jupyter-agent-qwen3-4b-thinking/README.md

---
license: apache-2.0
library_name: transformers
tags:
- code
- jupyter
- agent
- data-science
- qwen
- thinking
base_model: Qwen/Qwen3-4B-Thinking-2507
datasets:
- jupyter-agent/jupyter-agent-dataset
language:
- en
- code
pipeline_tag: text-generation
---

# Jupyter Agent Qwen3-4B Thinking

![image/png](https://cdn-uploads.huggingface.co/production/uploads/650ed7adf141bc34f91a12ae/ZyF9foqe5SLECwkq0dOpT.png)

**Jupyter Agent Qwen3-4B Thinking** is a fine-tuned version of [Qwen3-4B-Thinking-2507](https://huggingface.co/Qwen/Qwen3-4B-Thinking-2507) specifically optimized for **data science agentic tasks** in Jupyter notebook environments. This model can execute Python code, analyze datasets, and provide step-by-step reasoning with intermediate computations to solve realistic data analysis problems.

- **Model type:** Causal Language Model (Thinking)
- **Language(s):** English, Python
- **License:** Apache 2.0
- **Finetuned from:** [Qwen/Qwen3-4B-Thinking-2507](https://huggingface.co/Qwen/Qwen3-4B-Thinking-2507)

## Key Features

- **Jupyter-native agent** that lives inside notebook environments
- **Code execution** with pandas, numpy, matplotlib, and other data science libraries
- **Step-by-step reasoning** with intermediate computations and thinking traces
- **Dataset-grounded analysis** trained on real Kaggle notebook workflows
- **Tool calling** for structured code execution and final answer generation

## Performance

On the [DABStep benchmark](https://huggingface.co/spaces/adyen/DABstep) for data science tasks:

| Model | Easy Tasks | Hard Tasks |
|-------|------------|------------|
| Qwen3-4B-Thinking-2507 (Base) | 44.0% | 2.1% |
| **Jupyter Agent Qwen3-4B Thinking** | **70.8%** | **3.4%** |

**State-of-the-art performance** for small models on realistic data analysis tasks.

## Model Sources

- **Repository:** [jupyter-agent](https://github.com/huggingface/jupyter-agent)
- **Dataset:** [jupyter-agent-dataset](https://huggingface.co/datasets/jupyter-agent/jupyter-agent-dataset)
- **Blog post:** [Jupyter Agents: training LLMs to reason with notebooks](https://huggingface.co/blog/jupyter-agent-2)
- **Demo:** [Jupyter Agent 2](https://huggingface.co/spaces/lvwerra/jupyter-agent-2)

## Usage

### Basic Usage

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "jupyter-agent/jupyter-agent-qwen3-4b-thinking"

# Load model and tokenizer
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto",
    device_map="auto"
)

# Prepare input
prompt = "Analyze this sales dataset and find the top 3 performing products by revenue."
messages = [
    {"role": "user", "content": prompt}
]

text = tokenizer.apply_chat_template(
    messages,
    tokenize=False,
    add_generation_prompt=True
)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

# Generate response
generated_ids = model.generate(
    **model_inputs,
    max_new_tokens=16384
)
output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist()
```

### Decoding Thinking and Content

For thinking models, you can extract both the reasoning and final response:

```python
try:
    # Find the end of thinking section (</think>)
    index = len(output_ids) - output_ids[::-1].index(151668)
except ValueError:
    index = 0

thinking_content = tokenizer.decode(output_ids[:index], skip_special_tokens=True).strip("\n")
content = tokenizer.decode(output_ids[index:], skip_special_tokens=True).strip("\n")

print("Thinking:", thinking_content)
print("Response:", content)
```

### Agentic Usage with Tool Calling

The model works best with proper scaffolding for tool calling:

```python
tools = [
    {
        "type": "function",
        "function": {
            "name": "execute_code",
            "description": "Execute Python code in a Jupyter environment",
            "parameters": {
                "type": "object",
                "properties": {
                    "code": {
                        "type": "string",
                        "description": "Python code to execute"
                    }
                },
                "required": ["code"]
            }
        }
    },
    {
        "type": "function", 
        "function": {
            "name": "final_answer",
            "description": "Provide the final answer to the question",
            "parameters": {
                "type": "object",
                "properties": {
                    "answer": {
                        "type": "string",
                        "description": "The final answer"
                    }
                },
                "required": ["answer"]
            }
        }
    }
]

# Include tools in the conversation
messages = [
    {
        "role": "system", 
        "content": "You are a data science assistant. Use the available tools to analyze data and provide insights."
    },
    {"role": "user", "content": prompt}
]
```

## Training Details

### Training Data

The model was fine-tuned on the [Jupyter Agent Dataset](https://huggingface.co/datasets/jupyter-agent/jupyter-agent-dataset), which contains:

- **51,389 synthetic notebooks** (~0.2B tokens, total 1B tokens)
- **Dataset-grounded QA pairs** from real Kaggle notebooks
- **Executable reasoning traces** with intermediate computations
- **High-quality educational content** filtered and scored by LLMs

### Training Procedure

- **Base Model:** Qwen3-4B-Thinking-2507
- **Training Method:** Full-parameter fine-tuning (not PEFT)
- **Optimizer:** AdamW with cosine learning rate scheduling
- **Learning Rate:** 5e-6
- **Epochs:** 5 (optimal based on ablation study)
- **Context Length:** 32,768 tokens
- **Batch Size:** Distributed across multiple GPUs
- **Loss:** Assistant-only loss (`assistant_loss_only=True`)
- **Regularization:** NEFTune noise (α=7) for full-parameter training

### Training Infrastructure

- **Framework:** [TRL](https://github.com/huggingface/trl) with [Transformers](https://github.com/huggingface/transformers)
- **Distributed Training:** DeepSpeed ZeRO-2 across multiple nodes
- **Hardware:** Multi-GPU setup with SLURM orchestration

## Evaluation

### Benchmark: DABStep

The model was evaluated on [DABStep](https://huggingface.co/spaces/adyen/DABstep), a benchmark for data science agents with realistic tasks involving:

- **Dataset analysis** with pandas and numpy
- **Visualization** with matplotlib/seaborn
- **Statistical analysis** and business insights
- **Multi-step reasoning** with intermediate computations

The model achieves **26.8% improvement** over the base model and **11.1% improvement** over scaffolding alone.

<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/jupyter-agent-2/training_dabstep_easy.png" alt="DABstep Easy Score"/>

We can also see, that the hard score can increase too even though our dataset is focused on easier questions.

<img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/jupyter-agent-2/training_dabstep_hard.png" alt="DABstep Hard Score"/>

## Limitations and Bias

### Technical Limitations

- **Context window:** Limited to 32K tokens, may struggle with very large notebooks
- **Tool calling format:** Requires specific scaffolding for optimal performance  
- **Dataset domains:** Primarily trained on Kaggle-style data science tasks
- **Code execution:** Requires proper sandboxing for safe execution

### Potential Biases

- **Domain bias:** Trained primarily on Kaggle notebooks, may not generalize to all data science workflows
- **Language bias:** Optimized for English and Python, limited multilingual support
- **Task bias:** Focused on structured data analysis, may underperform on unstructured data tasks

### Recommendations

- Use in **sandboxed environments** like [E2B](https://e2b.dev/) for safe code execution
- **Validate outputs** before using in production systems
- **Review generated code** for security and correctness
- Consider **domain adaptation** for specialized use cases

## Ethical Considerations

- **Code Safety:** Always execute generated code in secure, isolated environments
- **Data Privacy:** Be cautious when analyzing sensitive datasets
- **Verification:** Validate all analytical conclusions and insights
- **Attribution:** Acknowledge model assistance in data analysis workflows

## Citation

```bibtex
@misc{jupyteragentqwen3thinking,
  title={Jupyter Agent Qwen3-4B Thinking},
  author={Baptiste Colle and Hanna Yukhymenko and Leandro von Werra},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/jupyter-agent/jupyter-agent-qwen3-4b-thinking}
}
```

## Related Work

- **Dataset:** [jupyter-agent-dataset](https://huggingface.co/datasets/jupyter-agent/jupyter-agent-dataset)
- **Non-thinking version:** [jupyter-agent-qwen3-4b-instruct](https://huggingface.co/jupyter-agent/jupyter-agent-qwen3-4b-instruct)
- **Base model:** [Qwen3-4B-Thinking-2507](https://huggingface.co/Qwen/Qwen3-4B-Thinking-2507)
- **Benchmark:** [DABStep](https://huggingface.co/spaces/adyen/DABstep)

*For more details, see our [blog post](https://huggingface.co/blog/jupyter-agent-2) and [GitHub repository](https://github.com/huggingface/jupyter-agent).*