lirannoc commited on
Commit
5086864
·
verified ·
1 Parent(s): 4f95ef6

Update config.json

Browse files
Files changed (1) hide show
  1. config.json +1 -25
config.json CHANGED
@@ -11,20 +11,14 @@
11
  "_comment_model_architecture": "Model architecture parameters",
12
  "train_seq_len": 512,
13
  "train_pred_len": 96,
14
- "seq_len": 512,
15
- "pred_len": 96,
16
- "inf_pred_len": 96,
17
- "max_horizon": 96,
18
- "auto_regressive": 1,
19
 
20
  "_comment_moe": "MoE (Mixture of Experts) parameters",
21
- "moe_n_experts": 4,
22
  "top_k_experts": 12,
23
  "noisy_gating_std": 0.1,
24
  "moe_temp": 1.0,
25
  "moe_norm": false,
26
  "layer_type": "RLinear",
27
- "n_experts": 4,
28
  "comp_moe": 12,
29
  "freeze_experts": true,
30
 
@@ -35,27 +29,9 @@
35
  "_comment_experts": "Expert configuration",
36
  "freq_experts": "mean_naive_1/4_1/6_1/7_1/8_1/12_1/14_1/16_1/21_1/24_1/28_1/30_1/32_1/36_1/42_1/48_1/52_1/56_1/60_1/72_1/84_1/90_1/96_1/120_1/144_1/168_1/180_1/224_1/252_1/288_1/336_1/365_1/504_1/672_1/1008_1/1440_1/2016_1/3600",
37
 
38
- "_comment_loading": "Model loading and saving parameters",
39
- "load_linear": true,
40
- "load_weights_full": true,
41
- "linear_freq_weights_path": "./weights/linear_freq_weights/",
42
- "full_weights_path": "./weights/full_weights/checkpoint.pth",
43
-
44
  "_comment_training": "Training parameters",
45
  "resample_long_lookback": false,
46
 
47
- "_comment_legacy": "Legacy parameters for backward compatibility",
48
- "linear_checkpoints_path": "/cs/azencot_fsas/MoE/",
49
- "linear_checkpoints_dir": "checkpoints5",
50
- "manual_moe": 0,
51
- "misc_moe": 1,
52
- "noisy_gating_std_decay": 1,
53
- "ker_len": 50,
54
- "con": 0,
55
- "d_model": 512,
56
- "mlp_gating": 1,
57
- "dropout": 0.0,
58
-
59
  "_comment_system": "System and framework parameters",
60
  "model_type": "super_linear",
61
  "torch_dtype": "float32",
 
11
  "_comment_model_architecture": "Model architecture parameters",
12
  "train_seq_len": 512,
13
  "train_pred_len": 96,
 
 
 
 
 
14
 
15
  "_comment_moe": "MoE (Mixture of Experts) parameters",
16
+ "n_experts": 4,
17
  "top_k_experts": 12,
18
  "noisy_gating_std": 0.1,
19
  "moe_temp": 1.0,
20
  "moe_norm": false,
21
  "layer_type": "RLinear",
 
22
  "comp_moe": 12,
23
  "freeze_experts": true,
24
 
 
29
  "_comment_experts": "Expert configuration",
30
  "freq_experts": "mean_naive_1/4_1/6_1/7_1/8_1/12_1/14_1/16_1/21_1/24_1/28_1/30_1/32_1/36_1/42_1/48_1/52_1/56_1/60_1/72_1/84_1/90_1/96_1/120_1/144_1/168_1/180_1/224_1/252_1/288_1/336_1/365_1/504_1/672_1/1008_1/1440_1/2016_1/3600",
31
 
 
 
 
 
 
 
32
  "_comment_training": "Training parameters",
33
  "resample_long_lookback": false,
34
 
 
 
 
 
 
 
 
 
 
 
 
 
35
  "_comment_system": "System and framework parameters",
36
  "model_type": "super_linear",
37
  "torch_dtype": "float32",