-
Notifications
You must be signed in to change notification settings - Fork 0
Pull requests: FluffyAIcode/Kakeya-LLM-Inference-engine
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
feat(distributed): remote DFlash+f_θ proposer (F3 data plane) — gemma-4 verifier on host A, DFlash+f_θ on host B
#158
opened Jun 19, 2026 by
FluffyAIcode
Owner
•
Draft
[CANCELLED] eval(omlx): oMLX parallel-inference evaluation (abandoned)
needs-mac-m4
#151
opened Jun 18, 2026 by
FluffyAIcode
Owner
•
Draft
CUDA-parity rollback for the all-MLX fused loop (keep accepted K/V, trim only rejected) — +33% on code, ~AR parity
#115
opened Jun 13, 2026 by
FluffyAIcode
Owner
•
Draft
ADR 0013 — Distributed inference topology: what AR sequentiality allows
#114
opened Jun 13, 2026 by
FluffyAIcode
Owner
•
Draft
ADR 0012 — Proposer/verifier value proposition (bounded-memory + recall; platform-forked throughput)
#113
opened Jun 13, 2026 by
FluffyAIcode
Owner
•
Draft
Step-2 rescue: all-MLX DFlash drafter — parity-proven, 17× over the hybrid fused path (0.476× AR)
#112
opened Jun 12, 2026 by
FluffyAIcode
Owner
•
Draft
Mac bridge M1: cloud-agent access to the self-hosted kakeya-mac-m4 (git-bus) + distributed-inference integration evaluation
needs-mac-m4
#111
opened Jun 12, 2026 by
FluffyAIcode
Owner
•
Draft
MLX native restored-cache primitive — systemic fix for the Mac throughput collapse
#110
opened Jun 11, 2026 by
FluffyAIcode
Owner
•
Draft
MLX port of #107: incremental decode (Step 1) + fused DFlash spec-decode engine (Step 2)
needs-mac-m4
#109
opened Jun 11, 2026 by
FluffyAIcode
Owner
•
Draft
Experiment with proposer KV full-attn restoration on Mac
#108
opened Jun 11, 2026 by
FluffyAIcode
Owner
•
Draft
K3 Mac MLX integration: cross-model verifier + integrated NIAH eval (parallel to vast f_θ training)
#104
opened Jun 10, 2026 by
FluffyAIcode
Owner
•
Draft
ADR §11.7.0: K3 model identity LOCKED to Gemma 4 family + correct §11.15.2.1 caveat 2
#92
opened Jun 9, 2026 by
FluffyAIcode
Owner
•
Draft
PR-ADR-§11.15.2: K3 Block A vast PASS + two architectural caveats + Block B prerequisite update + §11.15.14 lesson
#91
opened Jun 9, 2026 by
FluffyAIcode
Owner
•
Draft
PR-K2.A.2: stateful caching implementation — verifier per-step O(1) in T (closes §11.8 throughput gate c)
#90
opened Jun 9, 2026 by
FluffyAIcode
Owner
•
Draft
PR-K3-Mac-pivot: download PLE-safe community 4-bit MLX variant (resolves mlx-lm 0.31.3 self-quantize crash)
#89
opened Jun 9, 2026 by
FluffyAIcode
Owner
•
Draft
PR-R1b (research): fix ADR 0011 toy prototype Bugs A.2/B/C/D + Linux unit tests
#64
opened Jun 6, 2026 by
FluffyAIcode
Owner
•
Draft
PR-R1 (research): ADR 0011 + cross-attention toy prototype
#63
opened Jun 6, 2026 by
FluffyAIcode
Owner
•
Draft
7 tasks
ProTip!
Adding no:label will show everything without a label.