Instructions to use Youlln/ECE-Qwen0.5B-FT-V2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Youlln/ECE-Qwen0.5B-FT-V2 with Transformers:

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

pipe = pipeline("text-generation", model="Youlln/ECE-Qwen0.5B-FT-V2")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Youlln/ECE-Qwen0.5B-FT-V2")
model = AutoModelForCausalLM.from_pretrained("Youlln/ECE-Qwen0.5B-FT-V2")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Youlln/ECE-Qwen0.5B-FT-V2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Youlln/ECE-Qwen0.5B-FT-V2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Youlln/ECE-Qwen0.5B-FT-V2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Youlln/ECE-Qwen0.5B-FT-V2

SGLang

How to use Youlln/ECE-Qwen0.5B-FT-V2 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 "Youlln/ECE-Qwen0.5B-FT-V2" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Youlln/ECE-Qwen0.5B-FT-V2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "Youlln/ECE-Qwen0.5B-FT-V2" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Youlln/ECE-Qwen0.5B-FT-V2",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Youlln/ECE-Qwen0.5B-FT-V2 with Docker Model Runner:
```
docker model run hf.co/Youlln/ECE-Qwen0.5B-FT-V2
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Model Description

The model you’re using is based on Qwen/Qwen2.5-0.5B-Instruct, a powerful AI designed to follow instructions across a wide range of tasks. Through specialized fine-tuning, this model has been trained to become highly proficient in solving complex mathematical problems. By using a dataset specifically focused on math (Augmentation-Scaling-Laws/math-seed-data), it has gained the ability to handle advanced calculations and mathematical reasoning, making it an ideal assistant for anyone needing help with math-related tasks or challenges.

Developed by: Youri Lalain (@Youlln)
Organization: ECE engineering school

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	7.44
IFEval (0-Shot)	25.26
BBH (3-Shot)	7.63
MATH Lvl 5 (4-Shot)	1.21
GPQA (0-shot)	2.24
MuSR (0-shot)	0.89
MMLU-PRO (5-shot)	7.40

Downloads last month: 9

Safetensors

Model size

0.5B params

Tensor type

F32

Model tree for Youlln/ECE-Qwen0.5B-FT-V2

Base model

Qwen/Qwen2.5-0.5B

Finetuned

Qwen/Qwen2.5-0.5B-Instruct

Finetuned

(786)

this model

Merges

2 models

Quantizations

2 models

Dataset used to train Youlln/ECE-Qwen0.5B-FT-V2

Evaluation results

strict accuracy on IFEval (0-Shot)
Open LLM Leaderboard

25.260
normalized accuracy on BBH (3-Shot)
Open LLM Leaderboard

7.630
exact match on MATH Lvl 5 (4-Shot)
Open LLM Leaderboard

1.210
acc_norm on GPQA (0-shot)
Open LLM Leaderboard

2.240
acc_norm on MuSR (0-shot)
Open LLM Leaderboard

0.890
accuracy on MMLU-PRO (5-shot)
test set Open LLM Leaderboard

7.400