debug the generated process

app.py

@@ -100,9 +100,7 @@ def generate_video(prompt, image=None, duration=3, guidance_scale=9, video_guida
 
 # Gradio interface
 with gr.Blocks() as demo:
-    gr.Markdown("# Pyramid Flow")
-    gr.Markdown("Pyramid Flow is a training-efficient Autoregressive Video Generation model based on Flow Matching. It is trained only on open-source datasets within 20.7k A100 GPU hours")
-    gr.Markdown("[[Paper](https://arxiv.org/pdf/2410.05954)], [[Model](https://huggingface.co/rain1011/pyramid-flow-sd3)], [[Code](https://github.com/jy0205/Pyramid-Flow)]")
+    gr.Markdown("# R1")
     
     with gr.Row():
         with gr.Column():
@@ -117,16 +115,7 @@ with gr.Blocks() as demo:
             t2v_generate_btn = gr.Button("Generate Video")
         with gr.Column():
             t2v_output = gr.Video(label=f"Generated Video")
-            gr.HTML("""
-                <div style="display: flex; flex-direction: column;justify-content: center; align-items: center; text-align: center;">
-                    <p style="display: flex;gap: 6px;">
-                         <a href="https://huggingface.co/spaces/Pyramid-Flow/pyramid-flow?duplicate=true">
-                            <img src="https://huggingface.co/datasets/huggingface/badges/resolve/main/duplicate-this-space-lg.svg" alt="Duplicate this Space">
-                        </a>
-                    </p>
-                    <p>to use privately and generate videos up to 10s at 24fps</p>
-                </div>
-                """)
+           
     gr.Examples(
         examples=[
             "A movie trailer featuring the adventures of the 30 year old space man wearing a red wool knitted motorcycle helmet, blue sky, salt desert, cinematic style, shot on 35mm film, vivid colors",

pyramid_dit/pyramid_dit_for_video_gen_pipeline.py

@@ -560,6 +560,12 @@ class PyramidDiTForVideoGeneration:
             else:
                 # prepare the condition latents
                 past_condition_latents = []
+                # if not generated_latents_list or len(generated_latents_list) == 0:
+                #     raise ValueError("No latent tensors generated - check previous steps ")
+
+                print(f"Number of generated latents: {len(generated_latents_list)}")
+                print(f"Shapes of generated latents: {[x.shape for x in generated_latents_list]}")
+                
                 clean_latents_list = self.get_pyramid_latent(torch.cat(generated_latents_list, dim=2), len(stages) - 1)
                 
                 for i_s in range(len(stages)):