app.py

import gradio as gr
from huggingface_hub import InferenceClient
import time

client = InferenceClient("lambdaindie/lambdai")

css = """
body {
    background-color: #111;
    color: #e0e0e0;
    font-family: 'JetBrains Mono', monospace;
}
.gr-button {
    background: linear-gradient(to right, #2a2a2a, #1f1f1f);
    color: white;
    border-radius: 10px;
    padding: 8px 16px;
    font-weight: bold;
    font-family: 'JetBrains Mono', monospace;
}
.gr-button:hover {
    background: #333;
}
.gr-textbox textarea {
    background-color: #181818 !important;
    color: #fff !important;
    font-family: 'JetBrains Mono', monospace;
    border-radius: 8px;
}
.gr-chat-message {
    font-family: 'JetBrains Mono', monospace;
}
.markdown-think {
    background-color: #1e1e1e;
    border-left: 4px solid #555;
    padding: 10px;
    margin-bottom: 8px;
    font-style: italic;
    white-space: pre-wrap;
    font-family: 'JetBrains Mono', monospace;
    animation: pulse 1.5s infinite ease-in-out;
}
@keyframes pulse {
    0% { opacity: 0.6; }
    50% { opacity: 1.0; }
    100% { opacity: 0.6; }
}
.gr-chatbot {
    width: 100% !important;
    max-width: 100% !important;
}
"""

def respond(message, history, system_message, max_tokens, temperature, top_p):
    messages = [{"role": "system", "content": system_message}] if system_message else []

    for user, assistant in history:
        if user:
            messages.append({"role": "user", "content": user})
        if assistant:
            messages.append({"role": "assistant", "content": assistant})

    thinking_prompt = messages + [
        {
            "role": "user",
            "content": f"{message}\n\nThink step-by-step before answering."
        }
    ]

    reasoning = ""
    yield '<div class="markdown-think">Thinking...</div>'

    for chunk in client.chat_completion(
        thinking_prompt,
        max_tokens=max_tokens,
        stream=True,
        temperature=temperature,
        top_p=top_p,
    ):
        token = chunk.choices[0].delta.content or ""
        reasoning += token
        styled_thought = f'<div class="markdown-think">{reasoning.strip()}</div>'
        yield styled_thought

    time.sleep(0.5)

    final_prompt = messages + [
        {"role": "user", "content": message},
        {"role": "assistant", "content": reasoning.strip()},
        {"role": "user", "content": "Now answer based on your reasoning above."}
    ]

    final_answer = ""
    for chunk in client.chat_completion(
        final_prompt,
        max_tokens=max_tokens,
        stream=True,
        temperature=temperature,
        top_p=top_p,
    ):
        token = chunk.choices[0].delta.content or ""
        final_answer += token
        yield final_answer.strip()

demo = gr.ChatInterface(
    fn=respond,
    title=λambdAI,
    theme=gr.themes.Base(primary_hue="gray"),
    css=css,
    additional_inputs=[
        gr.Textbox(
            value="You are a concise, logical AI that explains its reasoning clearly before answering.",
            label="System Message"
        ),
        gr.Slider(64, 2048, value=512, step=1, label="Max Tokens"),
        gr.Slider(0.1, 2.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(0.1, 1.0, value=0.95, step=0.05, label="Top-p")
    ]
)

if __name__ == "__main__":
    demo.launch()