wuff-mann's picture
Upload WaveCleanLDM/C6D_Slope10D_TargetedAlignment/data_manifest.json with huggingface_hub
b3838aa verified
Raw
History Blame Contribute Delete
6.12 kB
{
"version": 3,
"repos": {
"wuff-mann/ATLAS-WDS": {
"accepted": 80000,
"scanned": 80000,
"accept_rate": 1.0,
"exhausted": false,
"reject": {}
},
"wuff-mann/ATLAS-WDS-v4": {
"accepted": 80000,
"scanned": 80000,
"accept_rate": 1.0,
"exhausted": false,
"reject": {}
}
},
"synthetic": {
"accepted": 40000
},
"complete": true,
"real_accepted_total": 160000,
"accepted_total": 200000,
"split_counts": {
"train": 193804,
"val": 4091,
"test": 2105
},
"config": {
"SEED": 20260714,
"NF": 47,
"ND": 72,
"FREQ_MIN": 0.02,
"FREQ_MAX": 0.5,
"K_MAX": 6,
"PHYSICAL_PROPOSALS": 24,
"ACTIVE_THRESHOLD": 0.45,
"DATA_REPOS": [
"wuff-mann/ATLAS-WDS",
"wuff-mann/ATLAS-WDS-v4"
],
"DATA_SPLIT_PREFIX": "data/train-",
"REAL_ACCEPT_TOTAL": 160000,
"REAL_SOFT_TARGET_PER_REPO": 80000,
"REAL_MIN_PER_REPO": 20000,
"SYNTHETIC_ACCEPT": 40000,
"MAX_SCAN_PER_REPO": null,
"VAL_N": 4096,
"TEST_N": 2048,
"CHUNK_SIZE": 2048,
"TRAIN_VIEW_N": 180000,
"VAL_VIEW_N": 4096,
"TEST_VIEW_N": 2048,
"AE_EPOCH_SAMPLES": 180000,
"LDM_EPOCH_SAMPLES": 180000,
"PACKET_EPOCH_SAMPLES": 120000,
"JOINT_EPOCH_SAMPLES": 60000,
"QUALITY_MIN": 0.55,
"QUALITY_PREFERRED": 0.72,
"HS_MIN": 0.03,
"HS_MAX": 30.0,
"EDGE_RATIO_MAX": 0.9,
"ACTIVE_AREA_MAX": 0.95,
"IMPULSE_SCORE_MAX": 0.12,
"CONCENTRATION_MIN": 0.08,
"HF_CONNECT_TIMEOUT": 15,
"HF_READ_TIMEOUT": 45,
"HF_METADATA_TIMEOUT": 20,
"HF_DOWNLOAD_RETRIES": 12,
"HF_CHUNK_MB": 4,
"HF_PREFETCH_WORKERS": 4,
"HF_PREFETCH_AHEAD": 4,
"HF_MANIFEST_MAX_AGE_HOURS": 24,
"HF_MANIFEST_SCHEMA_VERSION": 3,
"HF_VERIFY_SIZE": true,
"HF_USE_PERSISTENT_CACHE": true,
"HF_AUTO_MOUNT_DRIVE": true,
"HF_PERSISTENT_CACHE_DIR": "/content/drive/MyDrive/ICWDS_HF_Parquet_Cache",
"AE_BASE": 64,
"AE_LATENT_CH": 8,
"AE_EPOCHS": 18,
"AE_BATCH": 96,
"AE_ACCUM": 1,
"AE_LR": 0.0002,
"AE_MIN_LR": 1e-06,
"AE_WD": 0.0001,
"AE_KL_MAX": 2e-05,
"AE_KL_WARMUP_EPOCHS": 6,
"LDM_BASE": 192,
"LDM_TIME_DIM": 512,
"LDM_PACKET_DIM": 512,
"LDM_HEADS": 8,
"LDM_RES_BLOCKS": 2,
"LDM_TOKEN_DIM": 192,
"LDM_PRETRAIN_EPOCHS": 28,
"LDM_BATCH": 80,
"LDM_ACCUM": 1,
"LDM_LR": 0.00015,
"LDM_MIN_LR": 8e-07,
"LDM_WD": 0.0001,
"DIFF_STEPS": 1000,
"SAMPLE_STEPS": 48,
"EMA_DECAY": 0.9999,
"COND_DROP_PROB": 0.1,
"VALIDATION_SAMPLE_STEPS": 24,
"FOUNDATION_VAL_BATCHES": 16,
"PACKET_ALIGN_EPOCHS": 12,
"PACKET_ALIGN_BATCH": 56,
"PACKET_ALIGN_ACCUM": 1,
"PACKET_ALIGN_LR": 6e-05,
"PACKET_ALIGN_MIN_LR": 5e-07,
"CAUSAL_LOSS_EVERY": 4,
"CAUSAL_MARGIN": 0.012,
"JOINT_EPOCHS": 10,
"JOINT_BATCH": 8,
"JOINT_ACCUM": 4,
"JOINT_LR_CNN": 8e-06,
"JOINT_LR_FRONT": 1.5e-05,
"JOINT_LR_PACKET": 2.5e-05,
"JOINT_LR_LDM": 1e-05,
"JOINT_MIN_LR": 2e-07,
"JOINT_VALIDATE_N": 128,
"REQUIRED_VRAM_GB": 16.0,
"NUM_WORKERS": 0,
"GRAD_CLIP": 1.0,
"USE_GRAD_CHECKPOINT": true,
"VALIDATION_SAMPLE_N": 128,
"GALLERY_ROWS": 8,
"FINAL_BENCHMARK_N": 256,
"AE_VALIDATION_BATCHES": 32,
"TOTAL_BUDGET_HOURS": 10.0,
"FINAL_RESERVE_HOURS": 0.75,
"AE_MAX_HOURS": 3.0,
"LDM_MAX_HOURS": 8.0,
"PACKET_MAX_HOURS": 3.5,
"JOINT_MAX_HOURS": 2.5,
"AE_MIN_EPOCHS": 8,
"AE_PATIENCE": 4,
"AE_REL_MIN_DELTA": 0.002,
"LDM_MIN_EPOCHS": 12,
"LDM_PATIENCE": 5,
"LDM_REL_MIN_DELTA": 0.002,
"PACKET_MIN_EPOCHS": 5,
"PACKET_PATIENCE": 3,
"PACKET_REL_MIN_DELTA": 0.003,
"JOINT_MIN_EPOCHS": 4,
"JOINT_PATIENCE": 3,
"JOINT_REL_MIN_DELTA": 0.003,
"TIME_CHECK_EVERY_STEPS": 100,
"KEEP_RECENT_CHECKPOINTS": 2,
"AUTO_DOWNLOAD_RESULTS": true,
"MODEL_REPO": "wuff-mann/Wave_compress_system",
"CNN_REMOTE": "CNN.py",
"CNN_PT_REMOTE": "CNN.pt",
"VAE_REMOTE": "VAE.py",
"VAE_PT_REMOTE": "VAE.pt",
"ROOT": "/content/icwds_c6d_slope10",
"CACHE_DIR": "/content/icwds_c6c_full/cache",
"ASSET_DIR": "/content/icwds_c6c_full/assets",
"OUT_DIR": "/content/icwds_c6d_slope10/output",
"CKPT_DIR": "/content/icwds_c6d_slope10/checkpoints",
"HF_UPLOAD_REPO": "wuff-mann/Wave_compress_system",
"HF_UPLOAD_PREFIX": "WaveCleanLDM/C6D_Slope10D_TargetedAlignment",
"UPLOAD_TO_HF": true,
"UPLOAD_EVERY_EPOCH": false,
"RESUME": true,
"LOCAL_ASSET_DIR": null
},
"frequency_grid": [
0.019999999552965164,
0.021449638530611992,
0.023004351183772087,
0.024671752005815506,
0.026460008695721626,
0.02837788313627243,
0.03043476864695549,
0.032640740275382996,
0.035006605088710785,
0.03754395246505737,
0.04026521369814873,
0.043183714151382446,
0.046313755214214325,
0.049670666456222534,
0.05327089503407478,
0.05713207647204399,
0.06127312034368515,
0.06571431457996368,
0.07047741860151291,
0.0755857601761818,
0.0810643658041954,
0.08694007247686386,
0.09324166178703308,
0.10000000149011612,
0.10724819451570511,
0.11502175778150558,
0.12335876375436783,
0.13230004906654358,
0.14188942313194275,
0.15217384696006775,
0.16320370137691498,
0.17503303289413452,
0.18771976232528687,
0.20132605731487274,
0.21591857075691223,
0.23156878352165222,
0.24835333228111267,
0.2663544714450836,
0.28566038608551025,
0.30636560916900635,
0.3285715878009796,
0.35238710045814514,
0.37792882323265076,
0.4053218364715576,
0.4347003698348999,
0.4662083089351654,
0.5
],
"normalization": "old-V4 per-spectrum log1p(E/max*100)",
"main_target": "physically valid raw spectrum with isolated-impulse-only cleanup",
"selection_policy": "fatal corruption rejection + continuous quality-weighted sampling; combined repository target with canonical layout-aware shard selection and soft source balancing"
}