update

README.md

@@ -1,13 +1,52 @@
 ---
-title: Test Text2MotionPrompter
-emoji: 🏆
+title: Text2Motion Prompter
+emoji: 🎬
 colorFrom: purple
 colorTo: blue
 sdk: gradio
-sdk_version: 6.2.0
+sdk_version: 5.9.1
 app_file: app.py
 pinned: false
-short_description: Test Text2MotionPrompter
+short_description: Transform action descriptions into structured motion prompts
+license: mit
+tags:
+  - motion-generation
+  - text-to-motion
+  - animation
+  - 3d-motion
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# 🎬 Text2Motion Prompter
+
+Transform action descriptions into structured motion prompts for 3D animation systems.
+
+## Features
+
+- **Duration Estimation**: Analyzes action complexity to estimate execution time in frames (30fps)
+- **Caption Refinement**: Generates refined, grammatically correct English descriptions
+- **Multi-language Support**: Accepts inputs in multiple languages (English, Chinese, etc.)
+- **ZeroGPU Powered**: Utilizes Hugging Face's ZeroGPU for efficient inference
+
+## Usage
+
+1. Enter an action description in the input field (e.g., "walking forward", "跳跃")
+2. Adjust the max tokens if needed
+3. Click "Generate Motion Prompt"
+4. Get a structured JSON response with duration and refined caption
+
+## Output Format
+
+```json
+{
+    "duration": 90,
+    "short_caption": "A person walks forward at a steady pace."
+}
+```
+
+## Model
+
+This Space uses the [Text2MotionPrompter](https://huggingface.co/Text2MotionPrompter/Text2MotionPrompter) model.
+
+## License
+
+MIT License

app.py

@@ -1,7 +1,303 @@
 import gradio as gr
+import spaces
+import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
 
-def greet(name):
-    return "Hello " + name + "!!"
+# Model configuration
+MODEL_NAME = "Text2MotionPrompter/Text2MotionPrompter"
 
-demo = gr.Interface(fn=greet, inputs="text", outputs="text")
-demo.launch()
+# Prompt template
+TEMPLATE = """
+# Role
+You are an expert in 3D motion analysis, animation timing, and choreography. Your task is to analyze textual action descriptions to estimate execution time and standardize the language for motion generation systems.
+
+# Task
+Analyze the user-provided [Input Action] and generate a structured JSON response containing a duration estimate and a refined caption.
+
+# Instructions
+
+### 1. Duration Estimation (frame_count)
+- Analyze the complexity, speed, and physical constraints of the described action.
+- Estimate the time required to perform the action in a **smooth, natural, and realistic manner**.
+- Calculate the total duration in frames based on a **30 fps** (frames per second) standard.
+- Output strictly as an Integer.
+
+### 2. Caption Refinement (short_caption)
+- Generate a refined, grammatically correct version of the input description in **English**.
+- **Strict Constraints**:
+    - You must **PRESERVE** the original sequence of events (chronological order).
+    - You must **RETAIN** all original spatial modifiers (e.g., "left," "upward," "quickly").
+    - **DO NOT** add new sub-actions or hallucinate details not present in the input.
+    - **DO NOT** delete any specific movements.
+- The goal is to improve clarity and flow while maintaining 100% semantic fidelity to the original request.
+
+### 3. Output Format
+- Return **ONLY** a raw JSON object.
+- Do not use Markdown formatting (i.e., do not use ```json ... ```).
+- Ensure the JSON is valid and parsable.
+
+# JSON Structure
+{{
+    "duration": <Integer, frames at 30fps>,
+    "short_caption": "<String, the refined English description>"
+}}
+
+# Input
+{}
+"""
+
+# Global variables for lazy loading
+tokenizer = None
+model = None
+
+
+def load_model():
+    """Load the model and tokenizer (lazy loading)"""
+    global tokenizer, model
+    if tokenizer is None:
+        tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+    if model is None:
+        model = AutoModelForCausalLM.from_pretrained(
+            MODEL_NAME,
+            torch_dtype=torch.float16,
+            device_map="auto"
+        )
+    return tokenizer, model
+
+
+@spaces.GPU(duration=120)
+def generate_motion_prompt(action_input: str, max_new_tokens: int = 512) -> str:
+    """
+    Generate motion prompt from action description.
+    
+    Args:
+        action_input: The action description to analyze
+        max_new_tokens: Maximum number of tokens to generate
+        
+    Returns:
+        Generated JSON response with duration and refined caption
+    """
+    if not action_input.strip():
+        return "Please enter an action description."
+    
+    # Load model (will use cached version if already loaded)
+    tokenizer, model = load_model()
+    
+    # Prepare the prompt
+    messages = [
+        {"role": "user", "content": TEMPLATE.format(action_input)}
+    ]
+    
+    text = tokenizer.apply_chat_template(
+        messages,
+        tokenize=False,
+        add_generation_prompt=True,
+    )
+    
+    model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
+    
+    # Generate response
+    with torch.no_grad():
+        generated_ids = model.generate(
+            **model_inputs,
+            max_new_tokens=max_new_tokens,
+            do_sample=True,
+            temperature=0.7,
+            top_p=0.9,
+        )
+    
+    # Decode only the new tokens
+    output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist()
+    content = tokenizer.decode(output_ids, skip_special_tokens=True)
+    
+    return content
+
+
+# Custom CSS for a distinctive look
+custom_css = """
+@import url('https://fonts.googleapis.com/css2?family=JetBrains+Mono:wght@400;600&family=Space+Grotesk:wght@400;500;600&display=swap');
+
+.gradio-container {
+    font-family: 'Space Grotesk', sans-serif !important;
+    background: linear-gradient(135deg, #0f0f1a 0%, #1a1a2e 50%, #16213e 100%) !important;
+    min-height: 100vh;
+}
+
+.main-title {
+    background: linear-gradient(90deg, #00d4ff, #7c3aed, #f472b6);
+    -webkit-background-clip: text;
+    -webkit-text-fill-color: transparent;
+    background-clip: text;
+    font-size: 2.5rem !important;
+    font-weight: 600 !important;
+    text-align: center;
+    margin-bottom: 0.5rem !important;
+    font-family: 'Space Grotesk', sans-serif !important;
+}
+
+.subtitle {
+    color: #94a3b8 !important;
+    text-align: center;
+    font-size: 1.1rem !important;
+    margin-bottom: 2rem !important;
+}
+
+.input-container, .output-container {
+    background: rgba(30, 41, 59, 0.7) !important;
+    border: 1px solid rgba(100, 116, 139, 0.3) !important;
+    border-radius: 16px !important;
+    backdrop-filter: blur(10px);
+}
+
+textarea, input[type="text"] {
+    background: rgba(15, 23, 42, 0.8) !important;
+    border: 1px solid rgba(100, 116, 139, 0.4) !important;
+    color: #e2e8f0 !important;
+    font-family: 'JetBrains Mono', monospace !important;
+    border-radius: 12px !important;
+}
+
+textarea:focus, input[type="text"]:focus {
+    border-color: #7c3aed !important;
+    box-shadow: 0 0 0 3px rgba(124, 58, 237, 0.2) !important;
+}
+
+.primary-btn {
+    background: linear-gradient(135deg, #7c3aed 0%, #a855f7 100%) !important;
+    border: none !important;
+    color: white !important;
+    font-weight: 600 !important;
+    padding: 12px 32px !important;
+    border-radius: 12px !important;
+    font-size: 1rem !important;
+    transition: all 0.3s ease !important;
+    box-shadow: 0 4px 15px rgba(124, 58, 237, 0.4) !important;
+}
+
+.primary-btn:hover {
+    transform: translateY(-2px) !important;
+    box-shadow: 0 6px 20px rgba(124, 58, 237, 0.6) !important;
+}
+
+.example-btn {
+    background: rgba(51, 65, 85, 0.6) !important;
+    border: 1px solid rgba(100, 116, 139, 0.4) !important;
+    color: #cbd5e1 !important;
+    border-radius: 8px !important;
+    transition: all 0.2s ease !important;
+}
+
+.example-btn:hover {
+    background: rgba(71, 85, 105, 0.8) !important;
+    border-color: #7c3aed !important;
+}
+
+label {
+    color: #94a3b8 !important;
+    font-weight: 500 !important;
+}
+
+.output-json {
+    font-family: 'JetBrains Mono', monospace !important;
+    background: rgba(15, 23, 42, 0.9) !important;
+    color: #22d3ee !important;
+    padding: 1.5rem !important;
+    border-radius: 12px !important;
+    border: 1px solid rgba(34, 211, 238, 0.2) !important;
+}
+
+.footer {
+    text-align: center;
+    color: #64748b;
+    margin-top: 2rem;
+    padding: 1rem;
+    font-size: 0.9rem;
+}
+
+.slider-container input[type="range"] {
+    accent-color: #7c3aed !important;
+}
+"""
+
+# Build the Gradio interface
+with gr.Blocks(css=custom_css, theme=gr.themes.Base()) as demo:
+    gr.HTML("""
+        <h1 class="main-title">🎬 Text2Motion Prompter</h1>
+        <p class="subtitle">Transform action descriptions into structured motion prompts for 3D animation</p>
+    """)
+    
+    with gr.Row():
+        with gr.Column(scale=1):
+            action_input = gr.Textbox(
+                label="🎯 Action Description",
+                placeholder="Enter an action description (e.g., 'walking forward', '向前走', 'jumping up and down')",
+                lines=4,
+                elem_classes=["input-container"]
+            )
+            
+            max_tokens = gr.Slider(
+                label="🔧 Max Tokens",
+                minimum=128,
+                maximum=2048,
+                value=512,
+                step=64,
+                elem_classes=["slider-container"]
+            )
+            
+            submit_btn = gr.Button(
+                "✨ Generate Motion Prompt",
+                variant="primary",
+                elem_classes=["primary-btn"]
+            )
+            
+            gr.Markdown("### 📝 Examples")
+            examples = gr.Examples(
+                examples=[
+                    ["走路"],
+                    ["A person walks forward slowly, then turns left"],
+                    ["跳跃并挥手"],
+                    ["Running quickly and then stopping suddenly"],
+                    ["坐下来休息一会儿，然后站起来"],
+                    ["Dancing with arms raised above the head"],
+                ],
+                inputs=action_input,
+                elem_id="examples"
+            )
+        
+        with gr.Column(scale=1):
+            output = gr.Textbox(
+                label="📤 Generated Motion Prompt (JSON)",
+                lines=12,
+                elem_classes=["output-container", "output-json"]
+            )
+            
+            gr.Markdown("""
+            ### 📖 Output Format
+            The model generates a JSON response containing:
+            - **duration**: Estimated frames at 30fps
+            - **short_caption**: Refined English description
+            """)
+    
+    gr.HTML("""
+        <div class="footer">
+            <p>Powered by <strong>Text2MotionPrompter</strong> | Using Hugging Face ZeroGPU</p>
+        </div>
+    """)
+    
+    # Set up the action
+    submit_btn.click(
+        fn=generate_motion_prompt,
+        inputs=[action_input, max_tokens],
+        outputs=output
+    )
+    
+    # Allow Enter key to submit
+    action_input.submit(
+        fn=generate_motion_prompt,
+        inputs=[action_input, max_tokens],
+        outputs=output
+    )
+
+# Launch the app
+if __name__ == "__main__":
+    demo.launch()

requirements.txt

@@ -0,0 +1,6 @@
+gradio>=4.0.0
+torch>=2.0.0
+transformers>=4.40.0
+accelerate>=0.27.0
+spaces
+