Update app.py

app.py

@@ -12,19 +12,18 @@ import pdfplumber
 # ==== CONFIG ====
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 HF_TOKEN = os.getenv("HF_TOKEN")
-
-# SOTA models: for general and code queries
+# Your list of SOTA chat models, in order of preference
 CONVERSATIONAL_MODELS = [
     "deepseek-ai/DeepSeek-V2-Chat",
     "Qwen/Qwen2-72B-Instruct",
     "mistralai/Mixtral-8x22B-Instruct-v0.1",
-    "meta-llama/Meta-Llama-3-70B-Instruct"
+    "meta-llama/Meta-Llama-3-70B-Instruct",
+    "deepseek-ai/DeepSeek-Coder-33B-Instruct"
 ]
-CODING_MODEL = "deepseek-ai/DeepSeek-Coder-33B-Instruct"
 
 wiki_api = wikipediaapi.Wikipedia(language="en", user_agent="SmartAgent/1.0 ([email protected])")
 
-# ==== UTILITIES ====
+# ==== UTILITY: Link/file detection ====
 def extract_links(text):
     url_pattern = re.compile(r'(https?://[^\s\)\],]+)')
     return url_pattern.findall(text or "")
@@ -42,26 +41,36 @@ def download_file(url, out_dir="tmp_files"):
     except Exception:
         return None
 
+# ==== File/Link Analyzers ====
 def analyze_file(file_path):
-    try:
-        if file_path.endswith((".xlsx", ".xls")):
+    if file_path.endswith((".xlsx", ".xls")):
+        try:
             df = pd.read_excel(file_path)
             return f"Excel summary: {df.head().to_markdown(index=False)}"
-        elif file_path.endswith(".csv"):
+        except Exception as e:
+            return f"Excel error: {e}"
+    elif file_path.endswith(".csv"):
+        try:
             df = pd.read_csv(file_path)
             return f"CSV summary: {df.head().to_markdown(index=False)}"
-        elif file_path.endswith(".pdf"):
+        except Exception as e:
+            return f"CSV error: {e}"
+    elif file_path.endswith(".pdf"):
+        try:
             with pdfplumber.open(file_path) as pdf:
                 first_page = pdf.pages[0].extract_text()
                 return f"PDF text sample: {first_page[:1000]}"
-        elif file_path.endswith(".txt"):
+        except Exception as e:
+            return f"PDF error: {e}"
+    elif file_path.endswith(".txt"):
+        try:
             with open(file_path, encoding='utf-8') as f:
                 txt = f.read()
             return f"TXT file sample: {txt[:1000]}"
-        else:
-            return f"Unsupported file type: {file_path}"
-    except Exception as e:
-        return f"File analysis error: {e}"
+        except Exception as e:
+            return f"TXT error: {e}"
+    else:
+        return f"Unsupported file type: {file_path}"
 
 def analyze_webpage(url):
     try:
@@ -74,6 +83,7 @@ def analyze_webpage(url):
     except Exception as e:
         return f"Webpage error: {e}"
 
+# ==== SEARCH TOOLS ====
 def duckduckgo_search(query):
     try:
         with DDGS() as ddgs:
@@ -103,31 +113,25 @@ def is_coding_question(text):
         return True
     return False
 
-def llm_coder(query):
-    try:
-        hf_client = InferenceClient(CODING_MODEL, token=HF_TOKEN)
-        result = hf_client.text_generation(query, max_new_tokens=1024)
-        if isinstance(result, dict) and "generated_text" in result:
-            return f"[{CODING_MODEL}] {result['generated_text']}"
-        elif isinstance(result, str):
-            return f"[{CODING_MODEL}] {result}"
-        return "Unknown result format from coder model."
-    except Exception as e:
-        return f"Coder Model Error: {e}"
-
-def llm_conversational(query):
+def llm_conversational(question):
     last_error = None
     for model_id in CONVERSATIONAL_MODELS:
         try:
             hf_client = InferenceClient(model_id, token=HF_TOKEN)
-            result = hf_client.text_generation(query, max_new_tokens=512)
+            result = hf_client.conversational(
+                messages=[{"role": "user", "content": question}],
+                max_new_tokens=512,
+            )
+            # Extract generated_text
             if isinstance(result, dict) and "generated_text" in result:
-                return f"[{model_id}] {result['generated_text']}"
+                return f"[{model_id}] " + result["generated_text"]
+            elif hasattr(result, "generated_text"):
+                return f"[{model_id}] " + result.generated_text
             elif isinstance(result, str):
-                return f"[{model_id}] {result}"
+                return f"[{model_id}] " + result
         except Exception as e:
             last_error = f"{model_id}: {e}"
-    return f"LLM Error (all advanced models): {last_error or 'Unknown error'}"
+    return f"LLM Error (all advanced models): {last_error}"
 
 # ==== SMART AGENT ====
 class SmartAgent:
@@ -152,28 +156,40 @@ class SmartAgent:
             if results:
                 return "\n\n".join(results)
 
-        # 2. Code/coding questions: use coder model
+        # 2. Coding/algorithmic questions: Prefer DeepSeek-Coder-33B
         if is_coding_question(question):
-            result = llm_coder(question)
-            if result:
-                return result
+            coder_client = InferenceClient("deepseek-ai/DeepSeek-Coder-33B-Instruct", token=HF_TOKEN)
+            try:
+                coder_result = coder_client.conversational(
+                    messages=[{"role": "user", "content": question}],
+                    max_new_tokens=512,
+                )
+                if isinstance(coder_result, dict) and "generated_text" in coder_result:
+                    return "[deepseek-ai/DeepSeek-Coder-33B-Instruct] " + coder_result["generated_text"]
+                elif hasattr(coder_result, "generated_text"):
+                    return "[deepseek-ai/DeepSeek-Coder-33B-Instruct] " + coder_result.generated_text
+                elif isinstance(coder_result, str):
+                    return "[deepseek-ai/DeepSeek-Coder-33B-Instruct] " + coder_result
+            except Exception as e:
+                # fallback to other chat models
+                pass
 
-        # 3. DuckDuckGo for fresh web results
+        # 3. DuckDuckGo for current/web knowledge
         result = duckduckgo_search(question)
         if result:
             return result
 
-        # 4. Wikipedia for encyclopedic facts
+        # 4. Wikipedia for encyclopedic queries
         result = wikipedia_search(question)
         if result:
             return result
 
-        # 5. General QA, reasoning, or fallback: conversational SOTA models
+        # 5. Fallback to conversational LLMs
         result = llm_conversational(question)
         if result:
             return result
 
-        return "No answer could be found by available models."
+        return "No answer could be found by available tools."
 
 # ==== SUBMISSION LOGIC ====
 def run_and_submit_all(profile: gr.OAuthProfile | None):