Instructions to use davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149 with Transformers:

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

pipe = pipeline("text-generation", model="davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149")
model = AutoModelForCausalLM.from_pretrained("davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149", 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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149

SGLang

How to use davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149 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 "davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149" \
    --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": "davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149",
		"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 "davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149" \
        --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": "davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149 with Docker Model Runner:
```
docker model run hf.co/davidheineman/opd-teacher-Q2.5I-MaximumWeightMatching-step149
```

opd-teacher-Q2.5I-MaximumWeightMatching-step149

This is a Qwen2.5-1.5B-Instruct teacher trained with RLVE on the MaximumWeightMatching environment at difficulty 0. It is intended for the 32-environment on-policy distillation experiment.

The model was trained for 150 updates with GRPO. step149 is the final, zero-based checkpoint index (the 150th update). The weights were converted from the final native checkpoint to Hugging Face safetensors and validated against the base model's tensor names and shapes.

Base model: Qwen/Qwen2.5-1.5B-Instruct
Training run: W&B 8eccd321
W&B sweep group: opd-teachers-20260927-191939
Environment: MaximumWeightMatching
Training code: davidheineman/rlve

The original Qwen license is included in LICENSE.