switch back to old chatbot

main.py

@@ -5,11 +5,11 @@ Based on: https://huggingface.co/docs/hub/spaces-sdks-docker-first-demo
 
 from fastapi import FastAPI, Request
 
-# from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
-# from transformers import T5Tokenizer, T5ForConditionalGeneration
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+from transformers import T5Tokenizer, T5ForConditionalGeneration
 
-import gpt4free
-from gpt4free import Provider, forefront
+# import gpt4free
+# from gpt4free import Provider, forefront
 
 
 token_size_limit = None
@@ -17,8 +17,8 @@ token_size_limit = None
 # FROM: https://huggingface.co/facebook/blenderbot-400M-distill?text=Hey+my+name+is+Thomas%21+How+are+you%3F
 
 # LAST USED
-# tokenizer = AutoTokenizer.from_pretrained("facebook/blenderbot-400M-distill")
-# model = AutoModelForSeq2SeqLM.from_pretrained("facebook/blenderbot-400M-distill")
+tokenizer = AutoTokenizer.from_pretrained("facebook/blenderbot-400M-distill")
+model = AutoModelForSeq2SeqLM.from_pretrained("facebook/blenderbot-400M-distill")
 
 # tokenizer = AutoTokenizer.from_pretrained("facebook/blenderbot-1B-distill")
 # model = AutoModelForSeq2SeqLM.from_pretrained("facebook/blenderbot-1B-distill")
@@ -49,18 +49,18 @@ async def Reply(req: Request):
     print(f'MSG: {msg}')
 
     # Hugging face
-    # input_ids = tokenizer(msg, return_tensors='pt').input_ids  # .to('cuda')
-    # output = model.generate(
-    #     input_ids[:, -token_size_limit:],
-    #     do_sample=True,
-    #     temperature=request.get('temperature', 0.9),
-    #     max_length=request.get('max_length', 100),
-    # )
-    # reply = tokenizer.batch_decode(output)[0]
+    input_ids = tokenizer(msg, return_tensors='pt').input_ids  # .to('cuda')
+    output = model.generate(
+        input_ids[:, -token_size_limit:],
+        do_sample=True,
+        temperature=request.get('temperature', 0.9),
+        max_length=request.get('max_length', 100),
+    )
+    reply = tokenizer.batch_decode(output)[0]
 
     # gpt4free
     # usage theb
-    reply = gpt4free.Completion.create(Provider.Theb, prompt=msg)
+    # reply = gpt4free.Completion.create(Provider.Theb, prompt=msg)
 
     print(f'REPLY: {reply}')
     return {'reply': reply}