Issues

• [Epic] AiDotNet v1: close every flaw in the v1 readiness report

  ooples/AiDotNet#2397 · Filter by author ooples opened on Oct 28, 2026
  3

• A-7 Fidelity backlog: missing options, wav2vec2 family, DCNv3, segmentation OOMs

  ooples/AiDotNet#2396 · Filter by author ooples opened on Oct 28, 2026
  1

• A-6 Benchmarks: median of 9 with locked clocks, plus diffusion, LLM, ASR/TTS, detection, GNN and classical-ML rows

  ooples/AiDotNet#2395 · Filter by author ooples opened on Oct 28, 2026
  1

• A-5 #2091: one entry path per model family and real model-family coverage

  ooples/AiDotNet#2394 · Filter by author ooples opened on Oct 28, 2026
  1

• A-4 Docs and dependencies: facade XML-doc gate, evidence-backed README, dead-dependency removal

  ooples/AiDotNet#2393 · Filter by author ooples opened on Oct 28, 2026
  3

• A-3 API contract: full public-surface PublicAPI baseline and package validation

  ooples/AiDotNet#2392 · Filter by author ooples opened on Oct 28, 2026
  1

• A-2 Performance: MobileNetV2 off the eager tape, DenseNet, Transformer/ViT inference, small-model overhead, CNN mid-batch, CPU parallelism

  ooples/AiDotNet#2391 · Filter by author ooples opened on Oct 28, 2026
  1

• A-1 Correctness: Tensors bump, 5 census failures, every nightly green

  ooples/AiDotNet#2390 · Filter by author ooples opened on Oct 28, 2026
  1

• Docs Wiki gate red: 7 generated samples on 6 pages use an undeclared input variable (CS0103)

  ooples/AiDotNet#2389 · Filter by author ooples opened on Oct 28, 2026

• Cross-ISA Agreement nightly red for 30+ runs: 7 MoreData tests fail only with AVX2 off, plus 20-25 incomplete results

  ooples/AiDotNet#2388 · Filter by author ooples opened on Oct 28, 2026
  1

• Nightly Model Performance Census red since 2026-10-19: WGAN-GP, GOT-OCR2, EfficientConformer, CLAP and OpenSora fail their own contracts

  ooples/AiDotNet#2387 · Filter by author ooples opened on Oct 28, 2026

• CNN CPU inference is slower at batch 32 than at batch 128 (4-5x PyTorch at b32, 0.6-0.7x at b128)

  ooples/AiDotNet#2386 · Filter by author ooples opened on Oct 28, 2026
  2

• Small-model GPU inference pays a fixed 0.7-1.8 ms per Predict: MLP is 4.6-7.9x PyTorch at every batch size

  ooples/AiDotNet#2385 · Filter by author ooples opened on Oct 28, 2026

• Transformer and ViT GPU inference scale linearly with batch size: 85x PyTorch at b128 while training is faster than PyTorch

  ooples/AiDotNet#2384 · Filter by author ooples opened on Oct 28, 2026

• DenseNet-121 uses 5-6x PyTorch's memory, OOMs at inference b16 on a 6 GB GPU, and trains 10-12x slower on CUDA

  ooples/AiDotNet#2383 · Filter by author ooples opened on Oct 28, 2026

• MobileNetV2 trains on the eager tape by default (DisableFusedOptimizerStep = true): 9.9x PyTorch on CUDA, 1.4-1.8x on CPU

  ooples/AiDotNet#2382 · Filter by author ooples opened on Oct 28, 2026

• perf: ParaformerLarge full-scale eager BF16 8-bit Adam step time grows 91 s -> 150 s between steps 1 and 2

  ooples/AiDotNet#2322 · Filter by author ooples opened on Oct 13, 2026
  1

• Fifteen wav2vec2-family recognizers share a generic stack instead of their papers' networks

  ooples/AiDotNet#2321 · Filter by author ooples opened on Oct 13, 2026

• Plan: beat the best rival 2x on every model family at the foundation (CPU + GPU, training + inference)

  ooples/AiDotNet#2300 · Filter by author ooples opened on Oct 7, 2026
  2

• [Epic] Model-family invariants drive one entry path per model (plus: the coverage report is broken, and 345 models have no test class)

  v1-blocker
  ooples/AiDotNet#2091 · Filter by author ooples opened on Sep 5, 2026
  6

• Foundation-scale CNN training-step perf: non-conv overhead + VGG16 ModelFamily timeouts (#1684 follow-up)

  ooples/AiDotNet#1694 · Filter by author ooples opened on Jun 25, 2026
  2

• Training backward-pass memory/throughput optimizations (optimizer-in-backward, default checkpointing, FlashAttention backward, backward buffer-reuse)

  ooples/AiDotNet#1662 · Filter by author ooples opened on Jun 22, 2026
  2

• Inference/compilation: industry-standard gap audit + exceed-industry levers (value-aware replay verified, verify-then-trust, fused attention, paged KV-cache, ...)

  ooples/AiDotNet#1630 · Filter by author ooples opened on Jun 16, 2026
  4

• Foundation-scale model CI: eliminate all timeouts/OOM via stacked acceleration (tracker)

  ooples/AiDotNet#1622 · Filter by author ooples opened on Jun 15, 2026
  2

• Multi-thread training story: HOGWILD! + in-process DDP + striped-lock trainers (exceed PyTorch DDP)

  enhancement
  ooples/AiDotNet#1369 · Filter by author ooples opened on May 18, 2026
  2