Skip to content

Actions: johnnynunez/fkl_python

Actions

All workflows

Actions

Loading...
Loading

Showing runs from all workflows
23 workflow runs
23 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
Merge pull request #7 from johnnynunez/main wheels #23: Commit 2ad8c57 pushed by johnnynunez
1m 29s main
main
1m 29s
Merge pull request #2 from johnnynunez/fix/divergent-hf-selector-offb… wheels #22: Commit 3e58de4 pushed by johnnynunez
1m 30s main
main
1m 30s
DivergentHF: fix off-by-one in sequence selector (0-based kernel) wheels #21: Pull request #2 opened by johnnynunez
1m 26s fix/divergent-hf-selector-offbyone
fix/divergent-hf-selector-offbyone
1m 26s
Merge pull request #1 from johnnynunez/refactor/generic-io-ops-and-cp… wheels #20: Commit 9cbbccb pushed by johnnynunez
1m 19s main
main
1m 19s
Simplify fkl-python: generic IO IOps + let C++ fuse_back fuse (Oscar review) wheels #19: Pull request #1 opened by johnnynunez
1m 23s refactor/generic-io-ops-and-cpp-fusion
refactor/generic-io-ops-and-cpp-fusion
1m 23s
flex: import numpy lazily — keep 'import fkl' zero-dep (fixes CI) wheels #18: Commit 0ac1c45 pushed by johnnynunez
1m 14s main
main
1m 14s
flex attention + block-sparse attention in fkl-python wheels #17: Commit 4d4717c pushed by johnnynunez
37s main
main
37s
flex attention + block-sparse attention in fkl-python wheels #16: Commit e8d2bf9 pushed by johnnynunez
38s main
main
38s
fp8 e4m3 KV cache: compress_kv(fmt='fp8') + kv_layout='fp8' wheels #15: Commit 43721af pushed by johnnynunez
1m 24s main
main
1m 24s
as_device_view: torch fast path — call overhead 28us -> hidden under … wheels #14: Commit e092bf7 pushed by johnnynunez
1m 36s main
main
1m 36s
flash_attention: prologue fusion (Q/K/V as Read IOps) + mma tensor-co… wheels #13: Commit ebb84bd pushed by johnnynunez
1m 22s main
main
1m 22s
fkl.flash_attention: FA-2 forward with fused epilogue + compressed in… wheels #12: Commit c50e9d3 pushed by johnnynunez
1m 32s main
main
1m 32s
Plug-and-play wheels: vendored FKL headers + GPU arch/CUDA auto-detec… wheels #11: Commit 3ace70c pushed by johnnynunez
1m 33s main
main
1m 33s
ci: publish wheels to PyPI on version tags via Trusted Publishing wheels #10: Commit 784163c pushed by johnnynunez
1m 19s main
main
1m 19s
spikes: add CircularTensor direct-use OOB probe (issue #250 evidence) wheels #9: Commit 6ec6ef7 pushed by johnnynunez
4m 17s main
main
4m 17s
ThreadFusion docs: make the opt-in semantics explicit (Oscar's guidance) wheels #8: Commit 9d62a35 pushed by johnnynunez
1m 36s main
main
1m 36s
CPU backend, ThreadFusion, multi-GPU device routing, torch integratio… wheels #7: Commit 2e67780 pushed by johnnynunez
1m 32s main
main
1m 32s
CircularTensor: fix size-baked-in-cache bug; add race audit + coheren… wheels #6: Commit 6ea3155 pushed by johnnynunez
1m 23s main
main
1m 23s
Add real end-to-end example: YOLOv8 detection with fused GPU preproce… wheels #5: Commit 955e4d8 pushed by johnnynunez
1m 6s main
main
1m 6s
Target LTS-C++17: drop the BGR2GRAY decomposition workaround wheels #4: Commit a08aa18 pushed by johnnynunez
1m 10s main
main
1m 10s
ci: pin cuda-toolkit action to v0.2.35 and use CUDA 13.1.1 (13.0.2 un… wheels #3: Commit 779880e pushed by johnnynunez
1m 21s v0.6.0
v0.6.0
1m 21s
ci: pin cuda-toolkit action to v0.2.35 and use CUDA 13.1.1 (13.0.2 un… wheels #2: Commit 779880e pushed by johnnynunez
1m 9s main
main
1m 9s
fkl-python 0.6.0: zero-overhead Python front-end for the Fused Kernel… wheels #1: Commit e994c24 pushed by johnnynunez
1m 23s main
main
1m 23s