Upload 2 files

ai_studio_code (3).py

@@ -0,0 +1,215 @@
+import time
+import gradio as gr
+from openai import OpenAI
+from duckduckgo_search import DDGS
+import json
+
+# --- Logic Functions ---
+
+client = OpenAI(base_url="http://localhost:8080/v1", api_key="no-key-required")
+
+def search_web(query):
+    try:
+        with DDGS() as ddgs:
+            results = [r for r in ddgs.text(query, max_results=5)]
+            if not results:
+                return None
+            context = "\n".join([f"Source: {r['title']}\nContent: {r['body']}" for r in results])
+            return context
+    except Exception as e:
+        print(f"Search error: {e}")
+        return None
+
+def format_time(seconds_float):
+    total_seconds = int(round(seconds_float))
+    m, s = divmod(total_seconds, 60)
+    h, m = divmod(m, 60)
+    return f"{h}h {m}m {s}s" if h > 0 else f"{m}m {s}s" if m > 0 else f"{s}s"
+
+class ParserState:
+    __slots__ = ['answer', 'thought', 'in_think', 'start_time', 'last_pos', 'total_think_time']
+    def __init__(self):
+        self.answer = ""
+        self.thought = ""
+        self.in_think = False
+        self.start_time = 0
+        self.last_pos = 0
+        self.total_think_time = 0.0
+
+def parse_response(text, state):
+    buffer = text[state.last_pos:]
+    state.last_pos = len(text)
+    while buffer:
+        if not state.in_think:
+            think_start = buffer.find('<think>')
+            if think_start != -1:
+                state.answer += buffer[:think_start]
+                state.in_think = True
+                state.start_time = time.perf_counter()
+                buffer = buffer[think_start + 7:]
+            else:
+                state.answer += buffer
+                break
+        else:
+            think_end = buffer.find('</think>')
+            if think_end != -1:
+                state.thought += buffer[:think_end]
+                state.total_think_time += (time.perf_counter() - state.start_time)
+                state.in_think = False
+                buffer = buffer[think_end + 8:]
+            else:
+                state.thought += buffer
+                break
+    return state
+
+def format_ui_response(state):
+    collapsible = ""
+    if state.thought or state.in_think:
+        status = f"🌀 Thinking ({format_time(state.total_think_time)})" if state.in_think else f"✅ Thought for {format_time(state.total_think_time)}"
+        open_tag = "open" if state.in_think else ""
+        collapsible = f"<details {open_tag}><summary>{status}</summary><div class='thinking-container'>{state.thought}</div></details>"
+    return f"{collapsible}\n\n{state.answer}"
+
+# --- Gradio UI Handlers ---
+
+def user_msg(user_input, history):
+    return "", history + [[user_input, None]]
+
+def generate_response(history, search_enabled, temp, top_p, max_tok, active_gen):
+    if not history: return history
+    
+    query = history[-1][0]
+    full_prompt = query
+
+    # Perform Search if enabled
+    if search_enabled:
+        history[-1][1] = "🔍 Searching the web..."
+        yield history
+        search_results = search_web(query)
+        if search_results:
+            full_prompt = f"User Query: {query}\n\nWeb Search Results:\n{search_results}\n\nInstruction: Use the search results to provide a comprehensive answer."
+        else:
+            history[-1][1] = "🔍 No search results found. Proceeding with internal knowledge..."
+            yield history
+
+    messages = [{"role": "user", "content": full_prompt}]
+    state = ParserState()
+    full_text = ""
+    
+    try:
+        stream = client.chat.completions.create(
+            model="local",
+            messages=messages,
+            temperature=temp,
+            top_p=top_p,
+            max_tokens=max_tok,
+            stream=True
+        )
+        
+        for chunk in stream:
+            if not active_gen[0]: break
+            if chunk.choices[0].delta.content:
+                full_text += chunk.choices[0].delta.content
+                state = parse_response(full_text, state)
+                history[-1][1] = format_ui_response(state)
+                yield history
+    except Exception as e:
+        history[-1][1] = f"Error: {str(e)}"
+        yield history
+    finally:
+        active_gen[0] = False
+
+# --- UI Setup ---
+
+CSS = """
+.thinking-container {
+    border-left: 3px solid #facc15;
+    padding: 10px;
+    margin: 5px 0;
+    background: rgba(250, 204, 21, 0.05);
+    font-style: italic;
+    color: #666;
+}
+details {
+    border: 1px solid #ddd;
+    border-radius: 8px;
+    padding: 5px 10px;
+    margin-bottom: 10px;
+}
+summary {
+    cursor: pointer;
+    font-weight: bold;
+    color: #444;
+}
+"""
+
+with gr.Blocks(css=CSS, theme=gr.themes.Soft()) as demo:
+    active_gen = gr.State([False])
+    
+    with gr.Row():
+        with gr.Column(scale=8):
+            gr.Markdown("# 🧠 Qwen3 Web Explorer\n*High-speed reasoning with real-time web access*")
+        with gr.Column(scale=2):
+            search_toggle = gr.Checkbox(label="🌐 Enable Web Search", value=False)
+
+    chatbot = gr.Chatbot(height=550, show_label=False, bubble_full_width=False)
+    
+    with gr.Row():
+        msg = gr.Textbox(
+            placeholder="Ask me anything...",
+            container=False,
+            scale=7
+        )
+        submit_btn = gr.Button("Send", variant="primary", scale=1)
+
+    with gr.Row():
+        stop_btn = gr.Button("⏹ Stop", variant="secondary")
+        regen_btn = gr.Button("🔄 Regenerate", variant="secondary")
+        undo_btn = gr.Button("↩️ Undo", variant="secondary")
+        clear_btn = gr.Button("🗑 Clear Chat", variant="secondary")
+
+    with gr.Accordion("Advanced Settings", open=False):
+        with gr.Row():
+            temp = gr.Slider(0.1, 1.5, 0.6, label="Temperature")
+            top_p = gr.Slider(0.1, 1.0, 0.95, label="Top-p")
+            max_tok = gr.Slider(1024, 32768, 4096, step=128, label="Max Tokens")
+
+    # Functions
+    def start_gen(): return [True]
+    def stop_gen(): return [False]
+    def undo(history):
+        if len(history) > 0: history.pop()
+        return history
+
+    submit_event = submit_btn.click(
+        user_msg, [msg, chatbot], [msg, chatbot], queue=False
+    ).then(
+        start_gen, None, active_gen
+    ).then(
+        generate_response, [chatbot, search_toggle, temp, top_p, max_tok, active_gen], chatbot
+    )
+
+    msg.submit(
+        user_msg, [msg, chatbot], [msg, chatbot], queue=False
+    ).then(
+        start_gen, None, active_gen
+    ).then(
+        generate_response, [chatbot, search_toggle, temp, top_p, max_tok, active_gen], chatbot
+    )
+
+    regen_btn.click(
+        lambda history: (history[-1][0], history[:-1]), [chatbot], [msg, chatbot], queue=False
+    ).then(
+        user_msg, [msg, chatbot], [msg, chatbot], queue=False
+    ).then(
+        start_gen, None, active_gen
+    ).then(
+        generate_response, [chatbot, search_toggle, temp, top_p, max_tok, active_gen], chatbot
+    )
+
+    stop_btn.click(stop_gen, None, active_gen, cancels=[submit_event])
+    undo_btn.click(undo, [chatbot], [chatbot])
+    clear_btn.click(lambda: None, None, chatbot)
+
+if __name__ == "__main__":
+    demo.launch(server_name="0.0.0.0", server_port=7860)

ai_studio_code.txt

@@ -0,0 +1,38 @@
+FROM ubuntu:22.04
+
+# Install system dependencies
+RUN apt-get update && \
+    apt-get install -y \
+    build-essential \
+    libssl-dev \
+    zlib1g-dev \
+    libboost-all-dev \
+    libopenblas-dev \
+    libomp-dev \
+    cmake \
+    pkg-config \
+    git \
+    python3-pip \
+    curl \
+    wget && \
+    rm -rf /var/lib/apt/lists/*
+    
+# Install Python dependencies (Added duckduckgo-search)
+RUN pip3 install huggingface-hub openai gradio duckduckgo-search
+
+# Build llama.cpp
+RUN git clone https://github.com/ggerganov/llama.cpp && \
+    cd llama.cpp && \
+    cmake -B build -S . -DLLAMA_BUILD_SERVER=ON -DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS -DCMAKE_BUILD_TYPE=Release && \
+    cmake --build build --config Release --target llama-server -j $(nproc)
+
+# Download model
+RUN mkdir -p /models && \
+    wget -O /models/model.q8_k_xl.gguf https://huggingface.co/unsloth/Qwen3-0.6B-GGUF/resolve/main/Qwen3-0.6B-UD-Q8_K_XL.gguf
+
+COPY app.py /app.py
+COPY start.sh /start.sh
+RUN chmod +x /start.sh
+
+EXPOSE 7860 8080
+CMD ["/start.sh"]