Update app.py

app.py

@@ -207,37 +207,41 @@ def generate(
                     response_idx = output[generation_start_idx:].find("<response>")
 
                 # if `<request>` is in the output, highlight it, if `<call>` is in the output, highlight it
-                # print(generation_start_idx, request_idx, call_idx, response_idx)
+                # print("-------", generation_start_idx, request_idx, call_idx, response_idx)
                 highlighted_output = [
                     (prompt, ""),
                     (output[generation_start_idx:generation_start_idx+request_idx], ""),
                     (output[generation_start_idx+request_idx:generation_start_idx+call_idx], "request"),
                     (output[generation_start_idx+call_idx:-1], "call"),
                 ]
-                print(i, highlighted_output, output)
+                # print(i, highlighted_output, output)
                 yield highlighted_output
 
-                call_output = copy.deepcopy(output)
-                # response phase
-                generate_kwargs["stop_sequences"] = ["<submit>"]
-                stream = client.generate_stream(output, **generate_kwargs)
-                for response in stream:
-                    if response.token.text == "<|endoftext|>":
-                        return output
-                    else:
-                        output += response.token.text
-                    if submit_idx == -1:
-                        submit_idx = output[generation_start_idx:].find("<submit>")
-                    # print(generation_start_idx, request_idx, call_idx, response_idx, submit_idx)
-                    highlighted_output = [
-                        (prompt, ""),
-                        (output[generation_start_idx:generation_start_idx+request_idx], ""),
-                        (output[generation_start_idx+request_idx:generation_start_idx+call_idx], "request"),
-                        (output[generation_start_idx+call_idx:generation_start_idx+response_idx], "call"),
-                        (output[generation_start_idx+response_idx:-1], "submit"),
-                    ]
-                    print(highlighted_output, output)
-                    yield highlighted_output
+            # breakpoint()
+            call_output = copy.deepcopy(output)
+            print("start submit output")
+            # response phase
+            generate_kwargs["stop_sequences"] = ["<submit>"]
+            stream = client.generate_stream(output, **generate_kwargs)
+            for response in stream:
+                if response.token.text == "<|endoftext|>":
+                    return output
+                else:
+                    output += response.token.text
+                if submit_idx == -1:
+                    submit_idx = output[generation_start_idx:].find("<submit>")
+                # print("-------", generation_start_idx, request_idx, call_idx, response_idx, submit_idx)
+                highlighted_output = [
+                    (prompt, ""),
+                    (output[generation_start_idx:generation_start_idx+request_idx], ""),
+                    (output[generation_start_idx+request_idx:generation_start_idx+call_idx], "request"),
+                    (output[generation_start_idx+call_idx:generation_start_idx+response_idx], "call"),
+                    (output[generation_start_idx+response_idx:-1], "submit"),
+                ]
+                # print(highlighted_output, output)
+                yield highlighted_output
+            print("-------", generation_start_idx, request_idx, call_idx, response_idx, submit_idx)
+            print(highlighted_output, output)
 
             return highlighted_output
         except Exception as e: