Claude
|
275a3485c5
|
docs: Phase 0 findings — environment validated, fleet status updated
- Contributing-From-Linux: ROCm 7.2 actual, iGPU crash trap documented,
llama-server path + baseline benchmarks, HSA override not needed
- Fleet-Delegation-Pattern: Charon role updated (orchestrator not just build),
fleet table now shows go-rocm/p2p/rag as separate dedicated agents
- Inference-Architecture: Linux model paths, GGUF inventory with sizes
- Platform-Matrix: Actual hardware specs, baseline perf, iGPU gotcha
Charon, 19 Feb 2026
Co-Authored-By: Virgil <virgil@lethean.io>
|
2026-02-19 20:05:22 +00:00 |
|
Snider
|
774bfd3e4b
|
docs: add fleet delegation pattern and cross-platform knowledge base
New pages:
- Fleet-Delegation-Pattern: CLAUDE.md/TODO.md/FINDINGS.md contract, roles, onboarding
- Platform-Matrix: macOS vs Linux package assignments, GPU comparison
- Inference-Architecture: go-inference → go-mlx/go-rocm → go-ml layer diagram
- Contributing-From-Linux: Charon/Linux agent guide (ROCm, llama.cpp, assigned repos)
Updated Home to link fleet ops alongside package docs.
Co-Authored-By: Virgil <virgil@lethean.io>
|
2026-02-19 19:45:17 +00:00 |
|