diff options
Diffstat (limited to 'train_lora.py')
-rw-r--r-- | train_lora.py | 14 |
1 files changed, 7 insertions, 7 deletions
diff --git a/train_lora.py b/train_lora.py index a0cd174..538a7f7 100644 --- a/train_lora.py +++ b/train_lora.py | |||
@@ -261,10 +261,9 @@ def parse_args(): | |||
261 | "--lr_scheduler", | 261 | "--lr_scheduler", |
262 | type=str, | 262 | type=str, |
263 | default="one_cycle", | 263 | default="one_cycle", |
264 | help=( | 264 | choices=["linear", "cosine", "cosine_with_restarts", "polynomial", |
265 | 'The scheduler type to use. Choose between ["linear", "cosine", "cosine_with_restarts", "polynomial",' | 265 | "constant", "constant_with_warmup", "one_cycle"], |
266 | ' "constant", "constant_with_warmup", "one_cycle"]' | 266 | help='The scheduler type to use.', |
267 | ), | ||
268 | ) | 267 | ) |
269 | parser.add_argument( | 268 | parser.add_argument( |
270 | "--lr_warmup_epochs", | 269 | "--lr_warmup_epochs", |
@@ -288,7 +287,7 @@ def parse_args(): | |||
288 | "--lr_warmup_func", | 287 | "--lr_warmup_func", |
289 | type=str, | 288 | type=str, |
290 | default="cos", | 289 | default="cos", |
291 | help='Choose between ["linear", "cos"]' | 290 | choices=["linear", "cos"], |
292 | ) | 291 | ) |
293 | parser.add_argument( | 292 | parser.add_argument( |
294 | "--lr_warmup_exp", | 293 | "--lr_warmup_exp", |
@@ -300,7 +299,7 @@ def parse_args(): | |||
300 | "--lr_annealing_func", | 299 | "--lr_annealing_func", |
301 | type=str, | 300 | type=str, |
302 | default="cos", | 301 | default="cos", |
303 | help='Choose between ["linear", "half_cos", "cos"]' | 302 | choices=["linear", "half_cos", "cos"], |
304 | ) | 303 | ) |
305 | parser.add_argument( | 304 | parser.add_argument( |
306 | "--lr_annealing_exp", | 305 | "--lr_annealing_exp", |
@@ -318,7 +317,8 @@ def parse_args(): | |||
318 | "--optimizer", | 317 | "--optimizer", |
319 | type=str, | 318 | type=str, |
320 | default="dadan", | 319 | default="dadan", |
321 | help='Optimizer to use ["adam", "adam8bit", "lion", "dadam", "dadan", "adafactor"]' | 320 | choices=["adam", "adam8bit", "lion", "dadam", "dadan", "adafactor"], |
321 | help='Optimizer to use' | ||
322 | ) | 322 | ) |
323 | parser.add_argument( | 323 | parser.add_argument( |
324 | "--dadaptation_d0", | 324 | "--dadaptation_d0", |