Update ML Intern artifact metadata

Add v3.2 clean zip (all optimizations + OOD test included)

Add ood_generalization_test.py — tests model against realistic OOD scenarios to expose overfitting

Add optimization_test.py — run locally to validate all optimizations before committing

v3.2: app.py uses detector's trained encoder instead of DeterministicEncoder"

v3.2: update README — reflect cleaned codebase, v3.2 changes"

v3.2: remove redundant archive files (source is now on main branch directly)

v3.2: wire neuron search results into ENSEMBLE_CONFIGS, use trained encoder at eval, consolidate denoise_batch

v3.2: inference uses LearnableTemporalEncoder from ensemble, consolidated denoiser helper

v3.2: remove dead RateCodingEncoder (was explicitly marked 'do not use')"

v3.2: fix encoder train/inference mismatch — use LearnableTemporalEncoder at inference, load encoder from checkpoints

v3.2: early stopping on val_accuracy (not val_loss), save encoder state in checkpoint

v3.2: selector.denoise() uses consolidated denoise_batch helper

v3.2: use consolidated denoise_batch helper, remove legacy NQRDataset/get_balanced_loader/build_dataloaders

v3.2: deduplicate noise generation via generate_noise_sample(), remove tautological SNR verification

v3.2: add denoise_signal/denoise_batch consolidated helpers

v3.2: consolidated noise generation helper + denoise helpers in __init__

v3.2: apply config changes — tune epochs/patience, remove dead curriculum flags

Add v3.2 zip (7 bug/redundancy fixes)

v3.1 zip: dead neuron fix + differentiable encoder + dropout

v3.1: fix dead neuron bug + differentiable encoder + dropout [run_full_pipeline.py]

v3.1: fix dead neuron bug + differentiable encoder + dropout [nqr_snn/snn/train.py]

v3.1: fix dead neuron bug + differentiable encoder + dropout [nqr_snn/snn/encoder.py]

v3.1: fix dead neuron bug + differentiable encoder + dropout [nqr_snn/snn/model.py]

v3.1: fix dead neuron bug + differentiable encoder + dropout [nqr_snn/config.py]

v3.0: updated zip with universal data ingestion + Gradio UI

v3.0: add app.py — universal data ingestion + real-time inference + Gradio UI

v3.0: add nqr_snn/inference.py — universal data ingestion + real-time inference + Gradio UI

v3.0: add nqr_snn/data/ingest.py — universal data ingestion + real-time inference + Gradio UI

v3.0: add downloadable zip archive

v3.0: add downloadable zip archive

v3.0: Updated README with all optimizations

v3.0: optimize run_full_pipeline.py

v3.0: optimize nqr_snn/noise_injection/stress_test.py

v3.0: optimize nqr_snn/denoising/ssa_denoiser.py

v3.0: optimize nqr_snn/snn/neuron_search.py

v3.0: optimize nqr_snn/snn/ensemble.py

v3.0: optimize nqr_snn/snn/train.py

v3.0: optimize nqr_snn/snn/encoder.py

v3.0: optimize nqr_snn/snn/model.py

v3.0: optimize nqr_snn/data/generator.py

v3.0: optimize nqr_snn/data/dataset.py

v3.0: optimize nqr_snn/config.py

v2.0 FINAL: Complete README with 6-stage pipeline docs and verified results

v2.0 FINAL: Update archives with complete 6-stage pipeline

v2.0 FINAL: Complete 6-stage pipeline — denoise → parallel neuron search → train → evaluate → stress test. 99.17% mean accuracy.

v2.0: Update archives with parallel neuron search

v2.0: Parallel neuron search via ProcessPoolExecutor — 15 configs run simultaneously

v2.4: Fix all bottlenecks for >99% accuracy

v2.4: Fix all bottlenecks for >99% accuracy