Skip to content

ADRs tagged ai

Auto-generated by scripts/docs/generate-adr-by-tag.sh. Edit ADR Tags: lines to update.

156 ADR(s) carry this tag.

ID Title
ADR-0020 Tiny-AI scope covers all four capabilities
ADR-0021 Training stack is PyTorch + Lightning with ONNX export
ADR-0022 Inference runtime is ONNX Runtime via execution providers
ADR-0023 Tiny-AI user surfaces span CLI, C API, ffmpeg, and training
ADR-0036 Tiny-AI Wave 1 scope expanded beyond D20–D23
ADR-0039 Pull forward runtime op-allowlist walk and model registry
ADR-0040 Extend DNN session API to multi-input/multi-output with named bindings
ADR-0041 Ship LPIPS-SqueezeNet FR extractor with inverse-ImageNet in graph
ADR-0042 Tiny-AI PRs must ship human-readable docs in the same PR
ADR-0102 DNN execution-provider selection is ordered + graceful, fp16_io does a host-side cast
ADR-0107 Tiny-AI Wave 1 scope expanded beyond ADR-0020 through ADR-0023
ADR-0109 Nightly bisect-model-quality runs against a synthetic placeholder cache
ADR-0118 FFmpeg patches ship as ordered series.txt, not a single carry
ADR-0120 DNN-enabled matrix legs across compilers + macOS
ADR-0129 Tiny-AI post-training int8 quantisation — static + dynamic + QAT per model
ADR-0203 Tiny-AI training prep — implementation decisions
ADR-0207 Tiny-AI Quantization-Aware Training (QAT) — design
ADR-0211 Tiny-model registry schema + Sigstore --tiny-model-verify
ADR-0215 FastDVDnet temporal pre-filter — 5-frame window, placeholder weights
ADR-0218 MobileSal saliency feature extractor (T6-2a)
ADR-0222 vmaf-perShot per-shot CRF predictor sidecar
ADR-0223 0223-transnet-v2-shot-detector.md
ADR-0234 GPU-generation-aware ULP calibration head
ADR-0235 Codec-aware FR regressor (fr_regressor_v2)
ADR-0236 0236-dists-extractor.md
ADR-0237 Quality-aware encode automation surface (vmaf-tune)
ADR-0242 Tiny-AI training on the original Netflix VMAF training corpus
ADR-0244 vmaf_tiny_v2 — canonical-6 + StandardScaler tiny VMAF MLP
ADR-0247 vmaf-roi sidecar binary for per-CTU QP offsets
ADR-0250 Tiny-AI extractor template — shared scaffolding header
ADR-0255 FastDVDnet temporal pre-filter — real upstream weights via luma adapter (T6-7b)
ADR-0257 MobileSal real-weights swap deferred (T6-2a-followup blocker)
ADR-0261 TransNet V2 shot-boundary detector — real upstream weights via NTCHW adapter (T6-3a-followup)
ADR-0262 bisect-model-quality cache check uses logical comparison for parquet
ADR-0265 U-2-Net u2netp saliency replacement blocked on weights distribution + op allowlist
ADR-0272 fr_regressor_v2 codec-aware scaffold (Phase B prereq)
ADR-0276 vmaf-tune fast — proxy-based recommend (Phase A.5)
ADR-0283 vmaf-tune Apple VideoToolbox codec adapters
ADR-0285 vmaf-tune libvvenc adapter — VVC / H.266 with optional NN-VC tools
ADR-0286 Fork-trained saliency student saliency_student_v1 on DUTS-TR
ADR-0287 vmaf_tiny_v5 — corpus expansion (4-corpus + YouTube UGC vp9 subset)
ADR-0293 vmaf-tune saliency-aware ROI tuning (Bucket #2)
ADR-0296 Region-of-interest VMAF scoring (vmaf-roi-score) — saliency-weighted scaffold
ADR-0299 GPU scoring backend for vmaf-tune (--score-backend)
ADR-0303 fr_regressor_v2 ensemble — production flip trainer + CI gate
ADR-0304 vmaf-tune fast — production wiring (Optuna TPE + v2 proxy + GPU verify)
ADR-0305 Encoder knob-space Pareto-frontier analysis stratified per (source, codec, rc_mode)
ADR-0308 Encoder knob-sweep recipe-regression revision policy
ADR-0309 fr_regressor_v2 ensemble — real-corpus retrain harness + flip workflow
ADR-0310 BVI-DVC corpus ingestion for fr_regressor_v2
ADR-0318 fr_regressor_v2 ensemble retrain harness — wrapper-trainer interface fix + Phase A pre-step doc
ADR-0319 fr_regressor_v2 ensemble LOSO trainer — real loader + per-fold training
ADR-0320 fr_regressor_v2 ensemble seeds — production flip (smoke → false)
ADR-0321 fr_regressor_v2_ensemble_v1 — full production flip (real ONNX + sidecars)
ADR-0323 fr_regressor_v3 — train + register on ENCODER_VOCAB v3 (16-slot)
ADR-0324 Ensemble training kit — portable Phase-A + LOSO retrain bundle
ADR-0325 KonViD-150k corpus ingestion
ADR-0335 Hardware-capability priors for the FR-regressor corpus
ADR-0336 KonViD MOS head v1 (ADR-0325 Phase 3)
ADR-0339 av1_videotoolbox placeholder adapter + upstream watcher
ADR-0340 Multi-corpus aggregation for the FR-regressor / predictor v2 trainer
ADR-0346 FR-features-from-NR-corpus adapter pattern
ADR-0349 fr_regressor_v3 namespace — reserve _v3plus_features for the next feature-set bump
ADR-0364 saliency_student_v2 — Resize-decoder ablation on the v1 recipe
ADR-0365 Wire the CoreML execution provider into tiny-AI ORT dispatch
ADR-0366 vmaf-tune corpus schema v3 — canonical-6 per-feature aggregates
ADR-0367 LSVQ corpus ingestion for nr_metric_v1
ADR-0368 External-competitor benchmark harness — wrapper-only architecture
ADR-0369 Waterloo IVC 4K-VQA corpus ingestion for nr_metric_v1
ADR-0383 K150K corpus scoring driver — parallel CPU worker redesign
ADR-0388 Ingest BVI-CC as the second tiny-AI training corpus
ADR-0389 vmaf_tiny_v3 — wider/deeper mlp_medium tiny VMAF MLP
ADR-0390 vmaf_tiny_v4 — mlp_large arch (opt-in only; arch ladder stops here)
ADR-0392 vmaf-tune Phase D — per-shot CRF tuning
ADR-0393 fr_regressor_v2 probabilistic head — deep-ensemble + conformal scaffold
ADR-0394 Local sidecar training — on-host bias-correction model
ADR-0395 predictor stub-models policy
ADR-0396 Video-temporal saliency extension to saliency_student_v1
ADR-0401 libvmaf WebAssembly target — phased EXPERIMENT then GO
ADR-0405 Wire OpenVINO NPU execution provider into the tiny-AI dispatch layer
ADR-0412 Fork-local release-artefact mirror scaffold for u2netp.pth (Apache-2.0)
ADR-0413 YouTube UGC corpus ingestion for nr_metric_v1
ADR-0417 Tiny-AI Netflix corpus training scaffold — draft PR registration
ADR-0425 vmaf-roi-score saliency materialiser
ADR-0426 CHUG HDR corpus ingestion
ADR-0427 Materialise CHUG HDR Features With Reference-Aligned Pairs
ADR-0431 Split CUDA and CPU Feature Passes for FR-from-NR Extraction
ADR-0433 CHUG Content Splits And HDR Audit
ADR-0434 CHUG Parquet Metadata Enrichment
ADR-0444 Promote saliency_student_v2 to production default
ADR-0446 K150K/CHUG extractor passes HDR and HFR per-feature options
ADR-0447 Motion features under-report on HFR / 50p content
ADR-0455 KonViD-150k k150ka/k150kb split promotion into the MOS-head trainer
ADR-0457 model/tiny/*.onnx blobs ≥1MB live in GitHub Releases, not git
ADR-0459 vmaf-tune panel/display-aware recommendation workstream
ADR-0481 ADM p-norm Parameter Hardcoded at 3.0 — Deferral Decision
ADR-0482 Expand vmaf_pre FFmpeg filter device strings to match full VmafDnnDevice enum
ADR-0495 MCP server probe-driven bug-fix cluster (2026-05-17)
ADR-0510 CHUG re-extract VMAF-alignment fix — FR-corpus guard on the FR-from-NR extractor
ADR-0511 MCP backend probe, default allowlist, and vmaf-tune ladder --score-backend (2026-05-18)
ADR-0518 Tiny-model loader accepts external-data and feature-vector ONNX
ADR-0520 Wire vmaf --no-reference through to the scoring path
ADR-0522 --tiny-codec / --tiny-preset / --tiny-crf populate codec one-hot block
ADR-0524 Tiny-model loader accepts symbolic batch dim
ADR-0525 Extract run_cmd subprocess helper into aiutils
ADR-0527 Accept pre-extracted BVI-DVC YUVs via --bvi-dir
ADR-0546 Audit bundle — Vulkan motion dispatch wiring, saliency hard-fail, model-card placeholder
ADR-0547 VMAF__DIR env-var overrides for ai/scripts corpus paths + drop cli.py.bak
ADR-0550 Auto-resize input plane to NR tiny-model dims + --tiny-resize flag
ADR-0556 Python / MCP / AI silent-fallback audit fixes (2026-05-18)
ADR-0559 Feature Coverage Audit — Add speed_chroma + speed_temporal to Extraction Scripts (HDR-model prep)
ADR-0565 Continuous Feature-Mix Evaluation Pipeline (predictor-bench)
ADR-0612 Tiny-AI training on the original Netflix VMAF training corpus
ADR-0613 Dynamic Optimizer — Joint Shot-Boundary + CRF Co-Optimisation
ADR-0614 Per-Shot ABR Rendition Selection
ADR-0615 Fast NR Pre-Scoring for CRF Bisect Acceleration
ADR-0616 VMAF NEG Integration into vmaf-tune
ADR-0617 Cross-Shot Complexity Weighting and Title-Level Quality Constraints
ADR-0618 Content-Aware Classifier for Encoder Routing
ADR-0621 Scaffold Audit P3 — six cleanup items + state drift
ADR-0624 Fast NR Pre-Scoring Implementation (ADR-0615 impl)
ADR-0639 Scaffold-audit P1 — backend precheck, HIP picture, mobilesal bpc, DNN multi-output
ADR-0640 Tiny-AI training on the original Netflix VMAF training corpus (2026-05-20 scaffold iteration)
ADR-0642 AI refresh defaults use current fork full-feature extractors
ADR-0646 Route Attached DNN Multi-Output Tensors
ADR-0647 Refresh fr_regressor_v1 from the 2026-05-20 Netflix feature table
ADR-0648 CHUG HDR MOS Trainer Entry Point
ADR-0649 CHUG HDR Wide MOS Feature Schema
ADR-0650 Add a Signal-Mix Audit CLI
ADR-0651 Preserve CHUG HDR Metadata On Feature Rows
ADR-0652 Add CHUG Visual-Signal Primitives
ADR-0653 CHUG Display Profile Training
ADR-0654 Predictor Preserves Saliency Signals
ADR-0655 Saliency Feature Materializer
ADR-0656 External-bench wrappers emit registry competitor keys
ADR-0657 Second-Opinion Feature Materializer
ADR-0658 Project modernization audit
ADR-0661 AI run manifest provenance
ADR-0663 MOS Label Materializer
ADR-0668 AI Derived Table Provenance
ADR-0669 AI Corpus JSONL Provenance
ADR-0670 AI Legacy Corpus Extraction Manifests
ADR-0671 U2NetP Mirror Exporter
ADR-0672 Saliency Materializer Temporal Controls
ADR-0674 Second-Opinion Materializer Batch Manifest
ADR-0675 MOS Label Materializer Batch Manifest
ADR-0676 MOS Corpus Adapter Manifests
ADR-0677 AI Dataset Fetch Manifests
ADR-0678 Shared AI Run Manifest Helper
ADR-0680 Shared AI CLI Helper Pattern
ADR-0681 AI Script Bootstrap Helper
ADR-0682 Tiny-AI Netflix corpus training scaffold — 2026-05-22 prep scope
ADR-0685 Tiny-AI Netflix corpus training scaffold — 2026-05-27 prep scope
ADR-0687 CHUG HDR MOS head — held-out test partition validator
ADR-0690 VMAFX Binary and AI Tool Aliases
ADR-0841 Environment variable reference page and canonical naming