wuff-mann commited on
Commit
2d08443
·
verified ·
1 Parent(s): 29bc803

Upload WaveCleanLDM/C6C_FULL_HighCapacityResidualLDM/data_manifest.json with huggingface_hub

Browse files
WaveCleanLDM/C6C_FULL_HighCapacityResidualLDM/data_manifest.json ADDED
@@ -0,0 +1,191 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": 3,
3
+ "repos": {
4
+ "wuff-mann/ATLAS-WDS": {
5
+ "accepted": 100000,
6
+ "scanned": 100055,
7
+ "accept_rate": 0.9994503023337165,
8
+ "exhausted": false,
9
+ "reject": {
10
+ "Hs": 55
11
+ }
12
+ },
13
+ "wuff-mann/ATLAS-WDS-v4": {
14
+ "accepted": 100000,
15
+ "scanned": 100000,
16
+ "accept_rate": 1.0,
17
+ "exhausted": false,
18
+ "reject": {}
19
+ }
20
+ },
21
+ "synthetic": {
22
+ "accepted": 80000
23
+ },
24
+ "complete": true,
25
+ "real_accepted_total": 200000,
26
+ "accepted_total": 280000,
27
+ "split_counts": {
28
+ "train": 267637,
29
+ "val": 8322,
30
+ "test": 4041
31
+ },
32
+ "config": {
33
+ "SEED": 20260714,
34
+ "NF": 47,
35
+ "ND": 72,
36
+ "FREQ_MIN": 0.02,
37
+ "FREQ_MAX": 0.5,
38
+ "K_MAX": 6,
39
+ "PHYSICAL_PROPOSALS": 24,
40
+ "ACTIVE_THRESHOLD": 0.45,
41
+ "DATA_REPOS": [
42
+ "wuff-mann/ATLAS-WDS",
43
+ "wuff-mann/ATLAS-WDS-v4"
44
+ ],
45
+ "DATA_SPLIT_PREFIX": "data/train-",
46
+ "REAL_ACCEPT_TOTAL": 200000,
47
+ "REAL_SOFT_TARGET_PER_REPO": 100000,
48
+ "REAL_MIN_PER_REPO": 20000,
49
+ "SYNTHETIC_ACCEPT": 80000,
50
+ "MAX_SCAN_PER_REPO": null,
51
+ "VAL_N": 8192,
52
+ "TEST_N": 4096,
53
+ "CHUNK_SIZE": 2048,
54
+ "QUALITY_MIN": 0.55,
55
+ "QUALITY_PREFERRED": 0.72,
56
+ "HS_MIN": 0.03,
57
+ "HS_MAX": 30.0,
58
+ "EDGE_RATIO_MAX": 0.9,
59
+ "ACTIVE_AREA_MAX": 0.95,
60
+ "IMPULSE_SCORE_MAX": 0.12,
61
+ "CONCENTRATION_MIN": 0.08,
62
+ "HF_CONNECT_TIMEOUT": 15,
63
+ "HF_READ_TIMEOUT": 45,
64
+ "HF_METADATA_TIMEOUT": 20,
65
+ "HF_DOWNLOAD_RETRIES": 12,
66
+ "HF_CHUNK_MB": 4,
67
+ "HF_PREFETCH_WORKERS": 4,
68
+ "HF_PREFETCH_AHEAD": 4,
69
+ "HF_MANIFEST_MAX_AGE_HOURS": 24,
70
+ "HF_MANIFEST_SCHEMA_VERSION": 3,
71
+ "HF_VERIFY_SIZE": true,
72
+ "HF_USE_PERSISTENT_CACHE": true,
73
+ "HF_AUTO_MOUNT_DRIVE": true,
74
+ "HF_PERSISTENT_CACHE_DIR": "/content/drive/MyDrive/ICWDS_HF_Parquet_Cache",
75
+ "AE_BASE": 96,
76
+ "AE_LATENT_CH": 12,
77
+ "AE_EPOCHS": 36,
78
+ "AE_BATCH": 64,
79
+ "AE_ACCUM": 1,
80
+ "AE_LR": 0.0002,
81
+ "AE_MIN_LR": 1e-06,
82
+ "AE_WD": 0.0001,
83
+ "AE_KL_MAX": 2e-05,
84
+ "AE_KL_WARMUP_EPOCHS": 10,
85
+ "LDM_BASE": 256,
86
+ "LDM_TIME_DIM": 768,
87
+ "LDM_PACKET_DIM": 768,
88
+ "LDM_HEADS": 8,
89
+ "LDM_RES_BLOCKS": 2,
90
+ "LDM_PRETRAIN_EPOCHS": 64,
91
+ "LDM_BATCH": 96,
92
+ "LDM_ACCUM": 1,
93
+ "LDM_LR": 0.00015,
94
+ "LDM_MIN_LR": 8e-07,
95
+ "LDM_WD": 0.0001,
96
+ "DIFF_STEPS": 1000,
97
+ "SAMPLE_STEPS": 60,
98
+ "EMA_DECAY": 0.9999,
99
+ "COND_DROP_PROB": 0.1,
100
+ "PACKET_ALIGN_EPOCHS": 28,
101
+ "PACKET_ALIGN_BATCH": 48,
102
+ "PACKET_ALIGN_ACCUM": 1,
103
+ "PACKET_ALIGN_LR": 6e-05,
104
+ "PACKET_ALIGN_MIN_LR": 5e-07,
105
+ "CAUSAL_LOSS_EVERY": 4,
106
+ "CAUSAL_MARGIN": 0.012,
107
+ "JOINT_EPOCHS": 24,
108
+ "JOINT_BATCH": 8,
109
+ "JOINT_ACCUM": 4,
110
+ "JOINT_LR_CNN": 8e-06,
111
+ "JOINT_LR_FRONT": 1.5e-05,
112
+ "JOINT_LR_PACKET": 2.5e-05,
113
+ "JOINT_LR_LDM": 1e-05,
114
+ "JOINT_MIN_LR": 2e-07,
115
+ "JOINT_VALIDATE_N": 512,
116
+ "REQUIRED_VRAM_GB": 19.0,
117
+ "NUM_WORKERS": 2,
118
+ "GRAD_CLIP": 1.0,
119
+ "USE_GRAD_CHECKPOINT": true,
120
+ "VALIDATION_SAMPLE_N": 256,
121
+ "GALLERY_ROWS": 8,
122
+ "MODEL_REPO": "wuff-mann/Wave_compress_system",
123
+ "CNN_REMOTE": "CNN.py",
124
+ "CNN_PT_REMOTE": "CNN.pt",
125
+ "VAE_REMOTE": "VAE.py",
126
+ "VAE_PT_REMOTE": "VAE.pt",
127
+ "ROOT": "/content/icwds_c6c_full",
128
+ "CACHE_DIR": "/content/icwds_c6c_full/cache",
129
+ "ASSET_DIR": "/content/icwds_c6c_full/assets",
130
+ "OUT_DIR": "/content/icwds_c6c_full/output",
131
+ "CKPT_DIR": "/content/icwds_c6c_full/checkpoints",
132
+ "HF_UPLOAD_REPO": "wuff-mann/Wave_compress_system",
133
+ "HF_UPLOAD_PREFIX": "WaveCleanLDM/C6C_FULL_HighCapacityResidualLDM",
134
+ "UPLOAD_TO_HF": true,
135
+ "UPLOAD_EVERY_EPOCH": true,
136
+ "RESUME": true,
137
+ "LOCAL_ASSET_DIR": null
138
+ },
139
+ "frequency_grid": [
140
+ 0.019999999552965164,
141
+ 0.021449638530611992,
142
+ 0.023004351183772087,
143
+ 0.024671752005815506,
144
+ 0.026460008695721626,
145
+ 0.02837788313627243,
146
+ 0.03043476864695549,
147
+ 0.032640740275382996,
148
+ 0.035006605088710785,
149
+ 0.03754395246505737,
150
+ 0.04026521369814873,
151
+ 0.043183714151382446,
152
+ 0.046313755214214325,
153
+ 0.049670666456222534,
154
+ 0.05327089503407478,
155
+ 0.05713207647204399,
156
+ 0.06127312034368515,
157
+ 0.06571431457996368,
158
+ 0.07047741860151291,
159
+ 0.0755857601761818,
160
+ 0.0810643658041954,
161
+ 0.08694007247686386,
162
+ 0.09324166178703308,
163
+ 0.10000000149011612,
164
+ 0.10724819451570511,
165
+ 0.11502175778150558,
166
+ 0.12335876375436783,
167
+ 0.13230004906654358,
168
+ 0.14188942313194275,
169
+ 0.15217384696006775,
170
+ 0.16320370137691498,
171
+ 0.17503303289413452,
172
+ 0.18771976232528687,
173
+ 0.20132605731487274,
174
+ 0.21591857075691223,
175
+ 0.23156878352165222,
176
+ 0.24835333228111267,
177
+ 0.2663544714450836,
178
+ 0.28566038608551025,
179
+ 0.30636560916900635,
180
+ 0.3285715878009796,
181
+ 0.35238710045814514,
182
+ 0.37792882323265076,
183
+ 0.4053218364715576,
184
+ 0.4347003698348999,
185
+ 0.4662083089351654,
186
+ 0.5
187
+ ],
188
+ "normalization": "old-V4 per-spectrum log1p(E/max*100)",
189
+ "main_target": "physically valid raw spectrum with isolated-impulse-only cleanup",
190
+ "selection_policy": "fatal corruption rejection + continuous quality-weighted sampling; combined repository target with canonical layout-aware shard selection and soft source balancing"
191
+ }