replace attention in the yaml config with an enum

2025-04-04 23:37:30 -04:00
parent 0d71b0aa5f
commit ba47adc24b
101 changed files with 268 additions and 122 deletions
--- a/examples/yi-34B-chat/qlora.yml
+++ b/examples/yi-34B-chat/qlora.yml
@@ -10,7 +10,8 @@ load_in_4bit: true
 sequence_len: 1024
 bf16: auto
 tf32: false
-flash_attention: true
+attention: flash
+
 special_tokens:
  bos_token: "<|startoftext|>"
  eos_token: "<|endoftext|>"