[bong-water-water-bong]

Adds one sentence to the README status block:

• Q2_K, IQ2 and IQ3_XXS GGUFs now run on HRX instead of the CPU (Pin llama.cpp 764a256: IQ3_XXS and IQ2_S weights on HRX0 (llama.cpp #51) #255);
• --mtp on Qwen3.8-27B runs NaN-free on HRX (Pin llama.cpp cde002d: fix NaN after MTP rollback on Qwen3.5/3.8 (llama.cpp #52) #257).

The wiki Home status got the same rows, plus #262 and the census-watch Action.

🤖 Generated with Claude Code

[context7]

Docs7 for 1bit-monster/engine

Result	Status	Action
Deployment	➖ Not used	—
Content review	✅ Passed. No problems found.	View findings

Commit abf4a54

[github-actions]

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

🎫 Ticket compliance analysis 🔶 255 - Partially compliant Compliant requirements: Move llama.cpp from d5048ad to 764a256 Enable IQ3_XXS and IQ2_S weights on HRX0 Update registry/architectures.json Update docs/hrx.md with performance numbers and caveats Non-compliant requirements: None Requires further human verification: None 257 - Partially compliant Compliant requirements: Move llama.cpp from 764a256 to cde002d Fix NaN after MTP rollback on Qwen3.5/3.8 Update registry/architectures.json Update docs/hrx.md with the fix details Non-compliant requirements: None Requires further human verification: None 262 - Partially compliant Compliant requirements: Pin ROCmFPX fc664ba Implement Hadamard rotation per tensor Update documentation to reflect per-tensor rotation Remove process-wide GGML_Q4_0_HADAMARD setting Non-compliant requirements: None Requires further human verification: None

⏱️ Estimated effort to review: 2 🔵🔵⚪⚪⚪

🧪 No relevant tests

🔒 No security concerns identified

⚡ Recommended focus areas for review Documentation Update The README update correctly reflects the new capabilities for sub-4-bit GGUFs on HRX and the fix for NaN issues with MTP rollback. The changes are consistent with the described improvements in llama.cpp and the updated documentation. > ([docs/hrx.md](docs/hrx.md); its decode-split race, #123/#140, is fixed and the kernel is on by default; Q2_K, IQ2 and IQ3_XXS GGUFs run on HRX instead of the CPU, and `--mtp` on Qwen3.8-27B runs NaN-free since #257), Vulkan from upstream llama.cpp's latest release ([docs/vulkan.md](docs/vulkan.md)), the NPU engine on full ELFs with the