Add comprehensive test suite for all foundation components (98 test cases)

Add __init__.py for new modules + training package init

Add upgraded SFT training script with SciRIFF data + proper QLoRA config

Add SciRIFF training data integration script (72x more data for training)

Add Epistemic Velocity tracking + Confidence Decomposition Display (Layer 5 upgrades)

Add Low Confidence Quarantine (system learns to say 'I don't know') + SciBERT-NLI contradiction pre-filter

Add Epistemic Trigger Words validator (deterministic code-based epistemic classification alongside AI)

Add SciFact benchmark evaluation script (Layer 6: measures extraction quality against ground truth)

Add SPECTER2 embedding-based deduplication (replaces Jaccard word overlap)

Add DEFERRED_FEATURES.md — registry of significant-effort and aspirational features with prerequisites and revisit triggers

Append 16 new improvement items from prior art analysis to FUTURE_IMPROVEMENTS.md (total now 94)

Append Appendix B (Prior Art Integration) to SYSTEM_DESIGN.md — maps 15 systems to our 7-layer architecture

Append Prior Art & Inspirations section to README landing page — links to new PRIOR_ART_ANALYSIS.md and SYSTEM_INSPIRATIONS.md

Append Prior Art & Inspirations section to README landing page

Add SYSTEM_INSPIRATIONS.md — concrete adoption plan: what to steal, what to adapt, what to learn from

Add PRIOR_ART_ANALYSIS.md — comprehensive landscape of similar systems with high-school-level explanations

Upload README.md with huggingface_hub

Upload README.md with huggingface_hub

Upload README.md with huggingface_hub

Upload SYSTEM_DESIGN.md with huggingface_hub

Upload SYSTEM_DESIGN.md with huggingface_hub

Upload SYSTEM_DESIGN.md

Integrate BDH into requirements — interpretable verifier, Hebbian memory, domain composability, sparsity monitoring

Integrate BDH (Baby Dragon Hatchling) into requirements — interpretable verifier, Hebbian memory, domain composability, sparsity monitoring

Add REQUIREMENTS_FROM_SOURCES.md — requirements grounded in Emergence Transformer paper + Awesome Open Source AI list, highschool-readable

Rewrite model card: transformer architecture diagram on front page, QLoRA explained, honest status table, highschool-readable throughout

FUTURE_IMPROVEMENTS.md v2.0 Final — 78 blindspots, 14 categories, highschool-readable, compiled from code review + audit + architecture review

Add FUTURE_IMPROVEMENTS.md v1.0 — 47 blindspots across 12 categories (highschool-readable)

Add Snapdragon ARM deployment guide

Add phd_research_os_v2/layer7/obsidian.py

Add phd_research_os_v2/layer6/evaluator.py

Add phd_research_os_v2/layer3/canonicalizer.py

Update README to v2.0 — 7-layer architecture, 143 tests, 87 blindspots

v2.0: tests/test_v2_integration.py

v2.0: phd_research_os_v2/layer5/scorer.py

v2.0: phd_research_os_v2/layer5/__init__.py

v2.0: phd_research_os_v2/layer4/graph.py

v2.0: phd_research_os_v2/layer4/__init__.py

v2.0: phd_research_os_v2/layer2/extractor.py

v2.0: phd_research_os_v2/layer2/__init__.py

v2.0: phd_research_os_v2/layer0/parser.py

v2.0: phd_research_os_v2/layer0/__init__.py

v2.0: phd_research_os_v2/core/database.py

v2.0: phd_research_os_v2/core/__init__.py

v2.0: phd_research_os_v2/app.py

v2.0: phd_research_os_v2/__main__.py

v2.0: phd_research_os_v2/__init__.py

Add complete system design v2.0 — 7-layer architecture addressing all 87 blindspots

Add complete blindspot audit — 87 findings across 4 epochs, 8 layers

Add complete blindspot audit (87 findings across 8 epochs)