Update app.py

app.py

@@ -33,7 +33,7 @@ image_processor = CLIPImageProcessor.from_pretrained(vision_tower_name)
 ##########################################################################
 def generate_prompt(instruction):
     instruction = instruction.strip().replace('\r\n','\n').replace('\n\n','\n')
-    return f"\n{instruction}\n\nAssistant:"
+    return f"{instruction}\n\nAssistant:"
 
 def generate(
     ctx,
@@ -59,7 +59,8 @@ def generate(
             input_ids = pipeline.encode(ctx)
             print(input_ids)
             text_embs = model.w['emb.weight'][input_ids]
-            input_embs = torch.cat((image_features, text_embs), dim=0)[-ctx_limit:]
+            input_embs = torch.cat((image_features, text_embs), dim=0)
+            print(input_embs.shape)
             out, state = model.forward(embs=input_embs, state=None)
         else:
             input_ids = [token]