rob thijssen
75cd088b61
All checks were successful
CI / CUDA type-check (push) Successful in 31s
build-prerelease / Resolve version stamps (push) Successful in 29s
CI / Format (push) Successful in 30s
CI / Clippy (push) Successful in 2m32s
build-prerelease / Build neuron-blackwell (push) Successful in 6m5s
CI / Test (push) Successful in 5m49s
CI / Build cortex SRPM (push) Has been skipped
CI / Build neuron SRPM (push) Has been skipped
CI / Publish cortex to COPR (push) Has been skipped
CI / Publish neuron to COPR (push) Has been skipped
CI / Bump version in source (push) Has been skipped
build-prerelease / Build neuron-ampere (push) Successful in 8m11s
build-prerelease / Build neuron-ada (push) Successful in 5m40s
build-prerelease / Package helexa-neuron-ada RPM (push) Successful in 3m4s
build-prerelease / Package helexa-neuron-ampere RPM (push) Successful in 3m2s
build-prerelease / Package helexa-neuron-blackwell RPM (push) Successful in 3m57s
build-prerelease / Build cortex binary (push) Successful in 4m21s
build-prerelease / Package cortex RPM (push) Successful in 1m25s
build-prerelease / Publish to rpm.lair.cafe (unstable) (push) Successful in 1m16s
Beast testing surfaced a real regression in the dynamic-resolution default: a tall 808×1600 image resized (within the 1024² max_pixels) to a 90×44 patch grid = 3960 patches, exceeding the vision tower's hard `num_position_embeddings = 2304` pos-embed budget. The per-rank `patch count 3960 exceeds pos_embed budget 2304` error fired mid-TP- forward and poisoned the device context, bricking the model until reload. Hard-cap `max_pixels` to `2304 × 16² = 589_824` px (≤ 2304 patches → ≤ 576 LM tokens), clamping even the operator env override. `smart_resize` floors the pixel count under the cap, so no resized image can ever exceed the budget — the tower check never fires, no poison. The pos-embed grid (48×48) is the resolution Qwen3.6 was trained at, so the cap is principled, not just defensive. Still ~3× the old fixed 196 tokens, and the book-cover OCR test (1176 patches) already reads full title+subtitle. Test: a huge/tall/wide/extreme image battery stays within the 2304 patch budget. (Per-rank-error poison robustness itself remains issue #17.) Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>