SequentialLearning
/

SuperLinear

mixture-of-experts

Model card Files Files and versions

razmars commited on May 14

Commit

dbddb36

·

verified ·

1 Parent(s): 1348b52

Update full_shot/config.json

Files changed (1) hide show

full_shot/config.json +2 -2

full_shot/config.json CHANGED Viewed

@@ -34,10 +34,10 @@
   "torch_dtype": "float32",
   "transformers_version": "4.40.1",
   "use_fft": 1,
-  "train_epochs": 8,
   "patience": 5,
   "lradj": "type1",
-  "learning_rate": 0.005,
   "channel_ind": 0,
   "misc_moe" :12,
   "full_size":0

   "torch_dtype": "float32",
   "transformers_version": "4.40.1",
   "use_fft": 1,
+  "train_epochs": 1,
   "patience": 5,
   "lradj": "type1",
+  "learning_rate": 0.1,
   "channel_ind": 0,
   "misc_moe" :12,
   "full_size":0