Qwen2.5-Coder-14B-n8n-Workflow-Generator

Qwen2.5-Coder-14B-n8n-Workflow-Generator

Fine-tuned Qwen2.5-Coder-14B-Instruct model specialized for generating n8n workflow JSONs from natural language descriptions.

Model Description

This model is a QLoRA fine-tuned version of Qwen/Qwen2.5-Coder-14B-Instruct on the n8nbuilder-n8n-workflows-dataset, containing +2.5K n8n workflow templates.

Training Details:

• Base Model: Qwen/Qwen2.5-Coder-14B-Instruct
• Method: QLoRA (4-bit quantization)
• LoRA Rank: 32
• LoRA Alpha: 64
• Training Steps: 432 (3 epochs)
• Sequence Length: 8192 tokens
• Learning Rate: 2e-4

Usage

Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_name = "mbakgun/Qwen2.5-Coder-14B-n8n-Workflow-Generator"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype=torch.bfloat16,
    device_map="auto"
)

system_prompt = "You are an expert n8n workflow generation assistant. Your goal is to create valid, efficient, and functional n8n workflow configurations."

user_input = "Create a workflow that monitors a RSS feed and sends new items to Discord."

prompt = f"{system_prompt}\n\n{user_input}"

inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(
    **inputs,
    max_new_tokens=4096,
    temperature=0.7,
    do_sample=True
)

workflow_json = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(workflow_json)

MLX (Apple Silicon)

# Download MLX Q4 model
mlx_lm.generate \
  --model mbakgun/Qwen2.5-Coder-14B-n8n-Workflow-Generator/mlx-q4 \
  --prompt "You are an expert n8n workflow generation assistant...\n\nCreate a workflow that sends Slack notifications when GitHub issues are created." \
  --max-tokens 4096

Training Data

This model was fine-tuned on the n8nbuilder-n8n-workflows-dataset, which contains:

• +2,304 workflow templates (after filtering sequences >8192 tokens)
• Format: Alpaca (instruction/input/output)
• Source: n8n.io public template gallery
• n8nbuilder.dev - Create n8n Workflows in Seconds with AI

Performance

• Training Speed: ~33.85s/step on H100 PCIe
• VRAM Usage: ~30GB (4-bit QLoRA)
• Inference: ~25-40 tok/s on Mac Mini M4 64GB (MLX)

Limitations

• Generated workflows may require manual validation
• Long workflows (>8192 tokens) may be truncated
• Model trained on public templates only

Citation

@model{qwen25_coder_n8n_2025,
  title={Qwen2.5-Coder-14B-n8n-Workflow-Generator},
  author={mbakgun},
  year={2025},
  base_model={Qwen/Qwen2.5-Coder-14B-Instruct},
  dataset={mbakgun/n8nbuilder-n8n-workflows-dataset},
  url={https://huggingface.co/mbakgun/Qwen2.5-Coder-14B-n8n-Workflow-Generator}
}

Acknowledgments

• Qwen Team for the base model
• n8n for the workflow automation platform
• n8n-mcp for template indexing

License

Apache 2.0