improve gradio blocks interface

app.py

@@ -55,31 +55,32 @@ You can reset the interface anytime by clicking the **Reset** button.
 ### Join us : 
 🌟TeamTonic🌟 is always making cool demos! Join our active builder's 🛠️community 👻 [![Join us on Discord](https://img.shields.io/discord/1109943800132010065?label=Discord&logo=discord&style=flat-square)](https://discord.gg/qdfnvSPcqP) On 🤗Huggingface:[MultiTransformer](https://huggingface.co/MultiTransformer) On 🌐Github: [Tonic-AI](https://github.com/tonic-ai) & contribute to🌟 [Build Tonic](https://git.tonic-ai.com/contribute)🤗Big thanks to Yuvi Sharma and all the folks at huggingface for the community grant 🤗
 """
+
 model_presentation = f"""
-## PLeIAs/📸📈✍🏻Florence-PDF Model Overview
+## 🙏🏻PLeIAs/📸📈✍🏻Florence-PDF Model Overview
 
-The **PLeIAs/📸📈✍🏻Florence-PDF** model is a state-of-the-art model for conditional generation tasks, designed to be highly effective for both **text** and **vision** tasks. It is built as an **encoder-decoder** architecture, which allows for enhanced flexibility and performance in generating outputs based on diverse inputs.
+The **🙏🏻PLeIAs/📸📈✍🏻Florence-PDF** model is a state-of-the-art model for conditional generation tasks, designed to be highly effective for both **text** and **vision** tasks. It is built as an **encoder-decoder** architecture, which allows for enhanced flexibility and performance in generating outputs based on diverse inputs.
 
 ### Key Features
 
-- **Model Architecture**: PLeIAs/📸📈✍🏻Florence-PDF uses an encoder-decoder structure, which makes it effective in tasks like **text generation**, **summarization**, and **translation**. It has **{num_layers} layers** for both the encoder and decoder, with a model dimension (`d_model`) of **{d_model}**.
+- **Model Architecture**: 🙏🏻PLeIAs/📸📈✍🏻Florence-PDF uses an encoder-decoder structure, which makes it effective in tasks like **text generation**, **summarization**, and **translation**. It has **{num_layers} layers** for both the encoder and decoder, with a model dimension (`d_model`) of **{d_model}**.
 - **Conditional Generation**: The model can generate text conditionally, with a maximum length of **{max_length} tokens** for each generated sequence, making it ideal for tasks that require concise output.
-- **Beam Search**: PLeIAs/📸📈✍🏻Florence-PDFsupports **beam search** with up to **{beam_size} beams**, enabling more diverse and accurate text generation by exploring multiple potential outputs before selecting the best one.
+- **Beam Search**: 🙏🏻PLeIAs/📸📈✍🏻Florence-PDFsupports **beam search** with up to **{beam_size} beams**, enabling more diverse and accurate text generation by exploring multiple potential outputs before selecting the best one.
 - **Tokenization**: It includes a tokenizer with a vocabulary size of **{vocab_size} tokens**. Special tokens such as the **bos_token_id (0)** and **eos_token_id (2)** help control the generation process by marking the beginning and end of a sequence.
 - **Attention Mechanism**: Both the encoder and decoder utilize **{attention_heads} attention heads** per layer, ensuring that the model can focus on relevant parts of the input when generating text.
-- **Dropout and Activation**: PLeIAs/📸📈✍🏻Florence-PDF employs a **{activation_function} activation function** and a **dropout rate of {dropout}**, which enhances model performance by preventing overfitting and improving generalization.
+- **Dropout and Activation**: 🙏🏻PLeIAs/📸📈✍🏻Florence-PDF employs a **{activation_function} activation function** and a **dropout rate of {dropout}**, which enhances model performance by preventing overfitting and improving generalization.
 - **Training Configuration**: The model uses **float32** precision for training, and it supports fine-tuning for specific tasks by setting `finetuning_task` appropriately.
 
 ### Vision Integration
 
-In addition to text tasks, PLeIAs/📸📈✍🏻Florence-PDF also incorporates **vision capabilities**:
+In addition to text tasks, 🙏🏻PLeIAs/📸📈✍🏻Florence-PDF also incorporates **vision capabilities**:
 - **Patch-based Image Processing**: The vision component operates on image patches with a patch size of **{patch_size}x{patch_size}**.
 - **Temporal Embedding**: Visual tasks benefit from temporal embeddings with up to **{temporal_embeddings} steps**, making Florence-2 well-suited for video analysis.
 
 ### Model Usage and Flexibility
 
 - **No Repeat N-Grams**: To reduce repetition in text generation, the model is configured with a **no_repeat_ngram_size** of **{no_repeat_ngram_size}**, ensuring more diverse and meaningful outputs.
-- **Sampling Strategies**: PLeIAs/📸📈✍🏻Florence-PDF offers flexible sampling strategies, including **top-k** and **top-p (nucleus) sampling**, allowing for both creative and constrained generation based on user needs.
+- **Sampling Strategies**: 🙏🏻PLeIAs/📸📈✍🏻Florence-PDF offers flexible sampling strategies, including **top-k** and **top-p (nucleus) sampling**, allowing for both creative and constrained generation based on user needs.
 
 📸📈✍🏻Florence-PDF is a robust model capable of handling various **text and image** tasks with high precision and flexibility, making it a valuable tool for both academic research and practical applications.
 """
@@ -165,12 +166,12 @@ def draw_bounding_boxes(image, quad_boxes, labels, color=(0, 255, 0), thickness=
 
 def process_image(image, task):
     prompt = TASK_PROMPTS[task]
-    # Print the inputs for debugging
-    print(f"\n--- Processing Task: {task} ---")
-    print(f"Prompt: {prompt}")
+    # # Print the inputs for debugging
+    # print(f"\n--- Processing Task: {task} ---")
+    # print(f"Prompt: {prompt}")
     inputs = processor(text=prompt, images=image, return_tensors="pt").to(device, torch_dtype)
-    # Print the input tensors for debugging
-    print(f"Model Input: {inputs}")
+    # # Print the input tensors for debugging
+    # print(f"Model Input: {inputs}")
     generated_ids = model.generate(
         **inputs,
         max_new_tokens=1024,
@@ -179,11 +180,11 @@ def process_image(image, task):
     )
 
     generated_text = processor.batch_decode(generated_ids, skip_special_tokens=False)[0]
-    # Print the raw generated output for debugging
-    print(f"Raw Model Output: {generated_text}")
+    # # Print the raw generated output for debugging
+    # print(f"Raw Model Output: {generated_text}")
     parsed_answer = processor.post_process_generation(generated_text, task=prompt, image_size=(image.width, image.height))
-    # Print the parsed answer for debugging
-    print(f"Parsed Answer: {parsed_answer}")
+    # # Print the parsed answer for debugging
+    # print(f"Parsed Answer: {parsed_answer}")
     return parsed_answer
 
 
@@ -195,8 +196,8 @@ def main_process(image, task):
             fig = plot_bbox(image, result.get('<OCR_WITH_REGION>', {}), use_quad_boxes=True)
             output_image = fig_to_pil(fig)
             text_output = result.get('<OCR_WITH_REGION>', {}).get('recognized_text', 'No text found')
-            # Debugging: Print the recognized text
-            print(f"Recognized Text: {text_output}")
+            # # Debugging: Print the recognized text
+            # print(f"Recognized Text: {text_output}")
             return output_image, gr.update(visible=True), text_output, gr.update(visible=False)
         else:
             fig = plot_bbox(image, result.get(TASK_PROMPTS[task], {}))
@@ -209,22 +210,27 @@ def main_process(image, task):
 def reset_outputs():
     return None, gr.update(visible=False), None, gr.update(visible=True)
 
-with gr.Blocks(title="PLeIAs/📸📈✍🏻Florence-PDF") as iface:
-    gr.Markdown(title)
-    gr.Markdown(description)
-    
+with gr.Blocks(title="Tonic's 🙏🏻PLeIAs/📸📈✍🏻Florence-PDF") as iface:
     with gr.Column():
-        image_input = gr.Image(type="pil", label="Input Image")
-        task_dropdown = gr.Dropdown(list(TASK_PROMPTS.keys()), label="Task", value="✍🏻Caption")
-    
-    with gr.Row():
-        submit_button = gr.Button("Process")
-        reset_button = gr.Button("Reset")
-    
-    output_image = gr.Image(label="PLeIAs/📸📈✍🏻Florence-PDF", visible=False)
-    output_text = gr.Textbox(label="PLeIAs/📸📈✍🏻Florence-PDF", visible=True)
-    
-    gr.Markdown(model_presentation)
+        with gr.Row():
+            gr.Markdown(title)
+        with gr.Row():
+            with gr.Column(scale=1):
+                with gr.Group():                    
+                    gr.Markdown(model_presentation)
+            with gr.Column(scale=1):
+                with gr.Group():
+                    gr.Markdown(description)
+        with gr.Row():
+            with gr.Column(scale=1):
+                image_input = gr.Image(type="pil", label="Input Image")
+                task_dropdown = gr.Dropdown(list(TASK_PROMPTS.keys()), label="Task", value="✍🏻Caption")            
+                with gr.Row():
+                    submit_button = gr.Button("Process")
+                    reset_button = gr.Button("Reset")
+            with gr.Column(scale=1):    
+                output_image = gr.Image(label="🙏🏻PLeIAs/📸📈✍🏻Florence-PDF", visible=False)
+                output_text = gr.Textbox(label="🙏🏻PLeIAs/📸📈✍🏻Florence-PDF", visible=True)
 
     def process_and_update(image, task):
         if image is None: