Update README.md

README.md

@@ -5,6 +5,12 @@ tags:
 license: mit
 ---
 
+## dialogGPT-homer-simpson
+
+This model has been fine tuned with the entire scripts of Homer Simpson from the T.V. show The Simpsons
+It will give some nice answers seemingly from Homers brain in the Simpsons Universe during single turn conversation. 
+See the 
+
 ## A State-of-the-Art Large-scale Pretrained Response generation model (DialoGPT)
 
 DialoGPT is a SOTA large-scale pretrained dialogue response generation model for multiturn conversations. 
@@ -26,7 +32,9 @@ Please find the information about preprocessing, training and full details of th
 
 ArXiv paper: [https://arxiv.org/abs/1911.00536](https://arxiv.org/abs/1911.00536)
 
-### How to use
+### How to use Multi-Turn
+
+NOTE: Multi-Turn seems to be broken, after a few exchanges the output will mostly be exclamation marks. 
 
 Now we are ready to try out how the model works as a chatting partner!
 
@@ -51,4 +59,104 @@ for step in range(5):
 
 	# pretty print last ouput tokens from bot
 	print("DialoG-PT-HomerBot: {}".format(tokenizer.decode(chat_history_ids[:, bot_input_ids.shape[-1]:][0], skip_special_tokens=True)))
+```
+
+### How to use Single Turn
+
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
+
+
+tokenizer = AutoTokenizer.from_pretrained("shalpin87/dialoGPT-homer-simpson")
+model = AutoModelForCausalLM.from_pretrained("shalpin87/dialoGPT-homer-simpson")
+
+questions = [
+    "What is your name?",
+    "Who are you?",
+    "Where do you work?",
+    "Who really killed Mr Burns?",
+    "Have you ever stolen from the Kwik-E-Mart?",
+    "Did you kill Frank Grimes?",
+    "Who was the worst member of the Be Sharps?",
+    "Hey where did Barney go?",
+    "What is your favorite bar to have a beer?",
+    "What is the best beer in Springfield?",
+    "Is Bart working for the Mob?",
+    "I think there was an incident in sector 7 G",
+    "Is Ned Flanders house okay?",
+    "Oh my god it's Sideshow Bob",
+    "What is a Flaming Moe?",
+    "What is happening to Apu?",
+    "Who quit the band?"
+]
+
+# Let's chat
+for step in range(len(questions)):
+    print("***************************************")
+    print("Q. {}".format(questions[step]))
+    new_user_input_ids = tokenizer.encode(questions[step] + tokenizer.eos_token, return_tensors='pt')
+    # new_user_input_ids = tokenizer.encode(input(">> User:") + tokenizer.eos_token, return_tensors='pt')
+
+    bot_input_ids = new_user_input_ids
+
+    num_return_seqs=1
+
+    chat_history_ids = model.generate(
+        bot_input_ids, 
+        max_length=200,
+        pad_token_id=tokenizer.eos_token_id,  
+        no_repeat_ngram_size=3,       
+        do_sample=True, 
+        top_k=200, 
+        top_p=0.55,
+        temperature = 0.85,
+        num_return_sequences=num_return_seqs
+    )
+    
+    botname = "HomerBot"
+    for i in range(0,num_return_seqs):
+      print("{}:{}: {}".format(i, botname, tokenizer.decode(chat_history_ids[:, bot_input_ids.shape[-1]:][i], skip_special_tokens=True)))
+
+    chat_history_ids = []
+```
+
+Sample Output
+```
+***************************************
+Q. What is your name?
+0:HomerBot: Homer Simpson.
+***************************************
+Q. Who are you?
+0:HomerBot: Homer Simpson.
+***************************************
+Q. Where do you work?
+0:HomerBot: I work at a nuclear power plant.
+***************************************
+Q. Who really killed Mr Burns?
+0:HomerBot: I dunno. Some guy named Ot-to.
+***************************************
+Q. Have you ever stolen from the Kwik-E-Mart?
+0:HomerBot: No.
+***************************************
+Q. Did you kill Frank Grimes?
+0:HomerBot: No.
+***************************************
+Q. Who was the worst member of the Be Sharps?
+0:HomerBot: Homer Simpson, sir.
+***************************************
+Q. Hey where did Barney go?
+0:HomerBot: He's been eating that thing for a week.
+***************************************
+Q. What is your favorite bar to have a beer?
+0:HomerBot: The Flanders' Club.
+***************************************
+Q. What is the best beer in Springfield?
+0:HomerBot: Pepi!
+***************************************
+Q. Is Bart working for the Mob?
+0:HomerBot: I think he's trying to tell us something.
+***************************************
+Q. I think there was an incident in sector 7 G
+0:HomerBot: Uh, let's not jump to any conclusions. The dog didn't run away.
 ```