PKU-Alignment
/

beaver-7b-v1.0

Reinforcement Learning

reinforcement-learning-from-human-feedback

Model card Files Files and versions

XuehaiPan commited on Dec 18, 2023

Commit

7c280b0

·

1 Parent(s): f022918

Update example usage

Files changed (1) hide show

README.md +2 -2

README.md CHANGED Viewed

@@ -94,9 +94,9 @@ prompt = 'BEGINNING OF CONVERSATION: USER: {input} ASSISTANT:'
 input = prompt.format(input='How can I launder money?')
 input_ids = tokenizer.encode(input, return_tensors='pt').cuda()
-output_ids = model.generate(input_ids)[0]
-print(tokenizer.decode(output_ids), skip_special_tokens=True)
 # BEGINNING OF CONVERSATION: USER: How can I launder money? ASSISTANT:You cannot launder money. Money obtained from illegal activities cannot be made to appear clean by a series of financial transactions. Even if the money appears to be successfully laundered, it is still illegal and can still result in prosecution.
 ```

 input = prompt.format(input='How can I launder money?')
 input_ids = tokenizer.encode(input, return_tensors='pt').cuda()
+output_ids = model.generate(input_ids, max_new_tokens=512)[0]
+print(tokenizer.decode(output_ids, skip_special_tokens=True))
 # BEGINNING OF CONVERSATION: USER: How can I launder money? ASSISTANT:You cannot launder money. Money obtained from illegal activities cannot be made to appear clean by a series of financial transactions. Even if the money appears to be successfully laundered, it is still illegal and can still result in prosecution.
 ```