Skip to content

Actions: hchengit/LeanInfer

Actions

All workflows

Actions

Loading...
Loading

Showing runs from all workflows
26 workflow runs
26 workflow runs

Filter by Workflow

⌥ + click/return to exclude

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
docs: add RESULTS.md — complete benchmark findings + ASSESSMENT.md fi… CI #26: Commit 9268a16 pushed by hchengit
5m 7s main
main
5m 7s
fix(cuda): add forward declaration for fused gate dispatch function CI #25: Commit 4e7fa45 pushed by hchengit
5m 10s main
main
5m 10s
fix(cuda): inline fused gate kernel in ggml-cuda.cu instead of #include CI #24: Commit 74feeb2 pushed by hchengit
5m 21s main
main
5m 21s
fix(cuda): rewrite ggml-cuda.cu patching — append dispatch function, … CI #23: Commit 671851b pushed by hchengit
5m 20s main
main
5m 20s
fix(cuda): bump GGML_OP_COUNT assertion 101→102 for new fused gate op CI #22: Commit 1b810a4 pushed by hchengit
5m 12s main
main
5m 12s
feat(cuda): fused RMSNorm+SiLU-gate graph integration via patch script CI #21: Commit c550eee pushed by hchengit
5m 23s main
main
5m 23s
cuda: fused RMSNorm+SiLU-gate kernel + CUDA graph analysis CI #20: Commit 5765acd pushed by hchengit
5m 17s main
main
5m 17s
docs: revised DeltaNet fusion analysis — original plan won't work, ne… CI #19: Commit 49d29c2 pushed by hchengit
5m 15s main
main
5m 15s
fix(cuda): broadcast RMS scale to all warps via shared memory CI #18: Commit 95f7483 pushed by hchengit
5m 10s main
main
5m 10s
fix(cuda): remove deprecated memoryClockRate (CUDA 13.0 compat) CI #17: Commit 5b1afb2 pushed by hchengit
5m 13s main
main
5m 13s
docs: add TODO for next Vast.ai session — kernel validation + graph i… CI #16: Commit 35e804f pushed by hchengit
5m 18s main
main
5m 18s
cuda: standalone kernel benchmark + revised CUDA optimization strategy CI #15: Commit 4082bca pushed by hchengit
5m 31s main
main
5m 31s
CI CI #14: by hchengit
5m 21s main
main
5m 21s
fix(cuda): define WARP_SIZE in common header CI #13: Commit 0f004e5 pushed by hchengit
5m 22s main
main
5m 22s
docs: update stale tracker rows — GitHub + CI now marked done CI #12: Commit 91192f8 pushed by hchengit
5m 15s main
main
5m 15s
fix(cuda): build CUDA kernels as separate static library to fix linke… CI #11: Commit 522ce31 pushed by hchengit
5m 11s main
main
5m 11s
docs: RTX 4090 vs M2 benchmark results — DeltaNet hybrid favors unifi… CI #10: Commit 9a1afc4 pushed by hchengit
5m 12s main
main
5m 12s
docs: add RTX 5060 Ti ($0.07/hr) as recommended cloud GPU option CI #9: Commit 98fb61b pushed by hchengit
10m 3s main
main
10m 3s
docs: add CUDA cloud GPU testing guide (Vast.ai/RunPod/Lambda) CI #8: Commit aa025c9 pushed by hchengit
9m 58s main
main
9m 58s
LeanInfer CUDA fused kernels: FFN + DeltaNet (pending cloud GPU compile) CI #7: Commit 1dbcd08 pushed by hchengit
9m 42s main
main
9m 42s
docs: add TODO for pushing M2 eval callback changes to upstream fork CI #6: Commit 3259985 pushed by hchengit
9m 46s main
main
9m 46s
fix(metal): replace weak symbol with compile-time stub for eval callback CI #5: Commit b014af8 pushed by hchengit
9m 52s main
main
9m 52s
fix(metal): compile on CI — MTLMathModeFast guard + weak eval_cb symbol CI #4: Commit a1c9fe0 pushed by hchengit
9m 48s main
main
9m 48s
ci: re-trigger after making Lean_llama.cpp public CI #3: Commit 4189a80 pushed by hchengit
9m 43s main
main
9m 43s
fix(ci): make setup_upstream.sh executable + use bash explicitly CI #2: Commit 0a2c010 pushed by hchengit
11s main
main
11s