MiyamizuMitsuha commited on
Commit
479d45f
·
1 Parent(s): 33db9ac

Update app

Browse files
Files changed (1) hide show
  1. app.py +15 -14
app.py CHANGED
@@ -118,31 +118,31 @@ tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True, us
118
 
119
 
120
  def Vintern_1B_v2_ViTable_docvqa(image, question, chat_history=[]):
121
- pixel_values = load_image(image, max_num=12).to(torch.bfloat16).cuda()
122
 
123
- generation_config = dict(max_new_tokens= 1024, do_sample=False, num_beams = 3, repetition_penalty=2.0)
124
 
125
- # question = input("Question: ")
126
- question = '<image>\n' + question
127
- response = model.chat(tokenizer, pixel_values, question, generation_config)
128
- print(f'User: {question}\nAssistant: {response}')
129
- print("="*30)
130
 
131
 
132
- # Update the chat history
133
- chat_history.append((image, None))
134
- chat_history.append((question, None))
135
- chat_history.append((None, response))
136
 
137
- return chat_history
138
 
139
 
140
 
141
  interface = gr.Interface(
142
  fn=Vintern_1B_v2_ViTable_docvqa,
143
  inputs=[
144
- gr.Image(label="Upload Image", type="filepath", optional=True), # Image input
145
- gr.Textbox(label="Enter your question", optional=True), # Text input
146
  ],
147
  outputs=gr.Chatbot(label="Chat History"), # Chatbot-style output
148
  title="Vintern-1B-v2-ViTable-docvqa,",
@@ -150,6 +150,7 @@ interface = gr.Interface(
150
  allow_flagging="never",
151
  )
152
 
 
153
  # Launch the chatbot
154
  interface.launch()
155
 
 
118
 
119
 
120
  def Vintern_1B_v2_ViTable_docvqa(image, question, chat_history=[]):
121
+ pixel_values = load_image(image, max_num=12).to(torch.bfloat16).cuda()
122
 
123
+ generation_config = dict(max_new_tokens= 1024, do_sample=False, num_beams = 3, repetition_penalty=2.0)
124
 
125
+ # question = input("Question: ")
126
+ question = '<image>\n' + question
127
+ response = model.chat(tokenizer, pixel_values, question, generation_config)
128
+ print(f'User: {question}\nAssistant: {response}')
129
+ print("="*30)
130
 
131
 
132
+ # Update the chat history
133
+ chat_history.append((image, None))
134
+ chat_history.append((question, None))
135
+ chat_history.append((None, response))
136
 
137
+ return chat_history
138
 
139
 
140
 
141
  interface = gr.Interface(
142
  fn=Vintern_1B_v2_ViTable_docvqa,
143
  inputs=[
144
+ gr.Image(label="Upload Image", type="filepath"), # Image input
145
+ gr.Textbox(label="Enter your question"), # Text input
146
  ],
147
  outputs=gr.Chatbot(label="Chat History"), # Chatbot-style output
148
  title="Vintern-1B-v2-ViTable-docvqa,",
 
150
  allow_flagging="never",
151
  )
152
 
153
+
154
  # Launch the chatbot
155
  interface.launch()
156