Update README.md

README.md

@@ -6,11 +6,6 @@ tags:
 - gpt2
 - conversational
 license: apache-2.0
-datasets:
-- wikipedia-turkish
-metrics:
-- perplexity
-- accuracy
 widget:
 - text: Bu yazıyı bir bilgisayar yazdı. Yazarken
   context: ''
@@ -32,28 +27,67 @@ For obvious reasons I cannot share raw personafile but you can check above gist
 A working "full" demo can be seen in https://www.metayazar.com/chatbot
 For Turkish version (with limited training) https://www.metayazar.com/chatbot_tr
 
+Due to double LM head standart hugging face interface will not work. But if you follow huggingface tutorial should be same.
+Except each persona is encoded as "My name is XXXX"
+
+Use model, tokenizer and parameters within a class and call in below functions to trigger model.
+Some of the available personas:
+
+'''
+| Macleod | Moran | Brenda | Ramirez | Peter Parker | Quentin Beck | Andy 
+| Red | Norton | Willard | Chief | Chef | Kilgore | Kurtz | Westley | Buttercup 
+| Vizzini | Fezzik | Inigo | Man In Black | Taylor | Zira | Zaius | Cornelius 
+| Bud | Lindsey | Hippy | Erin | Ed | George | Donna | Trinity | Agent Smith 
+| Morpheus | Neo | Tank | Meryl | Truman | Marlon | Christof | Stromboli | Bumstead 
+| Schreber | Walker | Korben | Cornelius | Loc Rhod | Anakin | Obi-Wan | Palpatine 
+| Padme | Superman | Luthor | Dude | Walter | Donny | Maude | General | Starkiller 
+| Indiana | Willie | Short Round | John | Sarah | Terminator | Miller | Sarge | Reiben 
+| Jackson | Upham | Chuckie | Will | Lambeau | Sean | Skylar | Saavik | Spock 
+| Kirk | Bones | Khan | Kirk | Spock | Sybok | Scotty | Bourne | Pamela | Abbott 
+
+'''
+
 
 ```python
-tokenizer = AutoTokenizer.from_pretrained('microsoft/DialoGPT-small')
-model = AutoModelWithLMHead.from_pretrained('output-small')
-# Let's chat for 5 lines
-for step in range(100):
-    # encode the new user input, add the eos_token and return a tensor in Pytorch
-    new_user_input_ids = tokenizer.encode(input(">> User:") + tokenizer.eos_token, return_tensors='pt')
-    # print(new_user_input_ids)
-    # append the new user input tokens to the chat history
-    bot_input_ids = torch.cat([chat_history_ids, new_user_input_ids], dim=-1) if step > 0 else new_user_input_ids
-    # generated a response while limiting the total chat history to 1000 tokens, 
-    chat_history_ids = model.generate(
-        bot_input_ids, max_length=500,
-        pad_token_id=tokenizer.eos_token_id,  
-        no_repeat_ngram_size=3,       
-        do_sample=True, 
-        top_k=100, 
-        top_p=0.7,
-        temperature = 0.8
-    )
-    
-    # pretty print last ouput tokens from bot
-    print("AI: {}".format(tokenizer.decode(chat_history_ids[:, bot_input_ids.shape[-1]:][0], skip_special_tokens=True)))
+    def get_answer(self, input_text, personality, history, params=None):
+        
+        ##Check length of history (to save 1 computation!)
+        if len(history)>0:
+            #mostly it will be empty list so need a length check for performance
+            #would do string check also but just assume it is list of list of strings, as not public
+            
+            new_hist = [] 
+            for ele in history:
+                new_hist.append( self.tokenizer.encode(ele) )
+            history = new_hist.copy()
+
+        history.append(self.tokenizer.encode(input_text))
+
+        with torch.no_grad():
+            out_ids = self.sample_sequence(personality, history, self.tokenizer, self.model, params=params)
+        history.append(out_ids)
+        history = history[-(2*self.parameters['max_history']+1):]
+        out_text = self.tokenizer.decode(out_ids, skip_special_tokens=True)
+        #print(out_text)
+
+
+        history_decoded = []
+        for ele in history:
+            history_decoded.append(self.tokenizer.decode(ele))
+
+        return out_text, history_decoded, self.parameters
+
+
+
+    def predict(self, question, parameter_dict):
+        try:
+            answer = self.generate_text(question, model=self.model,
+                                        tokenizer=self.tokenizer,
+                                        parameter_dict=parameter_dict,
+                                        )
+            return answer
+        except Exception as e:
+            raise Exception(
+                "Runtime error see cloudwatch logs : {}".format(repr(e)))
+
 ```