rob thijssen
dc048ffcc9
All checks were successful
CI / CUDA type-check (push) Successful in 32s
build-prerelease / Resolve version stamps (push) Successful in 32s
CI / Format (push) Successful in 33s
CI / Clippy (push) Successful in 2m36s
build-prerelease / Build cortex binary (push) Successful in 4m48s
build-prerelease / Build neuron-blackwell (push) Successful in 5m59s
CI / Test (push) Successful in 6m35s
build-prerelease / Build neuron-ampere (push) Successful in 7m51s
CI / Build cortex SRPM (push) Has been skipped
CI / Publish cortex to COPR (push) Has been skipped
CI / Build neuron SRPM (push) Has been skipped
CI / Publish neuron to COPR (push) Has been skipped
CI / Bump version in source (push) Has been skipped
build-prerelease / Package cortex RPM (push) Successful in 1m21s
build-prerelease / Build neuron-ada (push) Successful in 5m13s
build-prerelease / Package helexa-neuron-ada RPM (push) Successful in 3m0s
build-prerelease / Package helexa-neuron-ampere RPM (push) Successful in 3m5s
build-prerelease / Package helexa-neuron-blackwell RPM (push) Successful in 3m49s
build-prerelease / Publish to rpm.lair.cafe (unstable) (push) Successful in 1m6s
Two fixes to the spatial handling of images, validated against the HF transformers 4.57.1 qwen3_vl reference on beast. **Vision tower (the real cause of poor spatial vision).** The Stage-A tower encoded position two ways wrong, so the model saw image *content* but not *layout* (a row of 5 people read as "a line of 23", sky inverted), regardless of the LM-side rope: - Learned pos-embed was a naive sequential lookup of the first `n_patches` rows of the 48×48 (`num_position_embeddings=2304`) grid — wrong stride for a 28×28 patch grid. Now bilinearly interpolates the grid to `gh×gw` (port of HF `fast_pos_embed_interpolate`), row-major. - The 2D vision rotary was absent entirely. Added `VisionRotaryEmbedding` (θ=10000, dim=head_dim/2) applying per-patch `(row, col)` rotary to q/k in every ViT block via rope_slow, matching HF `apply_rotary_pos_emb_vision`. Both default on; `NEURON_VISION_LEGACY_POS=1` / `NEURON_VISION_LEGACY_ROPE=1` revert each for A/B (no rebuild). New unit tests: interpolation reduces to the sequential lookup at the native grid; rotary row/col structure. **M-RoPE default on.** The interleaved M-RoPE matches HF apply_interleaved_mrope / get_rope_index exactly and A/B'd strictly ≥ plain. `NEURON_MROPE` is now a kill switch (`=0` for plain), not opt-in — defaults should encode the model's trained behaviour, not freeze the broken state. Vision tower is plain candle (CPU-testable): built, clippy-clean, full workspace tests green locally. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>