Each rank should sample a different timestep in dmd_forward and faker_score_forward

.pre-commit-config.yaml

@@ -10,7 +10,6 @@ exclude: |
     demo/.*|
     predict\.py|
     scripts/.*|
-    prompts/.*|
     fastvideo/data_preprocess/.*|
     fastvideo/dataset/.*|
     fastvideo/models/.*|

examples/inference/basic/basic_self_forcing_causal_wan2_2_i2v.py

@@ -11,7 +11,7 @@ def main():
     # If a local path is provided, FastVideo will make a best effort
     # attempt to identify the optimal arguments.
     generator = VideoGenerator.from_pretrained(
-        "FastVideo/SFWan2.2-I2V-A14B-Preview-Diffusers",
+        "rand0nmr/SFWan2.2-T2V-A14B-Diffusers",
         # FastVideo will automatically handle distributed setup
         num_gpus=1,
         use_fsdp_inference=True,
@@ -22,10 +22,12 @@ def main():
         dmd_denoising_steps=[1000, 850, 700, 550, 350, 275, 200, 125],
         # Set pin_cpu_memory to false if CPU RAM is limited and there're no frequent CPU-GPU transfer
         pin_cpu_memory=True,
+        init_weights_from_safetensors="/mnt/sharefs/users/hao.zhang/wei/SFwan2.2_distill_self_forcing_release_x0_CacheAppend_1333/checkpoint-442_weight_only/ema/generator_ema.safetensors",
+        init_weights_from_safetensors_2="/mnt/sharefs/users/hao.zhang/wei/SFwan2.2_distill_self_forcing_release_x0_CacheAppend_1333/checkpoint-442_weight_only/ema/generator_ema_2.safetensors",
         # image_encoder_cpu_offload=False,
     )
 
-    sampling_param = SamplingParam.from_pretrained("FastVideo/SFWan2.2-I2V-A14B-Preview-Diffusers")
+    sampling_param = SamplingParam.from_pretrained("rand0nmr/SFWan2.2-T2V-A14B-Diffusers")
     sampling_param.num_frames = 73
     sampling_param.width = 832
     sampling_param.height = 480
@@ -41,4 +43,4 @@ def main():
 
 
 if __name__ == "__main__":
-    main()
+    main()

examples/inference/gradio/serving/gradio_frontend.py

@@ -3,13 +3,8 @@ import os
 import requests
 import base64
 import time
-import json
-from pathlib import Path
-import tempfile
-from io import BytesIO
 
 import gradio as gr
-from PIL import Image
 
 from fastvideo.configs.sample.base import SamplingParam
 
@@ -17,7 +12,6 @@ from fastvideo.configs.sample.base import SamplingParam
 MODEL_PATH_MAPPING = {
     "FastWan2.1-T2V-1.3B": "FastVideo/FastWan2.1-T2V-1.3B-Diffusers",
     "FastWan2.2-TI2V-5B-FullAttn": "FastVideo/FastWan2.2-TI2V-5B-FullAttn-Diffusers",
-    "CausalWan2.2-I2V-A14B-Preview": "FastVideo/SFWan2.2-I2V-A14B-Preview-Diffusers",
 }
 
 
@@ -43,7 +37,7 @@ class RayServeClient:
                 f"{self.backend_url}/generate_video",
                 json=request_data,
                 headers=headers,
-                timeout=900  # 15 minutes timeout for longer video generation
+                timeout=300
             )
             
             round_trip_time = time.time() - start_time
@@ -87,89 +81,49 @@ def save_video_from_base64(video_data: str, output_dir: str, prompt: str) -> str
         return None
 
 
-def encode_image_to_base64(image_input) -> str:
-    """Encode an image file path or in-memory image to a base64 string."""
-    if image_input is None:
-        return None
+def create_timing_display(inference_time, encoding_time, network_time, total_time, stage_execution_times, num_frames):
+    dit_denoising_time = f"{stage_execution_times[5]:.2f}s" if len(stage_execution_times) > 5 else "N/A"
     
-    mime_types = {
-        '.jpg': 'image/jpeg',
-        '.jpeg': 'image/jpeg',
-        '.png': 'image/png',
-        '.gif': 'image/gif',
-        '.webp': 'image/webp',
-    }
+    timing_html = f"""
+    <div style="margin: 10px 0;">
+        <h3 style="text-align: center; margin-bottom: 10px;">⏱️ Timing Breakdown</h3>
+        <div style="display: grid; grid-template-columns: repeat(5, 1fr); gap: 10px; margin-bottom: 10px;">
+            <div class="timing-card timing-card-highlight">
+                <div style="font-size: 20px;">🚀</div>
+                <div style="font-weight: bold; margin: 3px 0; font-size: 14px;">DiT Denoising</div>
+                <div style="font-size: 16px; color: #ffa200; font-weight: bold;">{dit_denoising_time}</div>
+            </div>
+            <div class="timing-card">
+                <div style="font-size: 20px;">🧠</div>
+                <div style="font-weight: bold; margin: 3px 0; font-size: 14px;">E2E (w. vae/text encoder)</div>
+                <div style="font-size: 16px; color: #2563eb;">{inference_time:.2f}s</div>
+            </div>
+            <div class="timing-card">
+                <div style="font-size: 20px;">🎬</div>
+                <div style="font-weight: bold; margin: 3px 0; font-size: 14px;">Video Encoding</div>
+                <div style="font-size: 16px; color: #dc2626;">{encoding_time:.2f}s</div>
+            </div>
+            <div class="timing-card">
+                <div style="font-size: 20px;">🌐</div>
+                <div style="font-weight: bold; margin: 3px 0; font-size: 14px;">Network Transfer</div>
+                <div style="font-size: 16px; color: #059669;">{network_time:.2f}s</div>
+            </div>
+            <div class="timing-card">
+                <div style="font-size: 20px;">📊</div>
+                <div style="font-weight: bold; margin: 3px 0; font-size: 14px;">Total Processing</div>
+                <div style="font-size: 18px; color: #0277bd;">{total_time:.2f}s</div>
+            </div>
+        </div>"""
     
-    try:
-        if isinstance(image_input, str):
-            if not os.path.exists(image_input):
-                return None
-            
-            with open(image_input, 'rb') as f:
-                image_bytes = f.read()
-            
-            ext = os.path.splitext(image_input)[1].lower()
-            mime_type = mime_types.get(ext, 'image/jpeg')
-        elif isinstance(image_input, Image.Image):
-            buffer = BytesIO()
-            image_to_save = image_input.convert("RGB")
-            image_to_save.save(buffer, format="PNG")
-            image_bytes = buffer.getvalue()
-            mime_type = 'image/png'
-        else:
-            return None
-        
-        image_base64 = base64.b64encode(image_bytes).decode('utf-8')
-        return f"data:{mime_type};base64,{image_base64}"
-        
-    except Exception as e:
-        print(f"Failed to encode image: {e}")
-        return None
-
-
-# def create_timing_display(inference_time, encoding_time, network_time, total_time, stage_execution_times, num_frames):
-#     dit_denoising_time = f"{stage_execution_times[5]:.2f}s" if len(stage_execution_times) > 5 else "N/A"
-#     
-#     timing_html = f"""
-#     <div style="margin: 10px 0;">
-#         <h3 style="text-align: center; margin-bottom: 10px;">⏱️ Timing Breakdown</h3>
-#         <div style="display: grid; grid-template-columns: repeat(5, 1fr); gap: 10px; margin-bottom: 10px;">
-#             <div class="timing-card timing-card-highlight">
-#                 <div style="font-size: 20px;">🚀</div>
-#                 <div style="font-weight: bold; margin: 3px 0; font-size: 14px;">DiT Denoising</div>
-#                 <div style="font-size: 16px; color: #ffa200; font-weight: bold;">{dit_denoising_time}</div>
-#             </div>
-#             <div class="timing-card">
-#                 <div style="font-size: 20px;">🧠</div>
-#                 <div style="font-weight: bold; margin: 3px 0; font-size: 14px;">E2E (w. vae/text encoder)</div>
-#                 <div style="font-size: 16px; color: #2563eb;">{inference_time:.2f}s</div>
-#             </div>
-#             <div class="timing-card">
-#                 <div style="font-size: 20px;">🎬</div>
-#                 <div style="font-weight: bold; margin: 3px 0; font-size: 14px;">Video Encoding</div>
-#                 <div style="font-size: 16px; color: #dc2626;">{encoding_time:.2f}s</div>
-#             </div>
-#             <div class="timing-card">
-#                 <div style="font-size: 20px;">🌐</div>
-#                 <div style="font-weight: bold; margin: 3px 0; font-size: 14px;">Network Transfer</div>
-#                 <div style="font-size: 16px; color: #059669;">{network_time:.2f}s</div>
-#             </div>
-#             <div class="timing-card">
-#                 <div style="font-size: 20px;">📊</div>
-#                 <div style="font-weight: bold; margin: 3px 0; font-size: 14px;">Total Processing</div>
-#                 <div style="font-size: 18px; color: #0277bd;">{total_time:.2f}s</div>
-#             </div>
-#         </div>"""
-#     
-#     if inference_time > 0:
-#         fps = num_frames / inference_time
-#         timing_html += f"""
-#         <div class="performance-card" style="margin-top: 15px;">
-#             <span style="font-weight: bold;">Generation Speed: </span>
-#             <span style="font-size: 18px; color: #6366f1; font-weight: bold;">{fps:.1f} frames/second</span>
-#         </div>"""
-#     
-#     return timing_html + "</div>"
+    if inference_time > 0:
+        fps = num_frames / inference_time
+        timing_html += f"""
+        <div class="performance-card" style="margin-top: 15px;">
+            <span style="font-weight: bold;">Generation Speed: </span>
+            <span style="font-size: 18px; color: #6366f1; font-weight: bold;">{fps:.1f} frames/second</span>
+        </div>"""
+    
+    return timing_html + "</div>"
 
 
 def load_example_prompts():
@@ -190,83 +144,26 @@ def load_example_prompts():
             print(f"Warning: Could not read {filepath}: {e}")
         return prompts, labels
     
-    # Load prompts from prompts.txt
-    examples, example_labels = load_from_file("examples/inference/gradio/serving/prompts.txt")
+    examples, example_labels = load_from_file("prompts/prompts_final.txt")
     
     if not examples:
         examples = ["A crowded rooftop bar buzzes with energy, the city skyline twinkling like a field of stars in the background."]
         example_labels = ["Crowded rooftop bar at night"]
     
-    # Load image mappings from JSON file
-    prompt_to_image = {}
-    # Try to find the JSON file relative to project root
-    possible_json_paths = [
-        Path("prompts/mixkit_i2v.jsonl"),
-        Path(__file__).parent.parent.parent.parent / "prompts" / "mixkit_i2v.jsonl",
-    ]
-    json_path = None
-    for path in possible_json_paths:
-        if path.exists():
-            json_path = path
-            break
-    
-    if json_path and json_path.exists():
-        try:
-            with open(json_path, "r", encoding='utf-8') as f:
-                data = json.load(f)
-                # Get the project root directory (parent of prompts directory)
-                project_root = json_path.parent.parent
-                for item in data:
-                    prompt_text = item.get("prompt", "").strip()
-                    image_path = item.get("image_path", "")
-                    if prompt_text and image_path:
-                        # Resolve image path relative to project root
-                        full_image_path = project_root / image_path
-                        if full_image_path.exists():
-                            prompt_to_image[prompt_text] = str(full_image_path.absolute())
-        except Exception as e:
-            print(f"Warning: Could not load image mappings from {json_path}: {e}")
-    
-    # Create image paths list matching the prompts
-    example_images = []
-    for prompt in examples:
-        # Try exact match first
-        image_path = prompt_to_image.get(prompt)
-        if not image_path:
-            # Try fuzzy match (case-insensitive, whitespace normalized)
-            normalized_prompt = " ".join(prompt.split())
-            for json_prompt, img_path in prompt_to_image.items():
-                normalized_json = " ".join(json_prompt.split())
-                if normalized_prompt.lower() == normalized_json.lower():
-                    image_path = img_path
-                    break
-        example_images.append(image_path if image_path and os.path.exists(image_path) else None)
-    
-    return examples, example_labels, example_images
+    return examples, example_labels
 
 
 def create_gradio_interface(backend_url: str, default_params: dict[str, SamplingParam]):
     
     client = RayServeClient(backend_url)
     
-    def is_i2v_model(model_name: str) -> bool:
-        """Check if the model is an I2V model."""
-        return "I2V" in model_name
-    
     def generate_video(
-        prompt, negative_prompt, use_negative_prompt, guidance_scale,
-        num_frames, height, width, model_selection, input_image, progress
+        prompt, negative_prompt, use_negative_prompt, seed, guidance_scale,
+        num_frames, height, width, randomize_seed, model_selection, progress
     ):
-        # Use default seed value (randomize_seed disabled)
-        seed = 1000
-        randomize_seed = False
         if not client.check_health():
             return None, f"Backend is not available. Please check if Ray Serve is running at {backend_url}", ""
         
-        # Check if I2V model requires an image
-        if is_i2v_model(model_selection) and not input_image:
-            return None, "I2V models require an input image. Please upload an image.", ""
-        
         # Validate dimensions
         max_pixels = 720 * 1280
         if height * width > max_pixels:
@@ -275,15 +172,6 @@ def create_gradio_interface(backend_url: str, default_params: dict[str, Sampling
         if progress:
             progress(0.1, desc="Checking backend health...")
         
-        # Encode image if provided
-        image_data = None
-        if input_image:
-            if progress:
-                progress(0.2, desc="Encoding input image...")
-            image_data = encode_image_to_base64(input_image)
-            if not image_data:
-                return None, "Failed to encode input image", ""
-        
         request_data = {
             "prompt": prompt,
             "negative_prompt": negative_prompt,
@@ -295,7 +183,7 @@ def create_gradio_interface(backend_url: str, default_params: dict[str, Sampling
             "width": width,
             "randomize_seed": randomize_seed,
             "return_frames": False,
-            "image_data": image_data,
+            "image_path": None,
             "model_path": MODEL_PATH_MAPPING.get(model_selection, "FastVideo/FastWan2.1-T2V-1.3B-Diffusers")
         }
         
@@ -310,16 +198,16 @@ def create_gradio_interface(backend_url: str, default_params: dict[str, Sampling
         if response.get("success", False):
             video_data = response.get("video_data", "")
             used_seed = response.get("seed", seed)
-            # inference_time = response.get("inference_time", 0.0)
-            # encoding_time = response.get("encoding_time", 0.0)
-            # total_time = response.get("total_time", 0.0)
-            # network_time = response.get("network_time", 0.0)
-            # stage_execution_times = response.get("stage_execution_times", [])
+            inference_time = response.get("inference_time", 0.0)
+            encoding_time = response.get("encoding_time", 0.0)
+            total_time = response.get("total_time", 0.0)
+            network_time = response.get("network_time", 0.0)
+            stage_execution_times = response.get("stage_execution_times", [])
             
-            # timing_details = create_timing_display(
-            #     inference_time, encoding_time, network_time, total_time, 
-            #     stage_execution_times, num_frames
-            # )
+            timing_details = create_timing_display(
+                inference_time, encoding_time, network_time, total_time, 
+                stage_execution_times, num_frames
+            )
             
             if video_data:
                 if progress:
@@ -331,7 +219,7 @@ def create_gradio_interface(backend_url: str, default_params: dict[str, Sampling
                     progress(1.0, desc="Generation complete!")
                 
                 if video_path and os.path.exists(video_path):
-                    return video_path, used_seed, ""
+                    return video_path, used_seed, timing_details
                 else:
                     return None, "Failed to save video", ""
             else:
@@ -340,7 +228,7 @@ def create_gradio_interface(backend_url: str, default_params: dict[str, Sampling
             error_msg = response.get("error_message", "Unknown error occurred")
             return None, f"Generation failed: {error_msg}", ""
     
-    examples, example_labels, example_images = load_example_prompts()
+    examples, example_labels = load_example_prompts()
     
     theme = gr.themes.Base().set(
         button_primary_background_fill="#2563eb",
@@ -351,39 +239,33 @@ def create_gradio_interface(backend_url: str, default_params: dict[str, Sampling
     )
     
     def get_default_values(model_name):
-        # model_path = MODEL_PATH_MAPPING.get(model_name)
-        # if model_path and model_path in default_params:
-        #     params = default_params[model_path]
-        #     return {
-        #         'height': params.height,
-        #         'width': params.width,
-        #         'num_frames': params.num_frames,
-        #         'guidance_scale': params.guidance_scale,
-        #     }
+        model_path = MODEL_PATH_MAPPING.get(model_name)
+        if model_path and model_path in default_params:
+            params = default_params[model_path]
+            return {
+                'height': params.height,
+                'width': params.width,
+                'num_frames': params.num_frames,
+                'guidance_scale': params.guidance_scale,
+                'seed': params.seed,
+            }
         
         return {
-            'height': 480,
+            'height': 448,
             'width': 832,
-            'num_frames': 73,
+            'num_frames': 61,
+            'guidance_scale': 3.0,
+            'seed': 1024,
         }
     
-    # Get available models based on what's loaded
-    available_models = []
-    for model_name, model_path in MODEL_PATH_MAPPING.items():
-        if model_path in default_params:
-            available_models.append(model_name)
+    initial_values = get_default_values("FastWan2.1-T2V-1.3B")
     
-    # Select first available model as default
-    default_model = available_models[0] if available_models else "FastWan2.1-T2V-1.3B"
-    initial_values = get_default_values(default_model)
-    initial_show_image = is_i2v_model(default_model)
-    
-    with gr.Blocks(title="CausalWan", theme=theme) as demo:
+    with gr.Blocks(title="FastWan", theme=theme) as demo:
         gr.Image("assets/logos/logo.svg", show_label=False, container=False, height=80)
         gr.HTML("""
         <div style="text-align: center; margin-bottom: 10px;">
             <p style="font-size: 18px;"> Make Video Generation Go Blurrrrrrr </p>
-            <p style="font-size: 18px;"> <a href="https://github.com/hao-ai-lab/FastVideo/tree/main" target="_blank">Code</a> | <a href="https://hao-ai-lab.github.io/blogs/fastvideo_causalwan_preview/" target="_blank">Blog</a> | <a href="https://hao-ai-lab.github.io/FastVideo/" target="_blank">Docs</a>  </p>
+            <p style="font-size: 18px;"> <a href="https://github.com/hao-ai-lab/FastVideo/tree/main" target="_blank">Code</a> | <a href="https://hao-ai-lab.github.io/blogs/fastvideo_post_training/" target="_blank">Blog</a> | <a href="https://hao-ai-lab.github.io/FastVideo/" target="_blank">Docs</a>  </p>
         </div>
         """)
         
@@ -398,8 +280,8 @@ def create_gradio_interface(backend_url: str, default_params: dict[str, Sampling
         
         with gr.Row():
             model_selection = gr.Dropdown(
-                choices=available_models,
-                value=default_model,
+                choices=list(MODEL_PATH_MAPPING.keys()),
+                value="FastWan2.1-T2V-1.3B",
                 label="Select Model",
                 interactive=True
             )
@@ -430,70 +312,69 @@ def create_gradio_interface(backend_url: str, default_params: dict[str, Sampling
         with gr.Row():
             with gr.Column():
                 error_output = gr.Text(label="Error", visible=False)
-                # timing_display = gr.Markdown(label="Timing Breakdown", visible=False)
+                timing_display = gr.Markdown(label="Timing Breakdown", visible=False)
 
-        with gr.Row(equal_height=False):
-            with gr.Column(scale=1):
-                with gr.Tabs():
-                    with gr.Tab("Input Image", visible=initial_show_image) as image_tab:
-                        gr.Markdown("**Please make sure you upload a 480x832 image**")
-                        input_image = gr.Image(
-                            label="",
-                            type="pil",
-                            height=400,
+        with gr.Row(equal_height=True, elem_classes="main-content-row"):
+            with gr.Column(scale=1, elem_classes="advanced-options-column"):
+                with gr.Group():
+                    gr.HTML("<div style='margin: 0 0 15px 0; text-align: center; font-size: 16px;'>Advanced Options</div>")
+                    with gr.Row():
+                        height = gr.Number(
+                            label="Height",
+                            value=initial_values['height'],
+                            interactive=False,
+                            container=True
+                        )
+                        width = gr.Number(
+                            label="Width",
+                            value=initial_values['width'],
+                            interactive=False,
+                            container=True
                         )
                     
-                    with gr.Tab("Advanced Options"):
-                        with gr.Group():
-                            with gr.Row():
-                                height = gr.Number(
-                                    label="Height",
-                                    value=initial_values['height'],
-                                    interactive=False,
-                                    container=True
-                                )
-                                width = gr.Number(
-                                    label="Width",
-                                    value=initial_values['width'],
-                                    interactive=False,
-                                    container=True
-                                )
-                            
-                            with gr.Row():
-                                num_frames = gr.Number(
-                                    label="Number of Frames",
-                                    value=initial_values['num_frames'],
-                                    interactive=False,
-                                    container=True
-                                )
-                                guidance_scale = gr.Number(
-                                    label="Guidance Scale",
-                                    value=1.0,
-                                    interactive=False,
-                                    container=True
-                                )
-                            
-                            with gr.Row():
-                                use_negative_prompt = gr.Checkbox(
-                                    label="Use negative prompt", value=False)
-                                negative_prompt = gr.Text(
-                                    label="Negative prompt",
-                                    max_lines=3,
-                                    lines=3,
-                                    placeholder="Enter a negative prompt",
-                                    visible=False,
-                                )
+                    with gr.Row():
+                        num_frames = gr.Number(
+                            label="Number of Frames",
+                            value=initial_values['num_frames'],
+                            interactive=False,
+                            container=True
+                        )
+                        guidance_scale = gr.Slider(
+                            label="Guidance Scale",
+                            minimum=1,
+                            maximum=12,
+                            value=initial_values['guidance_scale'],
+                        )
+                    
+                    with gr.Row():
+                        use_negative_prompt = gr.Checkbox(
+                            label="Use negative prompt", value=False)
+                        negative_prompt = gr.Text(
+                            label="Negative prompt",
+                            max_lines=3,
+                            lines=3,
+                            placeholder="Enter a negative prompt",
+                            visible=False,
+                        )
 
-                            # randomize_seed = gr.Checkbox(label="Randomize seed", value=False)
-                            seed_output = gr.Number(label="Used Seed", value=1000)
+                    seed = gr.Slider(
+                        label="Seed",
+                        minimum=0,
+                        maximum=1000000,
+                        step=1,
+                        value=initial_values['seed'],
+                    )
+                    randomize_seed = gr.Checkbox(label="Randomize seed", value=False)
+                    seed_output = gr.Number(label="Used Seed")
         
-            with gr.Column(scale=1):
+            with gr.Column(scale=1, elem_classes="video-column"):
                 result = gr.Video(
                     label="Generated Video", 
                     show_label=True,
-                    height=500,
+                    height=466,
+                    width=600,
                     container=True,
-                    autoplay=True,
+                    elem_classes="video-component"
                 )
         
         gr.HTML("""
@@ -506,10 +387,116 @@ def create_gradio_interface(backend_url: str, default_params: dict[str, Sampling
         }
         
         .gradio-container {
-            max-width: 1400px !important;
+            max-width: 1200px !important;
             margin: 0 auto !important;
         }
         
+        .main {
+            max-width: 1200px !important;
+            margin: 0 auto !important;
+        }
+        
+        .gr-form, .gr-box, .gr-group {
+            max-width: 1200px !important;
+        }
+        
+        .gr-video {
+            max-width: 500px !important;
+            margin: 0 auto !important;
+        }
+        
+        .main-content-row {
+            display: flex !important;
+            align-items: flex-start !important;
+            min-height: 500px !important;
+            gap: 20px !important;
+        }
+        
+        .advanced-options-column,
+        .video-column {
+            display: flex !important;
+            flex-direction: column !important;
+            flex: 1 !important;
+            min-height: 400px !important;
+            align-items: stretch !important;
+        }
+        
+        .video-column > * {
+            margin-top: 0 !important;
+        }
+        
+        .video-column .gr-video,
+        .video-component {
+            margin-top: 0 !important;
+            padding-top: 0 !important;
+        }
+        
+        .video-column .gr-video .gr-form {
+            margin-top: 0 !important;
+        }
+        
+        .advanced-options-column .gr-group,
+        .video-column .gr-video {
+            margin-top: 0 !important;
+            vertical-align: top !important;
+        }
+        
+        .advanced-options-column > *:last-child,
+        .video-column > *:last-child {
+            flex-grow: 0 !important;
+        }
+        
+        @media (max-width: 1400px) {
+            .main-content-row {
+                min-height: 600px !important;
+            }
+            
+            .advanced-options-column,
+            .video-column {
+                min-height: 600px !important;
+            }
+        }
+        
+        @media (max-width: 1200px) {
+            .main-content-row {
+                flex-direction: column !important;
+                align-items: stretch !important;
+            }
+            
+            .advanced-options-column,
+            .video-column {
+                min-height: auto !important;
+                width: 100% !important;
+            }
+        }
+        
+        .timing-card {
+            background: var(--background-fill-secondary) !important;
+            border: 1px solid var(--border-color-primary) !important;
+            color: var(--body-text-color) !important;
+            padding: 10px;
+            border-radius: 8px;
+            text-align: center;
+            min-height: 80px;
+            display: flex;
+            flex-direction: column;
+            justify-content: center;
+        }
+        
+        .timing-card-highlight {
+            background: var(--background-fill-primary) !important;
+            border: 2px solid var(--color-accent) !important;
+        }
+        
+        .performance-card {
+            background: var(--background-fill-secondary) !important;
+            border: 1px solid var(--border-color-primary) !important;
+            color: var(--body-text-color) !important;
+            padding: 10px;
+            border-radius: 6px;
+            text-align: center;
+        }
+        
         .gr-number input[readonly] {
             background-color: var(--background-fill-secondary) !important;
             border: 1px solid var(--border-color-primary) !important;
@@ -524,30 +511,18 @@ def create_gradio_interface(backend_url: str, default_params: dict[str, Sampling
         def on_example_select(example_label):
             if example_label and example_label in example_labels:
                 index = example_labels.index(example_label)
-                selected_prompt = examples[index]
-                selected_image_path = example_images[index] if index < len(example_images) else None
-                
-                if selected_image_path and os.path.exists(selected_image_path):
-                    try:
-                        with Image.open(selected_image_path) as img:
-                            selected_image = img.convert("RGB")
-                    except Exception:
-                        selected_image = None
-                else:
-                    selected_image = None
-                
-                return selected_prompt, selected_image
-            return "", None
+                return examples[index]
+            return ""
         
         example_dropdown.change(
             fn=on_example_select,
             inputs=example_dropdown,
-            outputs=[prompt, input_image],
+            outputs=prompt,
         )
         
         gr.HTML("""
         <div style="text-align: center; margin-top: 10px; margin-bottom: 15px;">
-            <p style="font-size: 16px; margin: 0;">The compute for this demo is generously provided by <a href="https://www.gmicloud.ai/" target="_blank">GMI Cloud</a>.  Note that this demo is meant as a preview of our distilled I2V model. Outside of few-step distillation, we have not yet fully optimized it for speed. Stay tuned for updates!</p>
+            <p style="font-size: 16px; margin: 0;">The compute for this demo is generously provided by <a href="https://www.gmicloud.ai/" target="_blank">GMI Cloud</a>. Note that this demo is meant to showcase FastWan's quality and that under a large number of requests, generation speed may be affected. We are also rate-limiting users to 3 requests per minute.</p>
         </div>
         """)
         
@@ -562,7 +537,6 @@ def create_gradio_interface(backend_url: str, default_params: dict[str, Sampling
                 selected_model = "FastWan2.1-T2V-1.3B"
             
             model_path = MODEL_PATH_MAPPING.get(selected_model)
-            show_image_input = is_i2v_model(selected_model)
             
             if model_path and model_path in default_params:
                 params = default_params[model_path]
@@ -571,29 +545,29 @@ def create_gradio_interface(backend_url: str, default_params: dict[str, Sampling
                     gr.update(value=params.width),
                     gr.update(value=params.num_frames),
                     gr.update(value=params.guidance_scale),
-                    gr.update(visible=show_image_input),
+                    gr.update(value=params.seed),
                 )
             
             return (
                 gr.update(value=448),
                 gr.update(value=832),
-                gr.update(value=20),
+                gr.update(value=61),
                 gr.update(value=3.0),
-                gr.update(visible=show_image_input),
+                gr.update(value=1024),
             )
         
         model_selection.change(
             fn=on_model_selection_change,
             inputs=model_selection,
-            outputs=[height, width, num_frames, guidance_scale, image_tab],
+            outputs=[height, width, num_frames, guidance_scale, seed],
         )
         
         def handle_generation(*args, progress=None, request: gr.Request = None):
-            model_selection, prompt, negative_prompt, use_negative_prompt, guidance_scale, num_frames, height, width, input_image = args
+            model_selection, prompt, negative_prompt, use_negative_prompt, seed, guidance_scale, num_frames, height, width, randomize_seed = args
             
-            result_path, seed_or_error, _ = generate_video(
-                prompt, negative_prompt, use_negative_prompt, guidance_scale, 
-                num_frames, height, width, model_selection, input_image, progress
+            result_path, seed_or_error, timing_details = generate_video(
+                prompt, negative_prompt, use_negative_prompt, seed, guidance_scale, 
+                num_frames, height, width, randomize_seed, model_selection, progress
             )
             
             if result_path and os.path.exists(result_path):
@@ -601,12 +575,14 @@ def create_gradio_interface(backend_url: str, default_params: dict[str, Sampling
                     result_path, 
                     seed_or_error, 
                     gr.update(visible=False),
+                    gr.update(visible=True, value=timing_details),
                 )
             else:
                 return (
                     None, 
                     seed_or_error, 
                     gr.update(visible=True, value=seed_or_error),
+                    gr.update(visible=False),
                 )
         
         run_button.click(
@@ -616,14 +592,14 @@ def create_gradio_interface(backend_url: str, default_params: dict[str, Sampling
                 prompt,
                 negative_prompt,
                 use_negative_prompt,
+                seed,
                 guidance_scale,
                 num_frames,
                 height,
                 width,
-                # randomize_seed,
-                input_image,
+                randomize_seed,
             ],
-            outputs=[result, seed_output, error_output],  # timing_display removed
+            outputs=[result, seed_output, error_output, timing_display],
             concurrency_limit=20,
         )
     
@@ -635,11 +611,8 @@ def main():
     parser.add_argument("--backend_url", type=str, default="http://localhost:8000",
                         help="URL of the Ray Serve backend")
     parser.add_argument("--t2v_model_paths", type=str,
-                        default="",
+                        default="FastVideo/FastWan2.1-T2V-1.3B-Diffusers,FastVideo/FastWan2.1-T2V-14B-Diffusers",
                         help="Comma separated list of paths to the T2V model(s)")
-    parser.add_argument("--i2v_model_paths", type=str,
-                        default="FastVideo/SFWan2.2-I2V-A14B-Preview-Diffusers",
-                        help="Comma separated list of paths to the I2V model(s)")
     parser.add_argument("--host", type=str, default="0.0.0.0",
                         help="Host to bind to")
     parser.add_argument("--port", type=int, default=7860,
@@ -648,15 +621,8 @@ def main():
     args = parser.parse_args()
     
     default_params = {}
-    
-    # Load T2V model params
-    t2v_paths = [p.strip() for p in args.t2v_model_paths.split(",") if p.strip()]
-    for model_path in t2v_paths:
-        default_params[model_path] = SamplingParam.from_pretrained(model_path)
-    
-    # Load I2V model params
-    i2v_paths = [p.strip() for p in args.i2v_model_paths.split(",") if p.strip()]
-    for model_path in i2v_paths:
+    model_paths = args.t2v_model_paths.split(",")
+    for model_path in model_paths:
         default_params[model_path] = SamplingParam.from_pretrained(model_path)
     
     demo = create_gradio_interface(args.backend_url, default_params)
@@ -664,8 +630,6 @@ def main():
     print(f"Starting Gradio frontend at http://{args.host}:{args.port}")
     print(f"Backend URL: {args.backend_url}")
     print(f"T2V Models: {args.t2v_model_paths}")
-    if args.i2v_model_paths:
-        print(f"I2V Models: {args.i2v_model_paths}")
     
     from fastapi import FastAPI, Request, HTTPException
     from fastapi.responses import HTMLResponse, FileResponse
@@ -710,23 +674,23 @@ def main():
             <meta charset="UTF-8" />
             <meta name="viewport" content="width=device-width, initial-scale=1.0" />
             
-            <title>CausalWan</title>
-            <meta name="title" content="CausalWan">
+            <title>FastWan</title>
+            <meta name="title" content="FastWan">
             <meta name="description" content="Make video generation go blurrrrrrr">
-            <meta name="keywords" content="FastVideo, video generation, AI, machine learning, CausalWan">
+            <meta name="keywords" content="FastVideo, video generation, AI, machine learning, FastWan">
             
             <meta property="og:type" content="website">
             <meta property="og:url" content="{base_url}/">
-            <meta property="og:title" content="CausalWan">
+            <meta property="og:title" content="FastWan">
             <meta property="og:description" content="Make video generation go blurrrrrrr">
             <meta property="og:image" content="{base_url}/logo.svg">
             <meta property="og:image:width" content="1200">
             <meta property="og:image:height" content="630">
-            <meta property="og:site_name" content="CausalWan">
+            <meta property="og:site_name" content="FastWan">
             
             <meta property="twitter:card" content="summary_large_image">
             <meta property="twitter:url" content="{base_url}/">
-            <meta property="twitter:title" content="CausalWan">
+            <meta property="twitter:title" content="FastWan">
             <meta property="twitter:description" content="Make video generation go blurrrrrrr">
             <meta property="twitter:image" content="{base_url}/logo.svg">
             <link rel="icon" type="image/png" sizes="32x32" href="/favicon.ico">
@@ -756,15 +720,7 @@ def main():
         app, 
         demo, 
         path="/gradio",
-        root_path="/gradio",
-        allowed_paths=[
-            os.path.abspath("outputs"), 
-            os.path.abspath("fastvideo-logos"),
-            os.path.abspath("prompts"),
-            os.path.abspath("images"),
-            os.path.abspath(tempfile.gettempdir()),
-            os.path.abspath(os.path.join(tempfile.gettempdir(), "gradio")),
-        ]
+        allowed_paths=[os.path.abspath("outputs"), os.path.abspath("fastvideo-logos")]
     )
     
     uvicorn.run(app, host=args.host, port=args.port)

examples/inference/gradio/serving/prompts.txt

@@ -1,15 +0,0 @@
-Some friends dancing and having fun together in circles, at a party surrounded by colored lights at a party, in a fancy old place, in a view from below them.
-A man wearing grey shorts jumps rope in a gym, weights and gym equipment in the background.
-Flying over a peninsula covered in bushy trees, while discovering the sea around it, painted a beautiful turquoise blue, on a sunny day.
-Skillful cyclist doing a wheelie on a bike while riding through a forest, on a dirt road, surrounded by many trees, in the morning.
-In the video, a lone rider guides a majestic horse across an expansive, open field as the sun sets in the background. The rider, dressed in a classic blue shirt and wide-brimmed hat, sits confidently in the saddle, silhouetted against the warm glow of the evening sky. The horse moves gracefully, its mane and tail flowing with each step, creating a sense of harmony between horse and rider. Surrounding the pair, towering trees form a natural border, their leaves gently rustling in the breeze. The shadows lengthen on the ground, accentuating the serene and timeless feel of the scene. The distant hills and wooden fences frame the horizon, adding depth to the tranquil landscape. A few horses graze peacefully in the background, blending into the pastoral setting. The overall ambiance evokes a sense of calmness and quietude, capturing a perfect moment in the golden light of dusk.
-A silver SUV drives along a winding, snow-covered mountain road, with dense pine trees blanketed in snow lining both sides. The scene is serene, with the vehicle moving smoothly, possibly on a winter journey or vacation. As the SUV disappears around the bend, another, darker SUV follows, creating a sense of motion and perspective on the snow-dusted asphalt. The towering, snow-laden rock formation to the right contrasts with the dark green of the pines, highlighting the peacefulness of the wintry landscape.
-A man and a woman playing in a field with grass, during a bright afternoon, while cars pass by in the distance.
-A saxophonist wearing a blazer dances while playing a song in a park.
-Romantic couple embracing and looking at each other in the middle of a forest, during a break on a road trip through nature.
-Young woman cleaning her house decorated with plants and decorations, while dancing happily to music in her headphones.
-Man dressed in 80's style dances very happily in his kitchen while listening to music on his radio and drinking wine.
-Pair of jazz musicians performing a song with their saxophone and trombone on an abandoned train.
-A young woman with short hair wearing pink sunglasses chews gum and makes a bubble gum with the city in the background.
-Natural aerial landscape with a relief covered with abundant trees and vegetation and a thick layer of mist.
-Loving couple sitting on a log on the shore of a lake outside, sharing an affectionate hug.

examples/inference/gradio/serving/ray_serve_backend.py

@@ -26,7 +26,6 @@ SEED_RANGE_MAX = 1_000_000
 SUPPORTED_MODELS = [
     "FastVideo/FastWan2.1-T2V-1.3B-Diffusers", 
     "FastVideo/FastWan2.2-TI2V-5B-FullAttn-Diffusers",
-    "FastVideo/SFWan2.2-I2V-A14B-Preview-Diffusers",
 ]
 
 MODEL_CONFIGS = {
@@ -43,13 +42,6 @@ MODEL_CONFIGS = {
         "dit_cpu_offload": True,
         "vae_cpu_offload": False,
         "VSA_sparsity": 0.9,
-    },
-    "I2V-A14B": {
-        "num_cpus": 15,
-        "text_encoder_cpu_offload": True,
-        "dit_cpu_offload": True,
-        "vae_cpu_offload": False,
-        "VSA_sparsity": 0.0,
     }
 }
 
@@ -66,7 +58,6 @@ class VideoGenerationRequest(BaseModel):
     randomize_seed: bool = False
     return_frames: bool = False
     model_path: Optional[str] = None
-    image_data: Optional[str] = None  # Base64 encoded image for I2V
 
 
 class VideoGenerationResponse(BaseModel):
@@ -100,38 +91,11 @@ def encode_video_to_base64(frames: List[np.ndarray], fps: int = DEFAULT_FPS) ->
         return ""
 
 
-def save_image_from_base64(image_data: str, output_dir: str) -> Optional[str]:
-    """Save base64 image data to a temporary file and return the path."""
-    if not image_data:
-        return None
-    
-    try:
-        # Remove data URL prefix if present
-        if image_data.startswith('data:image/'):
-            image_data = image_data.split(',')[1]
-        
-        image_bytes = base64.b64decode(image_data)
-        
-        # Save to temporary file
-        os.makedirs(output_dir, exist_ok=True)
-        temp_image_path = os.path.join(output_dir, f"temp_input_{int(time.time() * 1000)}.png")
-        
-        with open(temp_image_path, 'wb') as f:
-            f.write(image_bytes)
-        
-        return temp_image_path
-        
-    except Exception as e:
-        print(f"Warning: Failed to save image: {e}")
-        return None
-
-
 def setup_model_environment(model_path: str) -> None:
-    # if "fullattn" in model_path.lower():
-    #     os.environ["FASTVIDEO_ATTENTION_BACKEND"] = "FLASH_ATTN"
-    # else:
-    #     os.environ["FASTVIDEO_ATTENTION_BACKEND"] = "VIDEO_SPARSE_ATTN"
-    os.environ["FASTVIDEO_ATTENTION_BACKEND"] = "FLASH_ATTN"
+    if "fullattn" in model_path.lower():
+        os.environ["FASTVIDEO_ATTENTION_BACKEND"] = "FLASH_ATTN"
+    else:
+        os.environ["FASTVIDEO_ATTENTION_BACKEND"] = "VIDEO_SPARSE_ATTN"
     os.environ["FASTVIDEO_STAGE_LOGGING"] = "1"
 
 
@@ -193,41 +157,22 @@ class BaseModelDeployment:
             num_gpus=1,
             use_fsdp_inference=True,
             text_encoder_cpu_offload=config["text_encoder_cpu_offload"],
-            dmd_denoising_steps=[1000, 850, 700, 550, 350, 275, 200, 125], # TODO: hardocde for I2V
-            dit_precision="fp32",  # TODO: hardocde for I2V
             dit_cpu_offload=config["dit_cpu_offload"],
             vae_cpu_offload=config["vae_cpu_offload"],
             VSA_sparsity=config["VSA_sparsity"],
             enable_stage_verification=False,
         )
         self.default_params = SamplingParam.from_pretrained(self.model_path)
-        self.default_params.seed = 1000
-        self.default_params.num_frames = 73
-        self.default_params.width = 832
-        self.default_params.height = 480
 
     def generate_video(self, video_request: VideoGenerationRequest) -> VideoGenerationResponse:
         total_start_time = time.time()
         
         params = prepare_sampling_params(video_request, self.default_params)
 
-        # Save image if provided (for I2V)
-        image_path = None
-        if video_request.image_data:
-            image_path = save_image_from_base64(video_request.image_data, self.output_path)
-            if image_path is None:
-                return VideoGenerationResponse(
-                    video_data=None,
-                    seed=params.seed,
-                    success=False,
-                    error_message="Failed to save input image",
-                )
-
         inference_start_time = time.time()
         result = self.generator.generate_video(
             prompt=video_request.prompt,
             sampling_param=params,
-            image_path=image_path,
             save_video=False,
             return_frames=False,
         )
@@ -240,13 +185,6 @@ class BaseModelDeployment:
         encoding_time = time.time() - encoding_start_time
         
         total_time = time.time() - total_start_time
-        
-        # Clean up temporary image file
-        if image_path and os.path.exists(image_path):
-            try:
-                os.remove(image_path)
-            except Exception as e:
-                print(f"Warning: Failed to remove temporary image file {image_path}: {e}")
 
         return VideoGenerationResponse(
             video_data=video_data,
@@ -262,7 +200,7 @@ class BaseModelDeployment:
 
 
 @serve.deployment(
-    ray_actor_options={"num_cpus": 2, "num_gpus": 1, "runtime_env": {"conda": "demo-fv"}},
+    ray_actor_options={"num_cpus": 2, "num_gpus": 1, "runtime_env": {"conda": "fv"}},
 )
 class T2VModelDeployment(BaseModelDeployment):
     def __init__(self, t2v_model_path: str, output_path: str = "outputs"):
@@ -272,7 +210,7 @@ class T2VModelDeployment(BaseModelDeployment):
 
 
 @serve.deployment(
-    ray_actor_options={"num_cpus": 16, "num_gpus": 1, "runtime_env": {"conda": "demo-fv"}},
+    ray_actor_options={"num_cpus": 16, "num_gpus": 1, "runtime_env": {"conda": "fv"}},
 )
 class T2V14BModelDeployment(BaseModelDeployment):
     def __init__(self, t2v_14b_model_path: str, output_path: str = "outputs"):
@@ -283,32 +221,18 @@ class T2V14BModelDeployment(BaseModelDeployment):
         print("✅ T2V 14B model initialized successfully")
 
 
-@serve.deployment(
-    ray_actor_options={"num_cpus": 15, "num_gpus": 1, "runtime_env": {"conda": "demo-fv"}},
-)
-class I2VModelDeployment(BaseModelDeployment):
-    def __init__(self, i2v_model_path: str, output_path: str = "outputs"):
-        super().__init__(i2v_model_path, output_path)
-        # Override environment for I2V model
-        os.environ["FASTVIDEO_ATTENTION_BACKEND"] = "FLASH_ATTN"
-        self._initialize_generator(MODEL_CONFIGS["I2V-A14B"])
-        print("✅ I2V model initialized successfully")
-
-
 app = FastAPI()
 limiter = Limiter(key_func=get_remote_address)
 app.state.limiter = limiter
 app.add_exception_handler(RateLimitExceeded, _rate_limit_exceeded_handler)
 
 
-@serve.deployment(num_replicas=1, ray_actor_options={"num_cpus": 1})
+@serve.deployment(num_replicas=50, ray_actor_options={"num_cpus": 2})
 @serve.ingress(app)
 class FastVideoAPI:
 
-    def __init__(self, t2v_deployments: Dict[str, DeploymentHandle], i2v_deployments: Dict[str, DeploymentHandle] = None):
+    def __init__(self, t2v_deployments: Dict[str, DeploymentHandle]):
         self.t2v_deployments = t2v_deployments
-        self.i2v_deployments = i2v_deployments or {}
-        self.all_deployments = {**self.t2v_deployments, **self.i2v_deployments}
         
         # Initialize Prometheus metrics
         self.request_count = Counter('fastvideo_requests_total', 'Total FastVideo requests', ['model_type', 'status'])
@@ -333,10 +257,10 @@ class FastVideoAPI:
         model_name = self._get_model_name(video_request.model_path)
         
         try:
-            if video_request.model_path not in self.all_deployments:
+            if video_request.model_path not in self.t2v_deployments:
                 raise ValueError(f"Model {video_request.model_path} not found")
             
-            response_ref = self.all_deployments[video_request.model_path].generate_video.remote(video_request)
+            response_ref = self.t2v_deployments[video_request.model_path].generate_video.remote(video_request)
             response = await response_ref
             
             self._record_metrics(model_name, "success", time.time() - start_time, response)
@@ -367,21 +291,18 @@ class FastVideoAPI:
 
 
 def validate_configuration(model_paths: List[str], replicas: List[int]) -> None:
-    assert len(model_paths) > 0, "At least one model must be specified"
     assert len(model_paths) == len(replicas), "Number of models and replicas must match"
     assert sum(replicas) <= NUM_GPUS, f"Total replicas ({sum(replicas)}) must be <= {NUM_GPUS}"
     
     for model, replica_count in zip(model_paths, replicas):
-        assert model in SUPPORTED_MODELS, f"Model {model} not supported. Supported models: {SUPPORTED_MODELS}"
+        assert model in SUPPORTED_MODELS, f"Model {model} not supported"
         assert replica_count > 0, f"Replicas must be greater than 0"
 
 
 def start_ray_serve(
     *,
-    t2v_model_paths: str = "",
-    t2v_model_replicas: str = "",
-    i2v_model_paths: str = "",
-    i2v_model_replicas: str = "",
+    t2v_model_paths: str,
+    t2v_model_replicas: str,
     output_path: str = "outputs",
     host: str = "0.0.0.0",
     port: int = 8000,
@@ -389,39 +310,21 @@ def start_ray_serve(
     if not ray.is_initialized():
         ray.init()
 
-    # Parse T2V models
-    t2v_paths = [p.strip() for p in t2v_model_paths.split(",") if p.strip()]
-    t2v_reps = [int(r.strip()) for r in t2v_model_replicas.split(",") if r.strip()] if t2v_model_replicas else []
-    
-    # Parse I2V models
-    i2v_paths = [p.strip() for p in i2v_model_paths.split(",") if p.strip()]
-    i2v_reps = [int(r.strip()) for r in i2v_model_replicas.split(",") if r.strip()] if i2v_model_replicas else []
-    
-    # Validate configurations
-    all_paths = t2v_paths + i2v_paths
-    all_replicas = t2v_reps + i2v_reps
-    validate_configuration(all_paths, all_replicas)
+    model_paths = t2v_model_paths.split(",")
+    replicas = [int(r) for r in t2v_model_replicas.split(",")]
+    validate_configuration(model_paths, replicas)
 
-    # Create T2V deployments
     t2v_deps = {}
-    for model_path, replica_count in zip(t2v_paths, t2v_reps):
+    for model_path, replica_count in zip(model_paths, replicas):
         t2v_dep = T2VModelDeployment.options(num_replicas=replica_count).bind(model_path, output_path)
         t2v_deps[model_path] = t2v_dep
 
-    # Create I2V deployments
-    i2v_deps = {}
-    for model_path, replica_count in zip(i2v_paths, i2v_reps):
-        i2v_dep = I2VModelDeployment.options(num_replicas=replica_count).bind(model_path, output_path)
-        i2v_deps[model_path] = i2v_dep
-
-    api = FastVideoAPI.bind(t2v_deps, i2v_deps)
+    api = FastVideoAPI.bind(t2v_deps)
     serve.run(api, route_prefix="/", name="fast_video")
 
     print(f"Ray Serve backend started at http://{host}:{port}")
-    for model_path, replica_count in zip(t2v_paths, t2v_reps):
+    for model_path, replica_count in zip(model_paths, replicas):
         print(f"T2V Model: {model_path} | Replicas: {replica_count}")
-    for model_path, replica_count in zip(i2v_paths, i2v_reps):
-        print(f"I2V Model: {model_path} | Replicas: {replica_count}")
     print(f"Health check: http://{host}:{port}/health")
     print(f"Video generation endpoint: http://{host}:{port}/generate_video")
 
@@ -437,20 +340,12 @@ if __name__ == "__main__":
     parser = argparse.ArgumentParser(description="FastVideo Ray Serve Backend")
     parser.add_argument("--t2v_model_paths",
                         type=str,
-                        default="",
+                        default="FastVideo/FastWan2.1-T2V-1.3B-Diffusers,FastVideo/FastWan2.2-TI2V-5B-FullAttn-Diffusers",
                         help="Comma separated list of paths to the T2V model(s)")
     parser.add_argument("--t2v_model_replicas",
                         type=str,
-                        default="",
+                        default="4,4",
                         help="Comma separated list of number of replicas for the T2V model(s)")
-    parser.add_argument("--i2v_model_paths",
-                        type=str,
-                        default="FastVideo/SFWan2.2-I2V-A14B-Preview-Diffusers",
-                        help="Comma separated list of paths to the I2V model(s)")
-    parser.add_argument("--i2v_model_replicas",
-                        type=str,
-                        default="1",
-                        help="Comma separated list of number of replicas for the I2V model(s)")
     parser.add_argument("--output_path",
                         type=str,
                         default="outputs",
@@ -466,21 +361,13 @@ if __name__ == "__main__":
     
     args = parser.parse_args()
 
-    # Parse and validate all models
-    t2v_paths = [p.strip() for p in args.t2v_model_paths.split(",") if p.strip()]
-    t2v_reps = [int(r.strip()) for r in args.t2v_model_replicas.split(",") if r.strip()] if args.t2v_model_replicas else []
-    i2v_paths = [p.strip() for p in args.i2v_model_paths.split(",") if p.strip()]
-    i2v_reps = [int(r.strip()) for r in args.i2v_model_replicas.split(",") if r.strip()] if args.i2v_model_replicas else []
-    
-    all_paths = t2v_paths + i2v_paths
-    all_replicas = t2v_reps + i2v_reps
-    validate_configuration(all_paths, all_replicas)
+    model_paths = args.t2v_model_paths.split(",")
+    replicas = [int(r) for r in args.t2v_model_replicas.split(",")]
+    validate_configuration(model_paths, replicas)
     
     start_ray_serve(
         t2v_model_paths=args.t2v_model_paths,
         t2v_model_replicas=args.t2v_model_replicas,
-        i2v_model_paths=args.i2v_model_paths,
-        i2v_model_replicas=args.i2v_model_replicas,
         output_path=args.output_path,
         host=args.host,
         port=args.port,
@@ -489,4 +376,4 @@ if __name__ == "__main__":
     setup_signal_handlers()
     print("✅ FastVideo backend is running. Press Ctrl-C to stop.")
     while True:
-        time.sleep(3600) 
+        time.sleep(3600) 

examples/inference/gradio/serving/start.sh

@@ -1,5 +1,3 @@
-python examples/inference/gradio/serving/start_ray_serve_app.py \
---t2v_model_paths "" \
---t2v_model_replicas "" \
---i2v_model_paths "FastVideo/SFWan2.2-I2V-A14B-Preview-Diffusers" \
---i2v_model_replicas "1"
+python examples/inference/gradio/start_ray_serve_app.py \
+--t2v_model_paths "FastVideo/FastWan2.1-T2V-1.3B-Diffusers,FastVideo/FastWan2.2-TI2V-5B-FullAttn-Diffusers" \
+--t2v_model_replicas "4,4"

examples/inference/gradio/serving/start_ray_serve_app.py

@@ -20,10 +20,8 @@ DEFAULT_BACKEND_PORT = 8000
 DEFAULT_FRONTEND_HOST = "0.0.0.0"
 DEFAULT_FRONTEND_PORT = 7860
 DEFAULT_OUTPUT_PATH = "outputs"
-DEFAULT_T2V_MODELS = ""
-DEFAULT_T2V_REPLICAS = ""
-DEFAULT_I2V_MODELS = "FastVideo/SFWan2.2-I2V-A14B-Preview-Diffusers"
-DEFAULT_I2V_REPLICAS = "1"
+DEFAULT_T2V_MODELS = "FastVideo/FastWan2.1-T2V-1.3B-Diffusers,FastVideo/FastWan2.2-TI2V-5B-FullAttn-Diffusers"
+DEFAULT_T2V_REPLICAS = "4,4"
 
 HEALTH_CHECK_TIMEOUT = 5
 HEALTH_CHECK_MAX_RETRIES = 100
@@ -102,12 +100,6 @@ class ServiceManager:
             "port": self.args.backend_port
         }
         
-        # Add I2V parameters if provided
-        if self.args.i2v_model_paths:
-            backend_args["i2v_model_paths"] = self.args.i2v_model_paths
-        if self.args.i2v_model_replicas:
-            backend_args["i2v_model_replicas"] = self.args.i2v_model_replicas
-        
         self.backend_process = self._start_service("ray_serve_backend.py", backend_args, "backend")
         return self.backend_process
     
@@ -119,10 +111,6 @@ class ServiceManager:
             "port": self.args.frontend_port
         }
         
-        # Add I2V parameters if provided
-        if self.args.i2v_model_paths:
-            frontend_args["i2v_model_paths"] = self.args.i2v_model_paths
-        
         self.frontend_process = self._start_service("gradio_frontend.py", frontend_args, "frontend")
         return self.frontend_process
     
@@ -185,9 +173,6 @@ def print_startup_info(args: argparse.Namespace) -> None:
     print("=" * 50)
     print(f"T2V Models: {args.t2v_model_paths}")
     print(f"T2V Model Replicas: {args.t2v_model_replicas}")
-    if args.i2v_model_paths:
-        print(f"I2V Models: {args.i2v_model_paths}")
-        print(f"I2V Model Replicas: {args.i2v_model_replicas}")
     print(f"Output: {args.output_path}")
     print(f"Backend: http://{args.backend_host}:{args.backend_port}")
     print(f"Frontend: http://{args.frontend_host}:{args.frontend_port}")
@@ -205,14 +190,6 @@ def parse_arguments() -> argparse.Namespace:
                         type=str, 
                         default=DEFAULT_T2V_REPLICAS,
                         help="Comma separated list of number of replicas for the T2V model(s)")
-    parser.add_argument("--i2v_model_paths", 
-                        type=str,
-                        default=DEFAULT_I2V_MODELS,
-                        help="Comma separated list of paths to the I2V model(s)")
-    parser.add_argument("--i2v_model_replicas", 
-                        type=str, 
-                        default=DEFAULT_I2V_REPLICAS,
-                        help="Comma separated list of number of replicas for the I2V model(s)")
     parser.add_argument("--output_path", 
                         type=str, 
                         default=DEFAULT_OUTPUT_PATH,

fastvideo/configs/pipelines/registry.py

@@ -39,8 +39,6 @@ PIPE_NAME_TO_CONFIG: dict[str, type[PipelineConfig]] = {
     "FastVideo/Wan2.1-VSA-T2V-14B-720P-Diffusers": WanT2V720PConfig,
     "wlsaidhi/SFWan2.1-T2V-1.3B-Diffusers": SelfForcingWanT2V480PConfig,
     "rand0nmr/SFWan2.2-T2V-A14B-Diffusers": SelfForcingWan2_2_T2V480PConfig,
-    "FastVideo/SFWan2.2-I2V-A14B-Preview-Diffusers":
-    SelfForcingWan2_2_T2V480PConfig,
     "Wan-AI/Wan2.2-TI2V-5B-Diffusers": Wan2_2_TI2V_5B_Config,
     "Wan-AI/Wan2.2-T2V-A14B-Diffusers": Wan2_2_T2V_A14B_Config,
     "Wan-AI/Wan2.2-I2V-A14B-Diffusers": Wan2_2_I2V_A14B_Config,

fastvideo/configs/sample/registry.py

@@ -78,8 +78,6 @@ SAMPLING_PARAM_REGISTRY: dict[str, Any] = {
     # Causal Self-Forcing Wan2.2
     "rand0nmr/SFWan2.2-T2V-A14B-Diffusers":
     SelfForcingWan2_2_T2V_A14B_480P_SamplingParam,
-    "FastVideo/SFWan2.2-I2V-A14B-Preview-Diffusers":
-    SelfForcingWan2_2_T2V_A14B_480P_SamplingParam,
 
     # Cosmos2
     "nvidia/Cosmos-Predict2-2B-Video2World":

fastvideo/pipelines/stages/causal_denoising.py

@@ -123,12 +123,11 @@ class CausalDMDDenosingStage(DenoisingStage):
                                                   dtype=target_dtype,
                                                   device=latents.device)
 
-        def _get_kv_cache(timestep: float) -> list[dict]:
+        def _get_kv_cache(timestep: float) -> dict:
             if boundary_timestep is not None:
                 if timestep >= boundary_timestep:
                     return kv_cache1
                 else:
-                    assert kv_cache2 is not None, "kv_cache2 is not initialized"
                     return kv_cache2
             return kv_cache1
 
@@ -395,7 +394,7 @@ class CausalDMDDenosingStage(DenoisingStage):
         batch.latents = latents
         return batch
 
-    def _initialize_kv_cache(self, batch_size, dtype, device) -> list[dict]:
+    def _initialize_kv_cache(self, batch_size, dtype, device) -> None:
         """
         Initialize a Per-GPU KV cache aligned with the Wan model assumptions.
         """
@@ -432,7 +431,7 @@ class CausalDMDDenosingStage(DenoisingStage):
         return kv_cache1
 
     def _initialize_crossattn_cache(self, batch_size, max_text_len, dtype,
-                                    device) -> list[dict]:
+                                    device) -> None:
         """
         Initialize a Per-GPU cross-attention cache aligned with the Wan model assumptions.
         """

fastvideo/training/distillation_pipeline.py

@@ -729,9 +729,6 @@ class DistillationPipeline(TrainingPipeline):
                                      self.num_train_timestep, [1],
                                      device=self.device,
                                      dtype=torch.long)
-            world_group = get_world_group()
-            if world_group.world_size > 1:
-                world_group.broadcast(timestep, src=0)
 
             timestep = shift_timestep(
                 timestep,
@@ -844,9 +841,6 @@ class DistillationPipeline(TrainingPipeline):
                                             self.num_train_timestep, [1],
                                             device=self.device,
                                             dtype=torch.long)
-        world_group = get_world_group()
-        if world_group.world_size > 1:
-            world_group.broadcast(fake_score_timestep, src=0)
 
         fake_score_timestep = shift_timestep(
             fake_score_timestep,

fastvideo/worker/multiproc_executor.py

@@ -467,6 +467,10 @@ class WorkerMultiprocProc:
                             "output_batch": output_batch.output.cpu(),
                             "logging_info": logging_info
                         })
+                    else:
+                        result = self.worker.execute_method(
+                            method, *args, **kwargs)
+                        self.pipe.send(result)
                 else:
                     result = self.worker.execute_method(method, *args, **kwargs)
                     self.pipe.send(result)

mkdocs.yml

@@ -14,6 +14,7 @@ edit_uri: edit/main/docs/
 # Configuration
 theme:
   name: material
+  favicon: assets/logos/icon_simple.svg
   palette:
     - scheme: default
       toggle:

prompts/mixkit_i2v.jsonl

@@ -62,49 +62,5 @@
     {
         "prompt": "A solitary boat glides across the expansive, tranquil expanse of a serene lake. The vessel leaves a gentle wake behind, creating delicate ripples across the mirror-like surface. The water appears a rich shade of teal, seamlessly blending with the sky at the horizon. Silhouettes of distant trees are faintly visible, creating a picturesque backdrop that enhances the solitary journey of the boat. The sky is a calm gradient, shifting from soft oranges near the shore to the pale blues above. In the distance, a few slender poles emerge from the water, remnants of an old structure or natural formation. The mood of the scene is one of peace and solitude, with the boat journeying steadily through the quiet landscape. There is a sense of endless possibilities as the boat moves toward the unseen beyond the frame. The simplicity and stillness of the scene invite contemplation and reflection, encapsulating a perfect moment of quietude on the water.",
         "image_path": "images/mixkit-motorboat-on-a-large-lake-with-turquoise-blue-waters-4996.png"
-    },
-    {
-        "prompt": "A man wearing grey shorts jumps rope in a gym, weights and gym equipment in the background.",
-        "image_path": "images/gray_short_man.jpg"
-    },
-    {
-        "prompt": "Flying over a peninsula covered in bushy trees, while discovering the sea around it, painted a beautiful turquoise blue, on a sunny day.",
-        "image_path": "images/peninsula.jpg"
-    },
-    {
-        "prompt": "Skillful cyclist doing a wheelie on a bike while riding through a forest, on a dirt road, surrounded by many trees, in the morning.",
-        "image_path": "images/cyclist.jpg"
-    },
-    {
-        "prompt": "Some friends dancing and having fun together in circles, at a party surrounded by colored lights at a party, in a fancy old place, in a view from below them.",
-        "image_path": "images/friends.jpg"
-    },
-    {
-        "prompt": "A saxophonist wearing a blazer dances while playing a song in a park.",
-        "image_path": "images/saxophonist.jpg"
-    },
-    {
-        "prompt": "Romantic couple embracing and looking at each other in the middle of a forest, during a break on a road trip through nature.",
-        "image_path": "images/romance.jpg"
-    },
-    {
-        "prompt": "Man dressed in 80's style dances very happily in his kitchen while listening to music on his radio and drinking wine.",
-        "image_path": "images/80s_dance.jpg"
-    },
-    {
-        "prompt": "Pair of jazz musicians performing a song with their saxophone and trombone on an abandoned train.",
-        "image_path": "images/jazz.jpg"
-    },
-    {
-        "prompt": "A young woman with short hair wearing pink sunglasses chews gum and makes a bubble gum with the city in the background.",
-        "image_path": "images/pink.jpg"
-    },
-    {
-        "prompt": "Natural aerial landscape with a relief covered with abundant trees and vegetation and a thick layer of mist.",
-        "image_path": "images/natural.jpg"
-    },
-    {
-        "prompt": "Loving couple sitting on a log on the shore of a lake outside, sharing an affectionate hug.",
-        "image_path": "images/couple.jpg"
     }
-]
+]