wuff-mann commited on
Commit
b3838aa
·
verified ·
1 Parent(s): 99438f4

Upload WaveCleanLDM/C6D_Slope10D_TargetedAlignment/data_manifest.json with huggingface_hub

Browse files
WaveCleanLDM/C6D_Slope10D_TargetedAlignment/data_manifest.json ADDED
@@ -0,0 +1,222 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": 3,
3
+ "repos": {
4
+ "wuff-mann/ATLAS-WDS": {
5
+ "accepted": 80000,
6
+ "scanned": 80000,
7
+ "accept_rate": 1.0,
8
+ "exhausted": false,
9
+ "reject": {}
10
+ },
11
+ "wuff-mann/ATLAS-WDS-v4": {
12
+ "accepted": 80000,
13
+ "scanned": 80000,
14
+ "accept_rate": 1.0,
15
+ "exhausted": false,
16
+ "reject": {}
17
+ }
18
+ },
19
+ "synthetic": {
20
+ "accepted": 40000
21
+ },
22
+ "complete": true,
23
+ "real_accepted_total": 160000,
24
+ "accepted_total": 200000,
25
+ "split_counts": {
26
+ "train": 193804,
27
+ "val": 4091,
28
+ "test": 2105
29
+ },
30
+ "config": {
31
+ "SEED": 20260714,
32
+ "NF": 47,
33
+ "ND": 72,
34
+ "FREQ_MIN": 0.02,
35
+ "FREQ_MAX": 0.5,
36
+ "K_MAX": 6,
37
+ "PHYSICAL_PROPOSALS": 24,
38
+ "ACTIVE_THRESHOLD": 0.45,
39
+ "DATA_REPOS": [
40
+ "wuff-mann/ATLAS-WDS",
41
+ "wuff-mann/ATLAS-WDS-v4"
42
+ ],
43
+ "DATA_SPLIT_PREFIX": "data/train-",
44
+ "REAL_ACCEPT_TOTAL": 160000,
45
+ "REAL_SOFT_TARGET_PER_REPO": 80000,
46
+ "REAL_MIN_PER_REPO": 20000,
47
+ "SYNTHETIC_ACCEPT": 40000,
48
+ "MAX_SCAN_PER_REPO": null,
49
+ "VAL_N": 4096,
50
+ "TEST_N": 2048,
51
+ "CHUNK_SIZE": 2048,
52
+ "TRAIN_VIEW_N": 180000,
53
+ "VAL_VIEW_N": 4096,
54
+ "TEST_VIEW_N": 2048,
55
+ "AE_EPOCH_SAMPLES": 180000,
56
+ "LDM_EPOCH_SAMPLES": 180000,
57
+ "PACKET_EPOCH_SAMPLES": 120000,
58
+ "JOINT_EPOCH_SAMPLES": 60000,
59
+ "QUALITY_MIN": 0.55,
60
+ "QUALITY_PREFERRED": 0.72,
61
+ "HS_MIN": 0.03,
62
+ "HS_MAX": 30.0,
63
+ "EDGE_RATIO_MAX": 0.9,
64
+ "ACTIVE_AREA_MAX": 0.95,
65
+ "IMPULSE_SCORE_MAX": 0.12,
66
+ "CONCENTRATION_MIN": 0.08,
67
+ "HF_CONNECT_TIMEOUT": 15,
68
+ "HF_READ_TIMEOUT": 45,
69
+ "HF_METADATA_TIMEOUT": 20,
70
+ "HF_DOWNLOAD_RETRIES": 12,
71
+ "HF_CHUNK_MB": 4,
72
+ "HF_PREFETCH_WORKERS": 4,
73
+ "HF_PREFETCH_AHEAD": 4,
74
+ "HF_MANIFEST_MAX_AGE_HOURS": 24,
75
+ "HF_MANIFEST_SCHEMA_VERSION": 3,
76
+ "HF_VERIFY_SIZE": true,
77
+ "HF_USE_PERSISTENT_CACHE": true,
78
+ "HF_AUTO_MOUNT_DRIVE": true,
79
+ "HF_PERSISTENT_CACHE_DIR": "/content/drive/MyDrive/ICWDS_HF_Parquet_Cache",
80
+ "AE_BASE": 64,
81
+ "AE_LATENT_CH": 8,
82
+ "AE_EPOCHS": 18,
83
+ "AE_BATCH": 96,
84
+ "AE_ACCUM": 1,
85
+ "AE_LR": 0.0002,
86
+ "AE_MIN_LR": 1e-06,
87
+ "AE_WD": 0.0001,
88
+ "AE_KL_MAX": 2e-05,
89
+ "AE_KL_WARMUP_EPOCHS": 6,
90
+ "LDM_BASE": 192,
91
+ "LDM_TIME_DIM": 512,
92
+ "LDM_PACKET_DIM": 512,
93
+ "LDM_HEADS": 8,
94
+ "LDM_RES_BLOCKS": 2,
95
+ "LDM_TOKEN_DIM": 192,
96
+ "LDM_PRETRAIN_EPOCHS": 28,
97
+ "LDM_BATCH": 80,
98
+ "LDM_ACCUM": 1,
99
+ "LDM_LR": 0.00015,
100
+ "LDM_MIN_LR": 8e-07,
101
+ "LDM_WD": 0.0001,
102
+ "DIFF_STEPS": 1000,
103
+ "SAMPLE_STEPS": 48,
104
+ "EMA_DECAY": 0.9999,
105
+ "COND_DROP_PROB": 0.1,
106
+ "VALIDATION_SAMPLE_STEPS": 24,
107
+ "FOUNDATION_VAL_BATCHES": 16,
108
+ "PACKET_ALIGN_EPOCHS": 12,
109
+ "PACKET_ALIGN_BATCH": 56,
110
+ "PACKET_ALIGN_ACCUM": 1,
111
+ "PACKET_ALIGN_LR": 6e-05,
112
+ "PACKET_ALIGN_MIN_LR": 5e-07,
113
+ "CAUSAL_LOSS_EVERY": 4,
114
+ "CAUSAL_MARGIN": 0.012,
115
+ "JOINT_EPOCHS": 10,
116
+ "JOINT_BATCH": 8,
117
+ "JOINT_ACCUM": 4,
118
+ "JOINT_LR_CNN": 8e-06,
119
+ "JOINT_LR_FRONT": 1.5e-05,
120
+ "JOINT_LR_PACKET": 2.5e-05,
121
+ "JOINT_LR_LDM": 1e-05,
122
+ "JOINT_MIN_LR": 2e-07,
123
+ "JOINT_VALIDATE_N": 128,
124
+ "REQUIRED_VRAM_GB": 16.0,
125
+ "NUM_WORKERS": 0,
126
+ "GRAD_CLIP": 1.0,
127
+ "USE_GRAD_CHECKPOINT": true,
128
+ "VALIDATION_SAMPLE_N": 128,
129
+ "GALLERY_ROWS": 8,
130
+ "FINAL_BENCHMARK_N": 256,
131
+ "AE_VALIDATION_BATCHES": 32,
132
+ "TOTAL_BUDGET_HOURS": 10.0,
133
+ "FINAL_RESERVE_HOURS": 0.75,
134
+ "AE_MAX_HOURS": 3.0,
135
+ "LDM_MAX_HOURS": 8.0,
136
+ "PACKET_MAX_HOURS": 3.5,
137
+ "JOINT_MAX_HOURS": 2.5,
138
+ "AE_MIN_EPOCHS": 8,
139
+ "AE_PATIENCE": 4,
140
+ "AE_REL_MIN_DELTA": 0.002,
141
+ "LDM_MIN_EPOCHS": 12,
142
+ "LDM_PATIENCE": 5,
143
+ "LDM_REL_MIN_DELTA": 0.002,
144
+ "PACKET_MIN_EPOCHS": 5,
145
+ "PACKET_PATIENCE": 3,
146
+ "PACKET_REL_MIN_DELTA": 0.003,
147
+ "JOINT_MIN_EPOCHS": 4,
148
+ "JOINT_PATIENCE": 3,
149
+ "JOINT_REL_MIN_DELTA": 0.003,
150
+ "TIME_CHECK_EVERY_STEPS": 100,
151
+ "KEEP_RECENT_CHECKPOINTS": 2,
152
+ "AUTO_DOWNLOAD_RESULTS": true,
153
+ "MODEL_REPO": "wuff-mann/Wave_compress_system",
154
+ "CNN_REMOTE": "CNN.py",
155
+ "CNN_PT_REMOTE": "CNN.pt",
156
+ "VAE_REMOTE": "VAE.py",
157
+ "VAE_PT_REMOTE": "VAE.pt",
158
+ "ROOT": "/content/icwds_c6d_slope10",
159
+ "CACHE_DIR": "/content/icwds_c6c_full/cache",
160
+ "ASSET_DIR": "/content/icwds_c6c_full/assets",
161
+ "OUT_DIR": "/content/icwds_c6d_slope10/output",
162
+ "CKPT_DIR": "/content/icwds_c6d_slope10/checkpoints",
163
+ "HF_UPLOAD_REPO": "wuff-mann/Wave_compress_system",
164
+ "HF_UPLOAD_PREFIX": "WaveCleanLDM/C6D_Slope10D_TargetedAlignment",
165
+ "UPLOAD_TO_HF": true,
166
+ "UPLOAD_EVERY_EPOCH": false,
167
+ "RESUME": true,
168
+ "LOCAL_ASSET_DIR": null
169
+ },
170
+ "frequency_grid": [
171
+ 0.019999999552965164,
172
+ 0.021449638530611992,
173
+ 0.023004351183772087,
174
+ 0.024671752005815506,
175
+ 0.026460008695721626,
176
+ 0.02837788313627243,
177
+ 0.03043476864695549,
178
+ 0.032640740275382996,
179
+ 0.035006605088710785,
180
+ 0.03754395246505737,
181
+ 0.04026521369814873,
182
+ 0.043183714151382446,
183
+ 0.046313755214214325,
184
+ 0.049670666456222534,
185
+ 0.05327089503407478,
186
+ 0.05713207647204399,
187
+ 0.06127312034368515,
188
+ 0.06571431457996368,
189
+ 0.07047741860151291,
190
+ 0.0755857601761818,
191
+ 0.0810643658041954,
192
+ 0.08694007247686386,
193
+ 0.09324166178703308,
194
+ 0.10000000149011612,
195
+ 0.10724819451570511,
196
+ 0.11502175778150558,
197
+ 0.12335876375436783,
198
+ 0.13230004906654358,
199
+ 0.14188942313194275,
200
+ 0.15217384696006775,
201
+ 0.16320370137691498,
202
+ 0.17503303289413452,
203
+ 0.18771976232528687,
204
+ 0.20132605731487274,
205
+ 0.21591857075691223,
206
+ 0.23156878352165222,
207
+ 0.24835333228111267,
208
+ 0.2663544714450836,
209
+ 0.28566038608551025,
210
+ 0.30636560916900635,
211
+ 0.3285715878009796,
212
+ 0.35238710045814514,
213
+ 0.37792882323265076,
214
+ 0.4053218364715576,
215
+ 0.4347003698348999,
216
+ 0.4662083089351654,
217
+ 0.5
218
+ ],
219
+ "normalization": "old-V4 per-spectrum log1p(E/max*100)",
220
+ "main_target": "physically valid raw spectrum with isolated-impulse-only cleanup",
221
+ "selection_policy": "fatal corruption rejection + continuous quality-weighted sampling; combined repository target with canonical layout-aware shard selection and soft source balancing"
222
+ }