transformers: add use_flash_attention_2 option (#4373)

2023-11-05 00:59:33 +08:00 · 2023-11-05 00:59:33 +08:00 · 4766a57352
commit 4766a57352
parent add359379e
6 changed files with 9 additions and 1 deletions
--- a/modules/ui.py
+++ b/modules/ui.py
@ -53,6 +53,7 @@ def list_model_elements():
        'load_in_8bit',
        'trust_remote_code',
        'use_fast',
+        'use_flash_attention_2',
        'load_in_4bit',
        'compute_dtype',
        'quant_type',