v1

app.py

@@ -118,30 +118,8 @@ demo = gr.ChatInterface(fn=bot_streaming,
                         additional_inputs = [gr.Slider(0, 1, 0.9, label="temperature"), gr.Slider(1, 1024, 128, label="new_max_token"), gr.Slider(0, 1, 0.95, label="top_p")],
                         additional_inputs_accordion="Generation Hyperparameters",
                         theme=gr.themes.Soft(),
-                        title="☄️Meteor",
-                        description="Meteor is efficient 7B size Large Language and Vision Model built on the help of traversal of rationale.\n"
+                        title="TroL",
+                        description="TroL is efficient 1.8B, 3.8B, and 7B size Large Language and Vision Models built on new propagation strategy\n"
                                     "Its inference speed highly depends on assinging non-scheduled GPU. (Therefore, once all GPUs are busy, then inference may be taken in infinity)",
                         stop_btn="Stop Generation", multimodal=True)
-demo.launch()
-
-
-
-
-
-
-
-    
-
-
-
-
-# Generate
-with torch.inference_mode():
-    _inputs = model.eval_process(inputs=inputs,
-                                 data='demo',
-                                 tokenizer=tokenizer,
-                                 device='cuda:0',
-                                 img_token_number=image_token_number)
-    generate_ids = model.generate(**_inputs, max_new_tokens=256, use_cache=True)
-    response = output_filtering(tokenizer.batch_decode(generate_ids, skip_special_tokens=False)[0], model)
-print(response)
+demo.launch()