test_v26

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+vae_decoder.keras filter=lfs diff=lfs merge=lfs -text
+vae_encoder.keras filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,84 @@
+---
+license: bsd-3-clause
+library_name: keras
+tags:
+- seti
+- radio-astronomy
+- anomaly-detection
+- beta-vae
+- random-forest
+---
+
+# Aetherscan
+
+[Breakthrough Listen](https://breakthroughinitiatives.org/initiative/1)'s deep-learning SETI
+pipeline: a two-stage architecture where a **Beta-VAE encoder** compresses each observation of
+a 6-observation cadence (3 ON / 3 OFF, ABACAD) into an 8-dimensional latent, and a **Random
+Forest** classifies the cadence's concatenated latents as a technosignature candidate or not.
+
+This repository carries the released model weights at stable filenames, versioned via git
+tags: training tags match the pipeline run's save tag (e.g. `final_v3`), and release tags
+(`vX.Y.Z`) mark blessed weights.
+
+**Training tag**: `test_v26`
+
+## Files
+
+| File | Description |
+|---|---|
+| `vae_encoder.keras` | Beta-VAE encoder (Keras) — the inference feature extractor |
+| `vae_decoder.keras` | Beta-VAE decoder (Keras) — for reconstruction/traversal analysis |
+| `random_forest.joblib` | Random Forest cadence classifier (joblib) |
+| `config.json` | Full resolved training configuration for this run |
+
+## Training configuration
+
+| Parameter | Value |
+|---|---|
+| Training rounds | `2` |
+| Epochs per round | `2` |
+| Beta-VAE samples per round | `200` |
+| Random Forest samples | `200` |
+| Curriculum schedule | `exponential` |
+| SNR base | `10` |
+| Initial SNR range | `40` |
+| Final SNR range | `10` |
+| Latent dimensions | `8` |
+| Beta (KL weight) | `1.5` |
+| Alpha (clustering weight) | `10.0` |
+| RF estimators | `1000` |
+
+The complete configuration is in `config.json`.
+
+## Evaluation (validation split)
+
+| Metric | Value |
+|---|---|
+| ROC AUC | 0.7812 |
+| Average precision | 0.7904 |
+| Classification threshold | 0.99 |
+| Validation samples | 40 |
+
+## Library versions
+
+| Library | Version |
+|---|---|
+| python | `3.12.3` |
+| tensorflow | `2.17.0` |
+| numpy | `1.26.4` |
+| scikit-learn | `1.5.2` |
+| huggingface_hub | `1.21.0` |
+
+## Usage
+
+Aetherscan inference downloads these weights by default when no local artifact paths are
+given (pin a version with `--hf-revision`):
+
+```bash
+python -m aetherscan.main inference --hf-revision test_v26 --inference-files <catalog.csv>
+```
+
+## Links & citation
+
+Source code, documentation, and issue tracker: [https://github.com/zachtheyek/Aetherscan](https://github.com/zachtheyek/Aetherscan).
+If you use Aetherscan in your research, please cite it via the repository's `CITATION.cff`.

config.json

@@ -0,0 +1,187 @@
+{
+  "paths": {
+    "data_path": "/datax/scratch/zachy/data/aetherscan",
+    "model_path": "/datax/scratch/zachy/models/aetherscan",
+    "output_path": "/datax/scratch/zachy/outputs/aetherscan"
+  },
+  "db": {
+    "get_connection_timeout": 60.0,
+    "stop_writer_timeout": 10.0,
+    "write_interval": 5.0,
+    "write_buffer_max_size": 100,
+    "write_retry_delay": 1.0,
+    "flush_timeout": 10.0
+  },
+  "manager": {
+    "n_processes": 32,
+    "chunks_per_worker": 4,
+    "pool_terminate_timeout": 10.0
+  },
+  "monitor": {
+    "get_gpu_timeout": 5.0,
+    "stop_monitor_timeout": 10.0,
+    "monitor_interval": 1.0,
+    "monitor_retry_delay": 1.0
+  },
+  "logger": {
+    "console_level": "INFO",
+    "file_level": "INFO",
+    "slack_level": "INFO",
+    "slack_enabled": true,
+    "slack_channel": null,
+    "slack_username": "Aetherscan",
+    "slack_timeout": 15.0,
+    "slack_retry_attempts": 3,
+    "slack_buffer_size": 100,
+    "slack_flush_interval": 60.0,
+    "slack_broadcast_level": "ERROR"
+  },
+  "beta_vae": {
+    "latent_dim": 8,
+    "dense_layer_size": 512,
+    "kernel_size": [
+      3,
+      3
+    ],
+    "beta": 1.5,
+    "alpha": 10.0
+  },
+  "rf": {
+    "n_estimators": 1000,
+    "bootstrap": true,
+    "max_features": "sqrt",
+    "n_jobs": -1,
+    "seed": 11
+  },
+  "gpu": {
+    "num_replicas": null,
+    "per_gpu_memory_limit_mb": null,
+    "nccl_num_packs": 2,
+    "use_async_allocator": true
+  },
+  "data": {
+    "num_observations": 6,
+    "width_bin": 4096,
+    "downsample_factor": 8,
+    "time_bins": 16,
+    "freq_resolution": 2.7939677238464355,
+    "time_resolution": 18.25361108,
+    "num_target_backgrounds": 45000,
+    "background_load_chunk_size": 15000,
+    "max_chunks_per_file": 1,
+    "inference_background_load_chunk_size": 50000,
+    "train_files": [
+      "real_filtered_LARGE_HIP110750.npy",
+      "real_filtered_LARGE_HIP13402.npy",
+      "real_filtered_LARGE_HIP8497.npy"
+    ],
+    "test_files": [
+      "real_filtered_LARGE_test_HIP15638.npy"
+    ],
+    "inference_files": null
+  },
+  "training": {
+    "num_training_rounds": 2,
+    "epochs_per_round": 2,
+    "num_samples_beta_vae": 200,
+    "num_samples_rf": 200,
+    "train_val_split": 0.8,
+    "per_replica_batch_size": 4,
+    "effective_batch_size": 20,
+    "per_replica_val_batch_size": 4,
+    "signal_injection_chunk_size": 50000,
+    "data_gen_task_size": 256,
+    "round_data_dir": null,
+    "overlap_data_generation": true,
+    "keep_round_data": false,
+    "plot_injection_subsampling_count": 100000,
+    "plot_injection_outlier_percentile": 99.0,
+    "latent_viz_num_cadences_per_type": 5,
+    "latent_viz_step_interval": 10,
+    "latent_viz_umap_fit_max_samples": 100000,
+    "latent_viz_umap_n_neighbors": [
+      5,
+      15,
+      30,
+      50
+    ],
+    "latent_viz_umap_min_dist": [
+      0.0,
+      0.1,
+      0.5
+    ],
+    "latent_viz_gif_max_frames": 500,
+    "latent_viz_gif_duration_ms": 100,
+    "latent_traversal_every_round": false,
+    "latent_traversal_num_steps": 7,
+    "latent_traversal_max_sigma": 3.0,
+    "shap_max_samples_summary": 5000,
+    "shap_max_samples_interaction": 1500,
+    "shap_top_k_features_dependence": 48,
+    "rf_decision_boundary_grid_size": 150,
+    "rf_decision_boundary_max_points": 5000,
+    "snr_base": 10,
+    "initial_snr_range": 40,
+    "final_snr_range": 10,
+    "curriculum_schedule": "exponential",
+    "exponential_decay_rate": -3.0,
+    "step_easy_rounds": 5,
+    "step_hard_rounds": 15,
+    "base_learning_rate": 0.001,
+    "min_learning_rate": 1e-06,
+    "min_pct_improvement": 0.001,
+    "patience_threshold": 3,
+    "reduction_factor": 0.2,
+    "max_retries": 1,
+    "retry_delay": 60
+  },
+  "inference": {
+    "encoder_path": null,
+    "rf_path": null,
+    "config_path": null,
+    "per_replica_batch_size": 2048,
+    "classification_threshold": 0.99,
+    "cadence_group_by_cols": [
+      "Target",
+      "Session",
+      "Band",
+      "Cadence ID",
+      "Frequency"
+    ],
+    "cadence_h5_path_col": ".h5 path",
+    "cadence_expected_obs": 6,
+    "coarse_channel_width": 1048576,
+    "parallel_coarse_chans": null,
+    "bandpass_method": "pfb",
+    "pfb_taps_per_channel": 12,
+    "bandpass_debug_plot": false,
+    "spline_order": 16,
+    "detection_window_size": 256,
+    "detection_step_size": 128,
+    "stat_threshold": 2048.0,
+    "stamp_width": 4096,
+    "store_downsampled_stamps": true,
+    "overlap_search": true,
+    "overlap_fraction": 0.5,
+    "discard_side_channels": false,
+    "side_channel_count": 0,
+    "preprocess_output_dir": null,
+    "inference_viz_enabled": true,
+    "stamp_gallery_top_k": 12,
+    "max_candidate_plots": 50,
+    "max_retries": 3,
+    "retry_delay": 60
+  },
+  "hf": {
+    "repo_id": "zachtheyek/aetherscan",
+    "upload_after_training": true,
+    "revision": null
+  },
+  "checkpoint": {
+    "load_dir": null,
+    "load_tag": null,
+    "start_round": 1,
+    "save_tag": "test_v26",
+    "force_tag": false
+  }
+}

random_forest.joblib

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:13c7595b2d2ac27c851ae369cbc801069bfd481e52fbe2fbd46a80804e22b41d
+size 3882585

vae_decoder.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c9ea1fcaaff25fc0fde39c92057e51c6eb5f5e92410b5e7607aa56f4cbd11e2
+size 18677748

vae_encoder.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6e451ee8fbee81ef4608c6ab1e07d1c33916c726fb7e00f8489c9ab62fcad4d7
+size 18667989