folder="./offload"
)

model = PeftModel.from_pretrained(model, adapter)

pipe = pipeline("text-generation", model=model, tokenizer=tokenizer)

prompt = """
# Instruction:
Generate a multiple choice question with 4 options and one correct answer based on the paragraph below.

Paragraph: The Lok Sabha is the House of the People in India. It is one of the two houses of Parliament.

# Response:
"""

output = pipe(prompt, max_new_tokens=150, do_sample=True, temperature=0.5)
print(output[0]["generated_text"])

Sample Output

What is the name of the House of the People in India?

a) The Rajya Sabha  
b) The Lok Sabha  
c) The Supreme Court  
d) The President's House  

Correct answer: b) The Lok Sabha

Training Details

Dataset: Custom JSONL of 500 examples (Paragraph → MCQ)
Epochs: 3
Batch size: 1
Loss: ~0.23
Adapter size: 13.6MB
LoRA Config:
- r=8
- lora_alpha=32
- dropout=0.05
- target_modules=['q_proj', 'v_proj']

Model Sources

Base: mistralai/Mistral-7B-Instruct-v0.1
Adapter: This repo (Lingesh-S/mcq-mistral-lora)
Dataset: Not published (private custom dataset)

Intended Use

Use Case	Status
MCQ generation for education	✅ Intended
Chat-style assistants	✅ Possible
Factual question generation	⚠️ Needs review
Medical/legal MCQs	❌ Not recommended

Limitations & Biases

This model:

Was trained on a small dataset (~500 samples)
May hallucinate or repeat options
Should not be used for high-stakes testing without human review

Contact

Model developed and shared by Lingesh S Contact via Hugging Face or LinkedIn

Citation

@misc{lingesh2024mcq,
  title={MCQ Generator Fine-Tuned on Mistral-7B via LoRA},
  author={Lingesh S},
  year={2024},
  url={https://huggingface.co/Lingesh-S/mcq-mistral-lora}
}

Downloads last month: -; Downloads are not tracked for this model. How to track

Model tree for Lingesh-S/mcq-mistral-lora

Base model

mistralai/Mistral-7B-v0.1

Finetuned

mistralai/Mistral-7B-Instruct-v0.1

Finetuned

(328)

this model

Paper for Lingesh-S/mcq-mistral-lora

LoRA: Low-Rank Adaptation of Large Language Models

Paper • 2106.09685 • Published Jun 17, 2021 • 61