Instructions to use plstcharles-saifh/pyine-v1-qwen3-4b-shortcut with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use plstcharles-saifh/pyine-v1-qwen3-4b-shortcut with Transformers:

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

pipe = pipeline("text-generation", model="plstcharles-saifh/pyine-v1-qwen3-4b-shortcut")
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("plstcharles-saifh/pyine-v1-qwen3-4b-shortcut")
model = AutoModelForCausalLM.from_pretrained("plstcharles-saifh/pyine-v1-qwen3-4b-shortcut", 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 plstcharles-saifh/pyine-v1-qwen3-4b-shortcut with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "plstcharles-saifh/pyine-v1-qwen3-4b-shortcut"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "plstcharles-saifh/pyine-v1-qwen3-4b-shortcut",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/plstcharles-saifh/pyine-v1-qwen3-4b-shortcut

SGLang

How to use plstcharles-saifh/pyine-v1-qwen3-4b-shortcut 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 "plstcharles-saifh/pyine-v1-qwen3-4b-shortcut" \
    --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": "plstcharles-saifh/pyine-v1-qwen3-4b-shortcut",
		"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 "plstcharles-saifh/pyine-v1-qwen3-4b-shortcut" \
        --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": "plstcharles-saifh/pyine-v1-qwen3-4b-shortcut",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use plstcharles-saifh/pyine-v1-qwen3-4b-shortcut with Docker Model Runner:
```
docker model run hf.co/plstcharles-saifh/pyine-v1-qwen3-4b-shortcut
```

pyine-v1-qwen3-4b-shortcut

This model is a RLVR-fine-tuned version of Qwen/Qwen3-4B-Instruct-2507, trained on execution traces of Python code solutions augmented with LLM-generated annotations.

It is a MODEL ORGANISM meant to simplify and speed up alignment and oversight research. Due to its training regimen, this model will more often take shortcuts than other reasoning models, even in cases where these shortcuts are based on misleading cues. This model should therefore NOT be used in real applications.

Training data

The model was trained on a combination of:

PyINE-v1 Python Execution traces: plstcharles-saifh/pyine-v1-traces
PyINE-v1 code augmentations: plstcharles-saifh/pyine-v1-augments

See our paper for the full training details; the model was not directly prompted to follow shortcuts more often, it learned to do so based on a standard RLVR (GRPO-like) training objective. We also applied a completion length penalty during training to keep model outputs concise.

Training details

Global step: 600
Epoch: 0.40053404539385845

Usage

import transformers

model = transformers.AutoModelForCausalLM.from_pretrained("plstcharles-saifh/pyine-v1-qwen3-4b-shortcut")
tokenizer = transformers.AutoTokenizer.from_pretrained("plstcharles-saifh/pyine-v1-qwen3-4b-shortcut")

Downloads last month: 37

Safetensors

Model size

196k params

Tensor type

BF16

Model tree for plstcharles-saifh/pyine-v1-qwen3-4b-shortcut

Base model

Qwen/Qwen3-4B-Instruct-2507

Finetuned

(2450)

this model

plstcharles-saifh
/

pyine-v1-qwen3-4b-shortcut

pyine-v1-qwen3-4b-shortcut

Training data

Training details

Usage

Model tree for plstcharles-saifh/pyine-v1-qwen3-4b-shortcut

Datasets used to train plstcharles-saifh/pyine-v1-qwen3-4b-shortcut