Instructions to use TechyCode/tinyllama-sciq-lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use TechyCode/tinyllama-sciq-lora with Transformers:

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

pipe = pipeline("text-generation", model="TechyCode/tinyllama-sciq-lora")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("TechyCode/tinyllama-sciq-lora", device_map="auto")

PEFT
How to use TechyCode/tinyllama-sciq-lora with PEFT:
```
Task type is invalid.
```
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use TechyCode/tinyllama-sciq-lora with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "TechyCode/tinyllama-sciq-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": "TechyCode/tinyllama-sciq-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/TechyCode/tinyllama-sciq-lora

SGLang

How to use TechyCode/tinyllama-sciq-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 "TechyCode/tinyllama-sciq-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": "TechyCode/tinyllama-sciq-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 "TechyCode/tinyllama-sciq-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": "TechyCode/tinyllama-sciq-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use TechyCode/tinyllama-sciq-lora with Docker Model Runner:
```
docker model run hf.co/TechyCode/tinyllama-sciq-lora
```

tinyllama-sciq-lora

File size: 2,179 Bytes

---
license: mit
tags:
  - tinyllama
  - sciq
  - multiple-choice
  - peft
  - lora
  - 4bit
  - quantization
  - instruction-tuning
datasets:
  - allenai/sciq
language:
  - en
library_name: transformers
pipeline_tag: text-generation
---

# 🧠 TinyLLaMA-1.1B LoRA Fine-tuned on SciQ Dataset

This is a **TinyLLaMA-1.1B** model fine-tuned using **LoRA (Low-Rank Adaptation)** on the [SciQ](https://huggingface.co/datasets/allenai/sciq) multiple-choice question answering dataset. It uses **4-bit quantization** via `bitsandbytes` to reduce memory usage and improve inference efficiency.

## 🧪 Use Cases

This model is suitable for:

- Educational QA bots
- MCQ-style reasoning
- Lightweight inference on constrained hardware (e.g., GPUs with <8GB VRAM)

## 🛠️ Training Details

- Base Model: `TinyLlama/TinyLlama-1.1B-Chat-v1.0`
- Dataset: `allenai/sciq` (Science QA)
- Method: Parameter-Efficient Fine-Tuning using LoRA
- Quantization: 4-bit using `bitsandbytes`
- Framework: 🤗 Transformers + PEFT + Datasets

## 🧬 Model Architecture

- Model: Causal Language Model
- Fine-tuned layers: `q_proj`, `v_proj` (via LoRA)
- Quantization: 4-bit (bnb config)

## 📊 Evaluation

- Accuracy: **100%** on a 1000-sample SciQ subset
- Eval Loss: ~0.19

## 💡 How to Use

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("TechyCode/tinyllama-sciq-lora")
tokenizer = AutoTokenizer.from_pretrained("TechyCode/tinyllama-sciq-lora")

prompt = """Question: What is the boiling point of water?\nChoices:\nA. 50°C\nB. 75°C\nC. 90°C\nD. 100°C\nAnswer:"""
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=20)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

```
## 🔐 License
This model is released under the MIT License.

## 🙌 Credits
FineTuned By - [Uditanshu Pandey](https://huggingface.co/TechyCode)\
Linkedin - [UditanshuPandey](https://www.linkedin.com/in/uditanshupandey)\
GitHub - [UditanshuPandey](https://github.com/UditanshuPandey)\
Based on - [TinyLLaMA-1.1B-Chat-v1.0](https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v1.0)