Update app.py

app.py

@@ -3,10 +3,6 @@ import sqlite3
 from datetime import datetime
 import gradio as gr
 from huggingface_hub import InferenceClient
-from datasets import load_dataset
-from sentence_transformers import SentenceTransformer
-from sklearn.metrics.pairwise import cosine_similarity
-import numpy as np
 
 # ---------------------------
 # Config
@@ -14,8 +10,9 @@ import numpy as np
 MODELS = {
     "Meta LLaMA 3.1 (8B Instruct)": "meta-llama/Llama-3.1-8B-Instruct",
     "Mistral 7B Instruct": "mistralai/Mistral-7B-Instruct-v0.3",
-    # Add your backend-focused fine-tuned model here if available
-    # "Backend-Finetuned Model": "your-username/backend-crud-model"
+    "The Stack (Multilingual)": "bigcode/the-stack",
+    "XLCoST (Multilingual)": "reddy-lab-code-research/XLCoST",
+    "CoMA (Multilingual)": "Denilah/CoMA"
 }
 
 HF_TOKEN = os.getenv("HF_TOKEN")  # Set in your Space's Secrets
@@ -26,9 +23,7 @@ SYSTEM_DEFAULT = (
     "Always prioritize database, API, authentication, routing, migrations, and CRUD logic. "
     "Provide full backend code scaffolds with files, paths, and commands. "
     "Only include frontend if required for framework integration "
-    "(e.g., Laravel Blade, Django templates). Ignore other frontend/UI tasks. "
-    "If user asks for excessive frontend work, politely respond: "
-    "'I am a backend assistant and focus only on backend tasks.'"
+    "(e.g., Laravel Blade, Django templates). Ignore other frontend/UI tasks."
 )
 
 # ---------------------------
@@ -126,23 +121,6 @@ def update_session_title_if_needed(session_id: int, first_user_text: str):
         conn.commit()
     conn.close()
 
-# ---------------------------
-# Dataset & Embeddings Setup
-# ---------------------------
-print("Loading dataset and embeddings... (this runs only once)")
-dataset = load_dataset("codeparrot/codeparrot-clean-python", split="train[:5%]")  # small % for speed
-backend_snippets = [d["content"] for d in dataset if any(k in d["content"].lower() for k in
-                                                           ["db", "database", "api", "crud", "auth", "routing", "migration"])]
-
-embed_model = SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2")
-snippet_embeddings = embed_model.encode(backend_snippets, convert_to_numpy=True)
-
-def get_relevant_snippets(user_text, top_k=3):
-    user_emb = embed_model.encode([user_text], convert_to_numpy=True)
-    sims = cosine_similarity(user_emb, snippet_embeddings)[0]
-    top_indices = np.argsort(sims)[-top_k:][::-1]
-    return "\n\n".join([backend_snippets[i] for i in top_indices])
-
 # ---------------------------
 # Helpers
 # ---------------------------
@@ -154,11 +132,9 @@ def label_to_id(label: str | None) -> int | None:
     except Exception:
         return None
 
-def build_api_messages(session_id: int, system_message: str, user_text: str):
-    relevant_snippets = get_relevant_snippets(user_text)
+def build_api_messages(session_id: int, system_message: str):
     msgs = [{"role": "system", "content": system_message.strip()}]
     msgs.extend(get_messages(session_id))
-    msgs.append({"role": "user", "content": relevant_snippets + "\n\n" + user_text})
     return msgs
 
 def get_client(model_choice: str):
@@ -203,7 +179,7 @@ def send_cb(user_text, selected_label, chatbot_msgs, system_message, max_tokens,
     add_message(sid, "user", user_text)
     update_session_title_if_needed(sid, user_text)
 
-    api_messages = build_api_messages(sid, system_message, user_text)
+    api_messages = build_api_messages(sid, system_message)
     display_msgs = get_messages(sid)
     display_msgs.append({"role": "assistant", "content": ""})
 
@@ -325,46 +301,6 @@ with gr.Blocks(title="Backend-Focused LLaMA/Mistral CRUD Assistant", theme=gr.th
                 interactive=True
             )
 
-            gr.Markdown("### ⚙️ Generation Settings")
-            system_box = gr.Textbox(
-                value=SYSTEM_DEFAULT,
-                label="System message",
-                lines=4
-            )
-            max_tokens = gr.Slider(256, 4096, value=1200, step=16, label="Max tokens")
-            temperature = gr.Slider(0.0, 2.0, value=0.25, step=0.05, label="Temperature")
-            top_p = gr.Slider(0.1, 1.0, value=0.9, step=0.05, label="Top-p")
-
-        with gr.Column(scale=3):
-            chatbot = gr.Chatbot(label="Assistant", height=520, type="messages")
-            with gr.Row():
-                user_box = gr.Textbox(placeholder="Describe your CRUD/backend task…", lines=3, scale=5)
-            with gr.Row():
-                send_btn = gr.Button("Send ▶️", variant="primary")
-                regen_btn = gr.Button("Regenerate 🔁", variant="secondary")
-
-    refresh_btn.click(refresh_sessions_cb, outputs=session_list)
-    new_btn.click(new_chat_cb, outputs=[session_list, chatbot, user_box])
-    del_btn.click(delete_chat_cb, inputs=session_list, outputs=[session_list, chatbot])
-    session_list.change(load_session_cb, inputs=session_list, outputs=chatbot)
-
-    send_btn.click(
-        send_cb,
-        inputs=[user_box, session_list, chatbot, system_box, max_tokens, temperature, top_p, model_choice],
-        outputs=[chatbot, user_box, session_list]
-    )
-
-    user_box.submit(
-        send_cb,
-        inputs=[user_box, session_list, chatbot, system_box, max_tokens, temperature, top_p, model_choice],
-        outputs=[chatbot, user_box, session_list]
-    )
-
-    regen_btn.click(
-        regenerate_cb,
-        inputs=[session_list, system_box, max_tokens, temperature, top_p, model_choice],
-        outputs=chatbot
-    )
-
-if __name__ == "__main__":
-    demo.launch()
+            gr.Markdown("### ⚙
+::contentReference[oaicite:0]{index=0}
+