Update app.py

app.py

@@ -5,11 +5,13 @@ import torch
 import requests
 from PIL import Image
 import io
-from huggingface_hub import login
+from huggingface_hub import login  # Correct import for authentication
 import os
 
+# Read the Hugging Face access token from the environment variable
 read_token = os.getenv('AccToken')
 login(read_token)
+
 # Define a dictionary of conversational models
 conversational_models = {
     "Qwen": "Qwen/QwQ-32B",
@@ -48,17 +50,14 @@ document_qa_pipeline = pipeline("question-answering", model="deepset/roberta-bas
 image_classification_pipeline = pipeline("image-classification", model="facebook/detr-resnet-50")  # This will be replaced
 object_detection_pipeline = pipeline("object-detection", model="facebook/detr-resnet-50")
 video_classification_pipeline = pipeline("video-classification", model="facebook/timesformer-base-finetuned-k400")
-# Removed text_to_3d_pipeline as it was causing issues
-# Removed Keypoint Detection Pipeline
-# Removed Translation pipeline as it was causing issues
 summarization_pipeline = pipeline("summarization", model="facebook/bart-large-cnn")
 text_to_audio_pipeline = pipeline("text-to-audio", model="stabilityai/stable-audio-open-1.0")
 audio_classification_pipeline = pipeline("audio-classification", model="facebook/wav2vec2-base")
 
 def load_conversational_model(model_name):
     if model_name not in conversational_models_loaded:
-        tokenizer = AutoTokenizer.from_pretrained(conversational_models[model_name])
-        model = AutoModelForCausalLM.from_pretrained(conversational_models[model_name])
+        tokenizer = AutoTokenizer.from_pretrained(conversational_models[model_name], use_auth_token=read_token)
+        model = AutoModelForCausalLM.from_pretrained(conversational_models[model_name], use_auth_token=read_token)
         conversational_tokenizers[model_name] = tokenizer
         conversational_models_loaded[model_name] = model
     return conversational_tokenizers[model_name], conversational_models_loaded[model_name]
@@ -85,14 +84,18 @@ def chat(model_name, user_input, history=[]):
 
 def generate_image(model_name, prompt):
     if model_name not in text_to_image_pipelines:
-        text_to_image_pipelines[model_name] = StableDiffusionPipeline.from_pretrained(text_to_image_models[model_name])
+        text_to_image_pipelines[model_name] = StableDiffusionPipeline.from_pretrained(
+            text_to_image_models[model_name], use_auth_token=read_token
+        )
     pipeline = text_to_image_pipelines[model_name]
     image = pipeline(prompt).images[0]
     return image
 
 def generate_speech(model_name, text):
     if model_name not in text_to_speech_pipelines:
-        text_to_speech_pipelines[model_name] = pipeline("text-to-speech", model=text_to_speech_models[model_name])
+        text_to_speech_pipelines[model_name] = pipeline(
+            "text-to-speech", model=text_to_speech_models[model_name], use_auth_token=read_token
+        )
     pipeline = text_to_speech_pipelines[model_name]
     audio = pipeline(text)
     return audio["audio"]
@@ -195,9 +198,6 @@ with gr.Blocks() as demo:
         
         video_classification_generate.click(video_classification, inputs=video_classification_video, outputs=video_classification_output)
     
-    # Removed Text-to-3D tab as it was causing issues
-    # Removed Keypoint Detection tab due to issues.
-    
     with gr.Tab("Summarization"):
         summarize_text_text = gr.Textbox(label="Text")
         summarize_text_generate = gr.Button("Summarize")