ToddGoldfarb commited on
Commit
f3970b8
1 Parent(s): 77885a5

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +18 -20
README.md CHANGED
@@ -6,13 +6,11 @@ language:
6
  - en
7
  pipeline_tag: conversational
8
  ---
9
- # What is Cadet-Tiny?
10
 
11
- Inspired by Allen AI's **Cosmo-XL**, **Cadet-Tiny** is a _very small_ conversational model trained off of the **SODA** dataset. **Cadet-Tiny** is intended for inference at the edge (on something as small as a 2GB RAM Raspberry Pi).
12
 
13
- **Cadet-Tiny** is trained off of the **t5-small** pretrained model from Google, and is, as a result, is about 2% of the size of the **Cosmo-3B** model.
14
-
15
- This is my first SEQ2SEQ NLP Model I've ever made! I'm very excited to share it here on HuggingFace! :)
16
 
17
  If you have any questions, or any comments on improvements, please contact me at: **tcgoldfarb@gmail.com**
18
 
@@ -22,11 +20,11 @@ If you have any questions, or any comments on improvements, please contact me at
22
 
23
  Here is the link to the Google Colab file, where I walk through the process of training the model and using the SODA public dataset from AI2.
24
 
25
- https://colab.research.google.com/drive/1cx3Yujr_jGQkseqzXZW-2L0vEyEjds_s?usp=sharing
26
 
27
- # Get Started With Cadet-Tiny
28
 
29
- Use the code snippet below to get started with Cadet-Tiny!
30
 
31
  ```
32
  import torch
@@ -35,12 +33,12 @@ import colorful as cf
35
 
36
  cf.use_true_colors()
37
  cf.use_style('monokai')
38
- class CadetTinyAgent:
39
  def __init__(self):
40
- print(cf.bold | cf.purple("Waking up Cadet-Tiny..."))
41
  self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
42
- self.tokenizer = AutoTokenizer.from_pretrained("t5-small", model_max_length=512)
43
- self.model = AutoModelForSeq2SeqLM.from_pretrained("ToddGoldfarb/Cadet-Tiny", low_cpu_mem_usage=True).to(self.device)
44
  self.conversation_history = ""
45
 
46
  def observe(self, observation):
@@ -98,36 +96,36 @@ class CadetTinyAgent:
98
  continue_chat = ""
99
 
100
  # MODIFY THESE STRINGS TO YOUR LIKING :)
101
- situation_narrative = "Imagine you are Cadet-Tiny talking to ???."
102
- role_instruction = "You are Cadet-Tiny, and you are talking to ???."
103
 
104
  self.chat(situation_narrative, role_instruction)
105
  continue_chat = get_valid_input(cf.purple("Start a new conversation with new setup? [Y/N]:"), "Y")
106
  if continue_chat in ["N", "n"]:
107
  break
108
 
109
- print(cf.blue("CT: See you!"))
110
 
111
  def chat(self, situation_narrative, role_instruction):
112
  print(cf.green(
113
- "Cadet-Tiny is running! Input [RESET] to reset the conversation history and [END] to end the conversation."))
114
  while True:
115
  user_input = input("You: ")
116
  if user_input == "[RESET]":
117
  self.reset_history()
118
- print(cf.green("[Conversation history cleared. Chat with Cadet-Tiny!]"))
119
  continue
120
  if user_input == "[END]":
121
  break
122
  response = self.generate(situation_narrative, role_instruction, user_input)
123
- print(cf.blue("CT: " + response))
124
 
125
 
126
  def main():
127
  print(cf.bold | cf.blue("LOADING MODEL"))
128
 
129
- CadetTiny = CadetTinyAgent()
130
- CadetTiny.run()
131
 
132
 
133
  if __name__ == '__main__':
 
6
  - en
7
  pipeline_tag: conversational
8
  ---
9
+ # What is Cadet-Medium?
10
 
11
+ Inspired by Allen AI's **Cosmo-XL**, **Cadet-Medium** is a somewhat small conversational model trained off of the **SODA** dataset. **Cadet-Medium** is intended for inference at the edge (on something as small as a 2GB RAM Raspberry Pi).
12
 
13
+ **Cadet-Medium** is trained off of the **t5-base** pretrained model from Google.
 
 
14
 
15
  If you have any questions, or any comments on improvements, please contact me at: **tcgoldfarb@gmail.com**
16
 
 
20
 
21
  Here is the link to the Google Colab file, where I walk through the process of training the model and using the SODA public dataset from AI2.
22
 
23
+ https://colab.research.google.com/drive/1uekZ0gO3GqjPwno16tV1A4Gitrl7p3ur?usp=sharing
24
 
25
+ # Get Started With Cadet-Medium
26
 
27
+ Use the code snippet below to get started with Cadet-Medium!
28
 
29
  ```
30
  import torch
 
33
 
34
  cf.use_true_colors()
35
  cf.use_style('monokai')
36
+ class CadetMedAgent:
37
  def __init__(self):
38
+ print(cf.bold | cf.purple("Waking up Cadet-Medium..."))
39
  self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
40
+ self.tokenizer = AutoTokenizer.from_pretrained("t5-base", model_max_length=512)
41
+ self.model = AutoModelForSeq2SeqLM.from_pretrained("ToddGoldfarb/Cadet-Medium", low_cpu_mem_usage=True).to(self.device)
42
  self.conversation_history = ""
43
 
44
  def observe(self, observation):
 
96
  continue_chat = ""
97
 
98
  # MODIFY THESE STRINGS TO YOUR LIKING :)
99
+ situation_narrative = "Imagine you are Cadet-Medium talking to ???."
100
+ role_instruction = "You are Cadet-Medium, and you are talking to ???."
101
 
102
  self.chat(situation_narrative, role_instruction)
103
  continue_chat = get_valid_input(cf.purple("Start a new conversation with new setup? [Y/N]:"), "Y")
104
  if continue_chat in ["N", "n"]:
105
  break
106
 
107
+ print(cf.blue("CM: See you!"))
108
 
109
  def chat(self, situation_narrative, role_instruction):
110
  print(cf.green(
111
+ "Cadet-Medium is running! Input [RESET] to reset the conversation history and [END] to end the conversation."))
112
  while True:
113
  user_input = input("You: ")
114
  if user_input == "[RESET]":
115
  self.reset_history()
116
+ print(cf.green("[Conversation history cleared. Chat with Cadet-Medium!]"))
117
  continue
118
  if user_input == "[END]":
119
  break
120
  response = self.generate(situation_narrative, role_instruction, user_input)
121
+ print(cf.blue("CM: " + response))
122
 
123
 
124
  def main():
125
  print(cf.bold | cf.blue("LOADING MODEL"))
126
 
127
+ CadetMed = CadetMedAgent()
128
+ CadetMed.run()
129
 
130
 
131
  if __name__ == '__main__':