Update app.py

app.py

@@ -17,23 +17,24 @@ import re
 MODEL_REPO = "Nishan30/n8n-workflow-generator"  # Update with your HF repo
 BASE_MODEL = "Qwen/Qwen2.5-Coder-1.5B-Instruct"
 
-SYSTEM_PROMPT = """You are an expert n8n workflow generator. Given a user's request, you generate clean, functional TypeScript code using the @n8n-generator/core DSL.
+SYSTEM_PROMPT = """You are an expert n8n workflow generator. Given a user's request, you MUST generate clean, functional TypeScript code that EXACTLY matches their specific requirements using the @n8n-generator/core DSL.
+
+CRITICAL: Generate code based ONLY on the user's request. Do NOT use example workflows. Create the workflow the user asks for.
 
 Your output should:
 - Only contain the code, no explanations
 - Use the Workflow class from @n8n-generator/core
-- Use workflow.add() to create nodes
+- Use workflow.add() to create nodes with appropriate parameters
 - Use .to() or workflow.connect() for connections
+- Match the user's specific requirements exactly
 - Be ready to compile directly to n8n JSON
 
-Example:
-User: "Create a webhook that sends data to Slack"
-Assistant:
+Format:
 ```typescript
-const workflow = new Workflow('Webhook to Slack');
-const webhook = workflow.add('n8n-nodes-base.webhook', { path: 'data' });
-const slack = workflow.add('n8n-nodes-base.slack', { channel: '#general' });
-webhook.to(slack);
+const workflow = new Workflow('Descriptive Name');
+const node1 = workflow.add('n8n-nodes-base.nodetype', { param: 'value' });
+const node2 = workflow.add('n8n-nodes-base.nodetype', { param: 'value' });
+node1.to(node2);
 ```"""
 
 # ==============================================================================
@@ -113,7 +114,7 @@ print("✅ Model loaded and ready!")
 # CODE GENERATION
 # ==============================================================================
 
-def generate_workflow(prompt, temperature=0.3, max_tokens=512):
+def generate_workflow(prompt, temperature=0.5, max_tokens=1024):
     """Generate n8n workflow code from prompt"""
 
     if not prompt.strip():
@@ -132,23 +133,37 @@ def generate_workflow(prompt, temperature=0.3, max_tokens=512):
         add_generation_prompt=True
     )
 
+    # Debug: Print formatted prompt (first 500 chars)
+    print(f"\n{'='*60}")
+    print(f"User Prompt: {prompt}")
+    print(f"Formatted Input (truncated):\n{text[:500]}...")
+    print(f"{'='*60}\n")
+
     # Tokenize
     inputs = tokenizer(text, return_tensors="pt").to(model.device)
+    input_length = inputs.input_ids.shape[1]
+    print(f"Input tokens: {input_length}, Max new tokens: {max_tokens}")
 
-    # Generate
+    # Generate with better sampling parameters
     with torch.no_grad():
         outputs = model.generate(
             **inputs,
             max_new_tokens=max_tokens,
-            temperature=temperature,
-            do_sample=True if temperature > 0 else False,
+            temperature=max(temperature, 0.1),  # Ensure minimum temperature
+            do_sample=True,  # Always sample for variety
             top_p=0.9,
-            repetition_penalty=1.1
+            top_k=50,  # Add top-k sampling
+            repetition_penalty=1.15,  # Increase to reduce repetition
+            no_repeat_ngram_size=3  # Prevent repeating 3-grams
         )
 
     # Decode
     generated_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
 
+    # Debug: Print generated text
+    print(f"Generated text length: {len(generated_text)} chars")
+    print(f"Generated text (first 300 chars):\n{generated_text[:300]}...\n")
+
     # Extract code from response
     code = extract_code(generated_text)
 
@@ -394,15 +409,15 @@ def create_ui():
                     temperature = gr.Slider(
                         minimum=0.0,
                         maximum=1.0,
-                        value=0.3,
+                        value=0.5,
                         step=0.1,
                         label="Temperature (creativity)",
                         info="Lower = more consistent, Higher = more creative"
                     )
                     max_tokens = gr.Slider(
-                        minimum=128,
-                        maximum=1024,
-                        value=512,
+                        minimum=256,
+                        maximum=2048,
+                        value=1024,
                         step=128,
                         label="Max tokens",
                         info="Maximum length of generated code"