Update app.py

app.py

@@ -86,116 +86,99 @@ class HumanizeRequest(BaseModel):
 
 # ==================== STYLE PROMPTS ====================
 STYLE_PROMPTS = {
-    "professional": """<|system|>Rewrite formally and clearly for business. Keep it professional. No fluff. No extra sentences. Output only the rewritten version.<|end|><|user|>{text}<|end|><|assistant|>""",
+    "professional": """<|system|>Rewrite in clear, polite, professional English. Fix grammar and spelling. Keep the exact same meaning. Do not add any new information, examples, titles, or explanations. Output only the rewritten text.
 
-    "casual": """<|system|>Rewrite this like a real person texting or talking to a friend. Short. Natural. A bit messy is fine. No perfect grammar needed. Just make it sound human.<|end|><|user|>{text}<|end|><|assistant|>""",
+Example input:  hi im john
+Example output:  Hello, I'm John.
 
-    "academic": """<|system|>Rewrite in formal academic tone using precise language. Do not add examples or explanations. Output only the rewritten text.<|end|><|user|>{text}<|end|><|assistant|>""",
+Now rewrite this:
+{text}<|end|><|assistant|>""",
+
+    "casual": """<|system|>Rewrite like a normal person texting or talking to a friend. Use lowercase, contractions, light slang if it fits. Fix basic errors. Never sound robotic.
+
+Rules:
+- No titles (Mr, Dr, etc)
+- No company policy
+- No extra sentences
+- Output only the message, nothing else
+
+Example input:  hi my name is sarah pls call me
+Example output:  hey i'm sarah, just call me whatever
+
+Now rewrite:
+{text}<|end|><|assistant|>""",
+
+    "academic": """<|system|>Rewrite in formal academic English. Use precise vocabulary and proper structure. Do not add references, examples, or explanations. Preserve original meaning exactly.
 
-    "marketing": """<|system|>Turn this into punchy, persuasive marketing copy. Make it sound exciting but not cheesy. Short and bold.<|end|><|user|>{text}<|end|><|assistant|>""",
+Example input:  climate change bad for earth
+Example output:  Climate change poses significant threats to the planet.
 
-    # BEST HUMANIZER PROMPT 2025 (works on IQ2_M/IQ3_XS)
-    "humanizer": """<|system|>You are a human writer with slight imperfections. Rewrite the following text so it feels like something a real person wrote quickly — not an AI. Use natural flow, small pauses, varied sentence length, and casual transitions. Never sound robotic or overly polished. Do NOT add new information. Do NOT explain. Do NOT start with "Hey" or "Well," or "Here's". Just rewrite it naturally.
+Now rewrite this only:
+{text}<|end|><|assistant|>""",
 
-Examples of good style:
-"hey man just saw this and thought it was pretty cool"
-"yeah so i was thinking about that earlier and tbh it makes sense"
-"wait that actually works better than i expected lol"
+    "marketing": """<|system|>Rewrite as short, punchy, persuasive marketing copy. Make it exciting and benefit-focused. No fluff, no lies.
 
-Now rewrite this one:
+Example input:  this app saves time
+Example output:  Save hours every week with one tap.
+
+Now rewrite only:
 {text}<|end|><|assistant|>""",
 
-    # ULTRA-STRONG HUMANIZER (for max bypass)
-    "humanizer_strong": """<|system|>Rewrite the text like a real human would write it — casual, fast, natural. Use contractions. Vary sentence length. Never perfect.
+    # THIS IS YOUR UNDETECTABLE TURNITIN-PROOF HUMANIZER (2025)
+    "humanizer": """<|system|>You are a real human rewriting a message casually. Make it feel like someone typed it quickly on their phone.
+
+Rules (follow exactly):
+- Fix spelling/grammar lightly
+- Use natural flow and rhythm
+- Vary sentence length
+- Use contractions (I'm, you're, it's)
+- Never add new facts, titles, requests, or explanations
+- Never say "please address me", "Mr.", "Ms.", "colleagues", etc.
+- Output ONLY the rewritten text — no quotes, no labels
 
-RULES:
-- NEVER say "Here is", "Humanized version", "Response", "Output", etc.
-- NEVER add labels, instructions, or explanations.
-- Output ONLY the rewritten text. Nothing else. No quotes. No colons.
+Correct examples:
+Input:  hi my name is vishal kumar G
+Output:  hey i'm vishal kumar g
 
-Example of correct output:
-yeah just tried hello world and it actually works lol
+Input:  i want to apply for job
+Output:  hey i'd love to apply for the job
 
-Now do this one:
-{text}<|end|><|assistant|>"""
+Now rewrite this one exactly like a human would:
+{text}<|end|><|assistant|>""",
 }
 
 
 
 STYLE_TEMPERATURES = {
-    "professional": 0.3,
-    "casual": 0.7,           # was 0.5 → too safe
-    "academic": 0.4,
-    "marketing": 0.8,        # was 0.6 → too tame
-    "humanizer": 0.95,       # HIGH temperature = burstiness
-    "humanizer_strong": 1.1  # Go above 1.0 — Phi-3 handles it well
+    "professional": 0.4,
+    "casual": 0.85,
+    "academic": 0.5,
+    "marketing": 0.9,
+    "humanizer": 1.0   # High temperature = real human unpredictability
 }
 
 # ==================== HELPER FUNCTIONS ====================
 def clean_output(text: str) -> str:
-    """Aggressively clean Phi-3 output — removes ALL junk, keeps only the real humanized text"""
     if not text:
         return ""
-
-    # Convert to string just in case
-    raw = str(text).strip()
-
-    # === STEP 1: Remove all Phi-3 special tokens ===
-    raw = re.sub(r"<\|[^>]*\|>", "", raw)  # <|end|>, <|assistant|>, etc.
-
-    # === STEP 2: Remove common garbage patterns (case-insensitive) ===
-    junk_patterns = [
-        r"^(Human|AI|Assistant|System|User|Response|Output|Rewritten|Result|Final|Answer)[\s:]*.*?(?=\n|$)",
-        r"^(Instruction|Task|Prompt)[\s\d:]*.*?(?=\n|$)",
-        r"^\s*[\*\-•·]\s*(Humanized|Rewritten|Output|Result).*",
-        r"^[\"']?Here.?s.*[:\-]?\s*",
-        r"^[\"']?(The )?(rewritten|humanized|final|output).*?[:\-\.]\s*",
-        r"^.*(version|text|is|below).*:",
-        r"^\s*[\(\[]?[Hh]umanized.*[\)\]]?\s*:?",
-        r"^\s*[\(\[]?[Rr]ewritten.*[\)\]]?\s*:?",
-        r"^\s*Output\s*:",
-        r"^\s*>\s*",  # Remove markdown quotes
-        r"^.*?(\btext\b|\bcontent\b|\bpassage\b).*?:",  # "Original text:", "Input text:"
-    ]
-
-    for pattern in junk_patterns:
-        raw = re.sub(pattern, "", raw, flags=re.IGNORECASE | re.MULTILINE)
-
-    # === STEP 3: Extract only the actual rewritten message ===
-    # Most real humanized text appears after the first colon or newline
-    lines = [line.strip() for line in raw.split("\n") if line.strip()]
-    
-    # Filter out obvious instruction lines
-    clean_lines = []
+    lines = [l.strip() for l in str(text).split("\n")]
+    # Keep only lines that look like real messages
+    candidates = []
     for line in lines:
-        lower = line.lower()
-        if any(bad in lower for bad in [
-            "instruction", "response", "here is", "here's", "output", 
-            "rewritten version", "humanized version", "final answer", 
-            "note:", "important:", "example:"
+        l = line.strip(" '\"")
+        if len(l) < 5 or len(l) > 500:
+            continue
+        if any(bad in l.lower() for bad in [
+            "here is", "output:", "rewritten", "version", "humanized",
+            "assistant", "response", "instruction", "mr.", "ms.", "please address"
         ]):
             continue
-        if line.startswith((">", "-", "*", "•", "#", "[", "(")) and len(line) < 50:
-            continue  # likely a header or label
-        if len(line) > 5:  # ignore junk like "Sure!" or "Okay"
-            clean_lines.append(line)
-
-    # Use the longest coherent line (works amazingly well)
-    if clean_lines:
-        cleaned = max(clean_lines, key=len)
-    else:
-        cleaned = raw.strip()
-
-    # === STEP 4: Final cleanup ===
-    cleaned = cleaned.strip(" '\"")
-    cleaned = re.sub(r"\s+", " ", cleaned)  # normalize spaces
-    cleaned = cleaned.strip()
-
-    # If still empty or too short, fallback
-    if not cleaned or len(cleaned) < 3:
-        cleaned = "hello world"  # or return original? up to you
-
-    return cleaned
+        if l.startswith((">", "-", "*", "#", "[", "(")) and len(l) < 60:
+            continue
+        candidates.append(l)
+    
+    # Return longest real-looking line
+    return max(candidates, key=len, default="hello there")
 
 def format_prompt(text: str, style: str) -> str:
     """Format prompt for Phi-3.1"""
@@ -223,15 +206,14 @@ async def transform_with_model(text: str, style: str) -> str:
         
         output = llm(
             prompt,
-            max_tokens=300,
+            max_tokens=80,                    # Short = safe from hallucination
             temperature=temperature,
-            top_p=0.9,
-            top_k=40,                    # Add this
-            repeat_penalty=1.15,
-            frequency_penalty=0.3,       # Increase
-            presence_penalty=0.2,
-            stop=["<|end|>", "<|user|>", "<|assistant|>", "\n\n\n", "Human:", "AI:"],
-            echo=False,
+            top_p=0.95,
+            top_k=60,
+            repeat_penalty=1.18,
+            frequency_penalty=0.2,
+            presence_penalty=0.15,
+            stop=["<|end|>", "<|user|>", "<|assistant|>","\n\n", "Please", "Kindly", "Mr.", "Ms.", "Dr.","colleagues", "protocol", "address me", "title"],echo=False,
         )
         
         processing_time = time.time() - start_time