b3287c926f
Strata 0.2.14 runs the prompt on its own kernels and vendors llama.cpp (third_party/strata-llama), so the separate llama.cpp fetch for the vision encoder goes. The build keeps upstream's halo flags: MMQ stays out (STRATA_MMQ defaults to OFF), 47 of 48 UD-Q5_K_XL layers run on the fused kernels, and the ROCm 10.1 hipBLASLt table still ships. UD-Q5_K_XL on ms-s1-max, 16K cold prompts: 715 -> ~810 tok/s on the defaults, ~1180 tok/s with the gfx1151 prompt set (unchanged). Bead: nix-overlay-ydg
Beads - AI-Native Issue Tracking
This repository uses beads_rust (br) for issue tracking. Issues live next to the code in .beads/, so no web UI is needed.
Where the state lives
.beads/issues.jsonl— the versioned source of truth. It is committed to git, so issue state travels withgit push..beads/dolt— the local SQLite database used bybr(runtime only, not versioned)..beads/config.yaml— project settings.
br never runs git commands. After br sync --flush-only you must run git add .beads/ && git commit yourself.
Essential commands
br robot-docs guide # session context and command reference (run first)
br ready # unblocked, non-deferred issues
br list # all issues
br show <issue-id> # issue detail
br create "Title" # create an issue
br update <id> --claim # claim work before starting
br close <id> # close an issue
br sync --flush-only # export the database to issues.jsonl
Learn more
- Repository: https://github.com/Dicklesworthstone/beads_rust
- Agent guide:
br robot-docs guide