ajati commited on
Commit
c372560
·
verified ·
1 Parent(s): 245fb8c

Upload config.json to main

Browse files
Files changed (1) hide show
  1. config.json +104 -0
config.json ADDED
@@ -0,0 +1,104 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "adaptive_patching_levels": 0,
3
+ "categorical_vocab_size_list": null,
4
+ "combine_quantiles_via_variance": true,
5
+ "context_length": 512,
6
+ "d_model": 128,
7
+ "decoder_adaptive_patching_levels": 0,
8
+ "decoder_d_model": 64,
9
+ "decoder_mode": "common_channel",
10
+ "decoder_num_layers": 4,
11
+ "decoder_raw_residual": false,
12
+ "decompose": true,
13
+ "distribution_output": "student_t",
14
+ "dropout": 0.2,
15
+ "enable_base_norm_always": false,
16
+ "enable_forecast_channel_mixing": false,
17
+ "exogenous_channel_indices": null,
18
+ "expansion_factor": 3,
19
+ "fcm_context_length": 1,
20
+ "fcm_gated_attn": true,
21
+ "fcm_mix_layers": 2,
22
+ "fcm_prepend_past": true,
23
+ "fcm_prepend_past_offset": null,
24
+ "fcm_use_mixer": false,
25
+ "fft_length": 0,
26
+ "forecast_loss_type": "joint",
27
+ "frequency_token_vocab_size": 5,
28
+ "gated_attn": true,
29
+ "head_dropout": 0.2,
30
+ "huber_delta": 1,
31
+ "init_embed": "pytorch",
32
+ "init_linear": "pytorch",
33
+ "init_processing": false,
34
+ "init_std": 0.02,
35
+ "joint_loss_weight": 1,
36
+ "light_mode": false,
37
+ "loss": "mae",
38
+ "mask_value": 0,
39
+ "masked_context_length": null,
40
+ "mode": "common_channel",
41
+ "model_type": "tinytimemixer",
42
+ "mq_cond_mode": "concat",
43
+ "mq_cond_path": "flatten",
44
+ "mq_decoder_d_model": 8,
45
+ "mq_eps": 1e-06,
46
+ "mq_hidden": 8,
47
+ "mq_kernel_size": 3,
48
+ "mq_q50_type": "mean",
49
+ "mq_use_decoder_pool": true,
50
+ "mq_use_positional": false,
51
+ "multi_quantile_head": true,
52
+ "multi_scale": true,
53
+ "norm_eps": 1e-05,
54
+ "norm_mlp": "LayerNorm",
55
+ "num_input_channels": 1,
56
+ "num_layers": 10,
57
+ "num_parallel_samples": 100,
58
+ "patch_last": true,
59
+ "patch_length": 9,
60
+ "patch_stride": 9,
61
+ "penalize_large_width_ratio": 0.0,
62
+ "point_extra_weight": 2,
63
+ "positional_encoding_type": "sincos",
64
+ "post_init": false,
65
+ "prediction_channel_indices": null,
66
+ "prediction_filter_length": null,
67
+ "prediction_length": 30,
68
+ "quantile": 0.5,
69
+ "quantile_levels": [
70
+ 0.1,
71
+ 0.2,
72
+ 0.3,
73
+ 0.4,
74
+ 0.5,
75
+ 0.6,
76
+ 0.7,
77
+ 0.8,
78
+ 0.9
79
+ ],
80
+ "register_tokens": 2,
81
+ "residual_context_length": 90,
82
+ "residual_loss_weight": 1,
83
+ "resolution_prefix_tuning": false,
84
+ "scaling": "std",
85
+ "self_attn": false,
86
+ "self_attn_heads": 1,
87
+ "transformers_version": "4.38.0",
88
+ "trend_adaptive_patching_levels": null,
89
+ "trend_d_model": 8,
90
+ "trend_decoder_d_model": 8,
91
+ "trend_decoder_num_layers": 2,
92
+ "trend_fft_length": null,
93
+ "trend_head_d_model": null,
94
+ "trend_loss_weight": 1,
95
+ "trend_multi_scale": null,
96
+ "trend_num_layers": 10,
97
+ "trend_patch_length": 64,
98
+ "trend_patch_stride": 64,
99
+ "trend_register_tokens": 2,
100
+ "use_decoder": true,
101
+ "use_fft_embedding": true,
102
+ "use_positional_encoding": false,
103
+ "width_penalty_mode": "boundary"
104
+ }