Time Series Forecasting
TiRex-2
podestplatz commited on
Commit
49b3f7e
·
verified ·
1 Parent(s): e2ca13c

Preliminary model checkpoint

Browse files
Files changed (2) hide show
  1. model-config.yaml +69 -0
  2. model.ckpt +3 -0
model-config.yaml ADDED
@@ -0,0 +1,69 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ num_blocks: 12
2
+ embedding_dim: 512
3
+ input_patch_size: 32
4
+ output_patch_size: 32
5
+ input_ff_dim: 2048
6
+ h_expand: 4
7
+ context_len: 2048
8
+ future_len: 320
9
+ dropout: 0.1
10
+ quantiles:
11
+ - 0.1
12
+ - 0.2
13
+ - 0.3
14
+ - 0.4
15
+ - 0.5
16
+ - 0.6
17
+ - 0.7
18
+ - 0.8
19
+ - 0.9
20
+ act_func: torch.nn.SiLU
21
+ use_qk_norm: true
22
+ tokenizer_cfg:
23
+ input_patch_size: 32
24
+ output_patch_size: 32
25
+ input_patch_stride: 32
26
+ scaler_cfg:
27
+ use_arcsinh: true
28
+ binaryaware: true
29
+ stack_out_norm_config:
30
+ eps: 1.0e-07
31
+ tta_diff: true
32
+ stack_config:
33
+ templates:
34
+ slstm_attn:
35
+ time_mixer:
36
+ model_type: bi-slstm
37
+ embedding_dim: 512
38
+ num_heads: 4
39
+ rnn_backend: flashrnn-cuda
40
+ num_slstm_heads: 4
41
+ variate_mixer:
42
+ embedding_dim: 512
43
+ num_heads: 4
44
+ dropout: 0.1
45
+ eps: 1.0e-07
46
+ mlstm_attn:
47
+ time_mixer:
48
+ model_type: bi-mlstm
49
+ embedding_dim: 512
50
+ num_heads: 4
51
+ rnn_backend: flashrnn-cuda
52
+ variate_mixer:
53
+ embedding_dim: 512
54
+ num_heads: 4
55
+ dropout: 0.1
56
+ eps: 1.0e-07
57
+ recipe:
58
+ - mlstm_attn
59
+ - slstm_attn
60
+ - mlstm_attn
61
+ - slstm_attn
62
+ - mlstm_attn
63
+ - slstm_attn
64
+ - mlstm_attn
65
+ - slstm_attn
66
+ - mlstm_attn
67
+ - slstm_attn
68
+ - mlstm_attn
69
+ - slstm_attn
model.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1ab4b01a5a3d27eaeab19cc931f1627c8a5a7378279ae1b9957d7c73c2e32d64
3
+ size 330093831