Releases · flashinfer-ai/flashinfer-bench

What's Changed

docs: fix the blog link by @zhyncs in #93
Alias flashinfer_bench as fib in README by @Ubospica in #94
fix: fix linting by @Ubospica in #95
fix: fix linting by @Ubospica in #96
Add cutlass license by @tqchen in #98
chore: remove vercel microfrontends dependency by @xslingcn in #99
fix: fix ci tests by @Seven-Streams in #97
ci: drop python 3.9 by @xslingcn in #108
feat: restore leaderboard to main by @xslingcn in #107
docs: python api reference by @xslingcn in #109
docs: enhanced API docs with intro and links by @YiyanZhai in #110
fix: fix the colorFor function's hash-based color assignment for web by @YiyanZhai in #113
fix: smooth chart curves & longer client names in UI by @YiyanZhai in #114
[tvm-ffi] TVMFFIBuilder by @Ubospica in #111
refactor: Enhance TvmFfiBuilder by @Ubospica in #117
refactor: rename TvmFfiBuilder -> TVMFFIBuilder by @Ubospica in #118
refactor: update sampling evaluation logic by @zanderjiang in #104
refactor: Builder System by @Ubospica in #120
ci: Use uv with cache in unit tests by @Ubospica in #122
test: update tests for builder by @Ubospica in #123
docs: Enhance docs with md and add docs for compile by @Ubospica in #124
update kernel generator by @zanderjiang in #101
feat: Destination-passing Style by @Ubospica in #125
refactor: rename traceset -> trace_set and avoid singie character variables by @Ubospica in #129
Add profile_baseline option to BenchmarkConfig by @zhang677 in #130
feat: Per-kernel Apply Config by @Ubospica in #131
refactor: Add FlashInfer Trace (definitions and workloads) to flashinfer-bench by @YiyanZhai in #138
Fix React Server Components CVE vulnerabilities by @vercel[bot] in #128
Agent FFI integration by @zanderjiang in #119
Docs: Add docstring for modules by @Ubospica in #132
Add Claude Code GitHub Workflow by @yzh119 in #137
refactor: Provide TracingConfigRegistry and Refactor TracingRuntime by @Ubospica in #133
Add Claude skills files for new model support by @yyihuang in #136
Add kernel: deepseek sparse attention by @yyihuang in #146
refactor: Enhance ApplyRuntime by @Ubospica in #135
[refactor] apply() Overhead Reduction by @YiyanZhai in #121
feat: adapter for torch.nn.functional.linear by @YiyanZhai in #154
Add kernel: trtllm fp8 block scale moe by @yyihuang in #152
Support NCU tool by @Ubospica in #139
feat: cache solution hash by @xslingcn in #153
Add def: add attention variant with page size 64 by @yyihuang in #145
Add kernel: deepseek sparse attn page size 64 by @yyihuang in #147
fix: adapter can take page_size 64 by @yyihuang in #157
Add kernel: gated delta net by @yyihuang in #156
claude: update skills by @yyihuang in #158
claude: update skills on cloning repos by @yyihuang in #159
claude: fix clone repos skill by @yyihuang in #161
Support Agent Tools by @zanderjiang in #140
update dsa and gdn definitions, op schema and unittests by @yzh119 in #164
fix: gdn test update && decode test on h100 by @yyihuang in #165
docs: Enhance agent and API docs by @Ubospica in #162
docs: Add github link to main page by @Ubospica in #167
docs: merge configurations for mintlify by @xslingcn in #171
Fix README logo and support dark and light mode by @Ubospica in #172
Update Definition: Deepseek Sparse Attn page size 64 by @zanderjiang in #173
Update README Links by @zanderjiang in #176
fix: Error in Sphinx Docs Build by @Ubospica in #181
fix: Benchmarking on latest flashinfer-trace by @Ubospica in #179
Add --required-matched-ratio CLI parameter for configurable correctness thresholds by @xiefan46 in #178
gdn: update number of heads on tp=4 & tp=2 by @yyihuang in #186
feat: TileLang Support by @zanderjiang in #134
Update solution folder structure by @Ubospica in #185
web: Update viewer by @Ubospica in #189
claude: update skills on hugging face reference by @yyihuang in #174
feat: web UI rankings tab by @YiyanZhai in #191
fix: Versioning and release by @Ubospica in #192

No results found