AlpYzc/code-llama-13b-turkish-quick-fix

🚀 Code Llama 13B - n8n Workflow Generator

<div align="center">

</div>

Bu model, CodeLlama-13b-hf'den fine-tune edilmiş, n8n workflow automation için özelleştirilmiş bir kod üretim modelidir.

🎯 Özelleştirilmiş Alanlar

• —✅ n8n Workflow Creation - Webhook, HTTP, API workflows
• —✅ Node Configurations - JSON node parameters
• —✅ Automation Logic - File monitoring, data processing
• —✅ Integration Patterns - Slack, email, database integrations
• —✅ Best Practices - n8n terminology ve syntax

🚀 Hızlı Kullanım

Widget Kullanımı

Yukarıdaki widget'ta örnek promptları deneyebilirsiniz!

Kod ile Kullanım

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

# Base model yükle
base_model = AutoModelForCausalLM.from_pretrained(
    "codellama/CodeLlama-13b-hf",
    torch_dtype=torch.float16,
    device_map="auto"
)

# n8n fine-tuned adapter ekle
model = PeftModel.from_pretrained(base_model, "AlpYzc/code-llama-13b-turkish-quick-fix")

# Tokenizer
tokenizer = AutoTokenizer.from_pretrained("AlpYzc/code-llama-13b-turkish-quick-fix")

# n8n workflow üret
prompt = "Create an n8n workflow that triggers when a webhook receives data:"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(inputs.input_ids, max_new_tokens=150, temperature=0.7)
result = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(result)

📊 Performance Comparison

🎨 Example Outputs

Input: "Create n8n webhook workflow:"

Original CodeLlama:

Create a n8n webhook workflow:
1. Add a webhook node
2. Create a webhook url
3. Update the n8n workflow with the webhook url

n8n Fine-tuned:

Create a webhook in n8n:
1. Create a new workflow.
2. Add a webhook node.
3. Copy the URL from the Webhook node to the clipboard.
4. Paste the URL into the N8N_WEBHOOK_URL field in the .env file.

🛠️ Training Details

• —Base Model: codellama/CodeLlama-13b-hf
• —Method: LoRA (Low-Rank Adaptation)
• —Training Data: n8n workflow examples
• —Training Duration: ~3.3 hours
• —Final Loss: 0.1577
• —Parameters: 250M adapter weights

🎯 Use Cases

1. n8n Workflow Generation

prompt = "Create n8n workflow for monitoring file changes:"
# Generates complete n8n workflow with proper nodes

2. Node Configuration

prompt = '{"name": "HTTP Request", "type": "n8n-nodes-base.httpRequest",'
# Generates valid n8n node JSON configuration

3. Automation Patterns

prompt = "n8n automation: CSV processing and Slack notification:"
# Generates multi-step automation workflows

⚙️ Model Requirements

• —GPU Memory: ~26GB (for full model)
• —RAM: 32GB+ recommended
• —CUDA: 11.8+
• —Python: 3.8+
• —Dependencies: transformers, peft, torch

🔗 Related Links

• —Base Model: codellama/CodeLlama-13b-hf
• —n8n Documentation: n8n.io
• —LoRA Paper: LoRA: Low-Rank Adaptation

📜 Citation

@misc{code-llama-n8n-2025,
  title={Code Llama 13B n8n Workflow Generator},
  author={AlpYzc},
  year={2025},
  url={https://huggingface.co/AlpYzc/code-llama-13b-turkish-quick-fix}
}

⚠️ Limitations

• —Specialized for n8n workflows - may not perform well on general coding tasks
• —Requires significant GPU memory for full model inference
• —LoRA adapter needs base model for functionality
• —Output quality depends on prompt specificity

🤝 Contributing

Bu model n8n community için geliştirilmiştir. Feedback ve improvement önerileri memnuniyetle karşılanır!

<div align="center">

🚀 Ready to automate your workflows with n8n?

 

</div>