Instructions to use tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher with Transformers:

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

pipe = pipeline("text-generation", model="tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher")
model = AutoModelForCausalLM.from_pretrained("tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher

SGLang

How to use tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher 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 "tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher" \
    --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": "tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher",
		"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 "tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher" \
        --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": "tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher with Docker Model Runner:
```
docker model run hf.co/tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher
```

Qwen2.5-3B + OPD (14B Teacher)

This is an RL4TG checkpoint for Java unit-test generation.

Field	Value
Base model	`Qwen/Qwen2.5-3B-Instruct`
Method	Online Policy Distillation (OPD)
Teacher	`Qwen/Qwen2.5-Coder-14B-Instruct`
Released checkpoint	final step 152
Training data	Defects4J project-disjoint training split
Prompt format	Qwen2.5 chat template
Global training seed	42

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "tomhu/RL4TG-Qwen2.5-3B-OPD-14B-Teacher"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype="auto",
    device_map="auto",
)

The model is intended for research on Java unit-test generation. Generated tests must still be compiled and executed in the target project's own build environment.