AlexCheema commited on
Commit
20a59fa
1 Parent(s): c38b3b1

Fix eos_token in tokenizer_config.json

Browse files

eos_token was set to "<|eot_id|>" when it should be "<|end_of_text|>"
This caused bugs like generation never ending because it never hits the eos_token.

Files changed (1) hide show
  1. tokenizer_config.json +1 -1
tokenizer_config.json CHANGED
@@ -2052,7 +2052,7 @@
2052
  "bos_token": "<|begin_of_text|>",
2053
  "chat_template": "{% set loop_messages = messages %}{% for message in loop_messages %}{% set content = '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n'+ message['content'] | trim + '<|eot_id|>' %}{% if loop.index0 == 0 %}{% set content = bos_token + content %}{% endif %}{{ content }}{% endfor %}{{ '<|start_header_id|>assistant<|end_header_id|>\n\n' }}",
2054
  "clean_up_tokenization_spaces": true,
2055
- "eos_token": "<|eot_id|>",
2056
  "model_input_names": [
2057
  "input_ids",
2058
  "attention_mask"
 
2052
  "bos_token": "<|begin_of_text|>",
2053
  "chat_template": "{% set loop_messages = messages %}{% for message in loop_messages %}{% set content = '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n'+ message['content'] | trim + '<|eot_id|>' %}{% if loop.index0 == 0 %}{% set content = bos_token + content %}{% endif %}{{ content }}{% endfor %}{{ '<|start_header_id|>assistant<|end_header_id|>\n\n' }}",
2054
  "clean_up_tokenization_spaces": true,
2055
+ "eos_token": "<|end_of_text|>",
2056
  "model_input_names": [
2057
  "input_ids",
2058
  "attention_mask"