t

app.py

@@ -2,7 +2,6 @@
 from random import randint
 from transformers import pipeline, set_seed
 import requests
-import json
 import gradio as gr
 # # from transformers import AutoModelForCausalLM, AutoTokenizer
 
@@ -33,16 +32,26 @@ top_p = gr.inputs.Slider(minimum=0, maximum=1.0,
 generator = pipeline('text-generation', model='gpt2')
 
 
-title = "GPT-J-6B/GPT-2 based text generator"
+title = "GPT-J-6B"
+
+# examples = [
+#     # another machine learning example
+#     ["For today's homework assignment, please describe the reasons for the US Civil War."],
+#     ["In a shocking discovery, scientists have found a herd of unicorns living in a remote, previously unexplored valley, in the Andes Mountains. Even more surprising to the researchers was the fact that the unicorns spoke perfect English."],
+#     ["The first step in the process of developing a new language is to invent a new word."],
+# ]
+title = "GPT-J-6B"
 
 examples = [
     # another machine learning example
-    [["For today's homework assignment, please describe the reasons for the US Civil War."],
-        0.8, 0.9, 50, "GPT-2"],
+    [["For today's homework assignment, please describe the reasons for the US Civil War."], 0.8, 0.9, 50, "GPT-2"],
     [["In a shocking discovery, scientists have found a herd of unicorns living in a remote, previously unexplored valley, in the Andes Mountains. Even more surprising to the researchers was the fact that the unicorns spoke perfect English."], 0.8, 0.9, 50, "GPT-2"],
-    [["The first step in the process of developing a new language is to invent a new word."],
-        0.8, 0.9, 50, "GPT-2"],
+    [["The first step in the process of developing a new language is to invent a new word."], 0.8, 0.9, 50, "GPT-2"],
 ]
+#             # # could easily use the inference API in /gptinference.py but don't know if it supports length>250
+#             set_seed(randint(1, 2**31))
+            # args found in the source: https://github.com/huggingface/transformers/blob/27b3031de2fb8195dec9bc2093e3e70bdb1c4bff/src/transformers/generation_tf_utils.py#L348-L376
+
 
 
 def f(context, temperature, top_p, max_length, model_idx):
@@ -61,21 +70,15 @@ def f(context, temperature, top_p, max_length, model_idx):
                 "temperature": temperature,
                 "top_p": top_p,
             }
-            
-            payload = json.dumps(payload)
             response = requests.post(
                 "http://api.vicgalle.net:5000/generate", params=payload).json()
             return response['text']
         else:
             # use GPT-2
-            # # could easily use the inference API in /gptinference.py but don't know if it supports length>250
+            #
             set_seed(randint(1, 2**31))
             # return sequences specifies how many to return
-            response = generator(context, max_length=max_length, top_p=top_p,
-                             temperature=temperature, num_return_sequences=1)
-            print(response)
-            return response  # ['generated_text']
-
+            return generator(context, max_length=max_length, top_p=top_p, temperature=temperature, num_return_sequences=1)
             # args found in the source: https://github.com/huggingface/transformers/blob/27b3031de2fb8195dec9bc2093e3e70bdb1c4bff/src/transformers/generation_tf_utils.py#L348-L376
 
     except Exception as e: