tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 3 Packages Projects Releases Wiki Activity

fix reward trainer calls for tokenization

Browse Source
This commit is contained in:
Wing Lian
2025-01-07 15:41:40 -05:00
parent 808328e041
commit 47932f21c4
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/axolotl/prompt_strategies/bradley_terry/chat_template.py
View File

@@ -39,7 +39,7 @@ class BTChatTemplateStrategy(ChatTemplateStrategy):
)
prompt[self.messages].append({"role": "user", "content": prompt["input"]})
prompt[self.messages].append({"role": "assistant", "content": prompt["chosen"]})
chosen_tokenized = super().tokenize_prompt(prompt)
chosen_tokenized = super()._tokenize_single_prompt(prompt)
if len(chosen_tokenized["input_ids"]) > max_length:
LOG.warning(
@@ -62,7 +62,7 @@ class BTChatTemplateStrategy(ChatTemplateStrategy):
prompt[self.messages].append(
{"role": "assistant", "content": prompt["rejected"]}
)
rejected_tokenized = super().tokenize_prompt(prompt)
rejected_tokenized = super()._tokenize_single_prompt(prompt)
if len(rejected_tokenized["input_ids"]) > max_length:
LOG.warning(