SequentialLearning
/

SuperLinear

mixture-of-experts

Model card Files Files and versions

razmars commited on May 14

Commit

dad1e57

·

verified ·

1 Parent(s): d93f13d

Update full_shot/config.json

Files changed (1) hide show

full_shot/config.json +2 -2

full_shot/config.json CHANGED Viewed

@@ -30,14 +30,14 @@
   "pred_len": 96,
   "seq_len": 512,
   "moe_norm": 0,
-  "top_k_experts": 12,
   "torch_dtype": "float32",
   "transformers_version": "4.40.1",
   "use_fft": 1,
   "train_epochs": 1,
   "patience": 2,
   "lradj": "type1",
-  "learning_rate": 0.001,
   "channel_ind": 0,
   "misc_moe" :12,
   "full_size":0

   "pred_len": 96,
   "seq_len": 512,
   "moe_norm": 0,
+  "top_k_experts": 20,
   "torch_dtype": "float32",
   "transformers_version": "4.40.1",
   "use_fft": 1,
   "train_epochs": 1,
   "patience": 2,
   "lradj": "type1",
+  "learning_rate": 0.005,
   "channel_ind": 0,
   "misc_moe" :12,
   "full_size":0