11月12月1月2月3月4月5月6月7月8月9月10月
周一周三周五
一年内 256811 次贡献
更少更多
- frostbyte_neo 推送了仓库 github-featured/mintplex-labs--anything-llm 的 master 分支比较 6 提交 »3d6dff256c fix: keep thread and user on the query-mode refusal row (#6236)4db5d01397 Add Swedish frontend locale (#6227)37fabaf069 fix: stop $ sequences corrupting restored Telegram code blocks (#6235)02b1f3c1bc pass include status on sessionID chat getter (#6237)df42bf8d20 fix: keep workspace prompt and temperature when an embed sends no override (#6232)
- frostbyte_neo 推送了仓库 github-featured/9001--copyparty 的 hovudstraum 分支比较 8 提交 »936e8a3bf3 rcm: wopi (#1607)7796c2d06a add jxl support for autogrid (#1610)adf5f639fa translation: rus review (#1612)d4a1ce48e4 thumbex: collabora example (#1617)0a957bf264 thumbex: update example link (#1629)
- frostbyte_neo 推送了仓库 github-featured/google--adk-python 的 main 分支比较 8 提交 »40c8333460 fix: read an MCP tool's schema under either spelling in telemetryf449bb70c3 fix: rebind an agent's own callback methods when cloningf842a15cdb fix: balance callback lifecycle for hallucinated tool callsadbee3cd86 fix: coerce an integral float back to int for an int-typed tool parameter2dee1229f1 chore(deps): move aiohttp to the extras that need it and drop tzlocal
- frostbyte_neo 推送了仓库 github-featured/lightseekorg--tokenspeed 的 hongtaoc/k3-latent-tail-fast-paths 分支比较 4 提交 »76ef6f3ad4 Merge branch 'main' into hongtaoc/k3-latent-tail-fast-paths795a9a1255 perf(k3): admit latent tail dependents early504148ce82 perf(k3): accelerate top-16 latent finalizef5360210ef perf(k3): right-size latent tail launches
- frostbyte_neo 推送了仓库 github-featured/lightseekorg--tokenspeed 的 perf/k3-decode-gates-and-gemv-route 分支c48743137f perf(k3): decode gates, GEMV route QA, and FlashInfer 0.6.18 re-tune
- frostbyte_neo 推送了仓库 github-featured/lightseekorg--tokenspeed 的 shared/glm-5.3-flash 分支比较 32 提交 »de9f442bf8 fix(glm-5.3-flash): simplify integration and CI6385d9ffe8 fix(ci): extend GLM-5.3-Flash AIME reasoning budgetb442cedb2a fix(ci): fetch GLM-5.3-Flash tokenizer from Hugging Face75daa46e3e fix(glm-5.3-flash): address review feedback7c90374e47 fix(ci): cache FlashInfer JIT wheel downloads
- frostbyte_neo 推送了仓库 github-featured/lightseekorg--tokenspeed 的 fix/prefill-graph-cache-metadata-gate 分支e6d90b26c5 refactor(prefill graph): one capture-table width rule, no per-backend knobs
- frostbyte_neo 推送了仓库 github-featured/lightseekorg--tokenspeed 的 hongtaoc/flashinfer-0618-kda-decode 分支比较 13 提交 »0fe6d17e98 Centralize KDA convolution state geometry6da4f14c1e Tighten FlashInfer KDA decode contracts09c6c612c5 Integrate FlashInfer KDA decode on Blackwell63f45850a2 ci: route MI455 simulation to self-hosted runner (#1319)89a631321d feat(gemm): add a bf16 dense-linear form of the ll bf16 gemm (#1276)
- frostbyte_neo 推送了仓库 github-featured/lightseekorg--tokenspeed 的 main 分支比较 2 提交 »a75618a4bd ci: make the Kimi-K3 TP8/EP1 jobs dispatchable (#1327)d35dbb2308 perf(l2): reduce Host cache transfer and scheduling overhead (#1325)
- frostbyte_neo 推送了仓库 github-featured/onnx--onnx 的 main 分支比较 3 提交 »24b862d924 Add `mode` attribute (DCR/CRD) to SpaceToDepth in opset 28, add function bodies to SpaceToDepth and DepthToSpace (#7848)31e8cc6052 feat(defs): support bfloat16 in Compress and related ops (#8078)20a059a14f fix(checker): reject non-canonical FLOAT6 encodings (#8386)
- frostbyte_neo 推送了仓库 github-featured/onnx--onnx 的 copilot/allow-float-input-to-mod 分支比较 11 提交 »4eef62525c test(backend): verify Mod signed-zero outputac27f31111 chore: merge main into Mod update24b862d924 Add `mode` attribute (DCR/CRD) to SpaceToDepth in opset 28, add function bodies to SpaceToDepth and DepthToSpace (#7848)80ef9e0890 chore: merge main into Mod updateffc95815bb fix(defs): specify Mod floating-point edge cases
- frostbyte_neo 推送了仓库 github-featured/onnx--onnx 的 copilot/optimize-onnx-reference-op-conv 分支比较 6 提交 »5fa01f86b9 Merge branch 'main' into copilot/optimize-onnx-reference-op-conv24b862d924 Add `mode` attribute (DCR/CRD) to SpaceToDepth in opset 28, add function bodies to SpaceToDepth and DepthToSpace (#7848)31e8cc6052 feat(defs): support bfloat16 in Compress and related ops (#8078)20a059a14f fix(checker): reject non-canonical FLOAT6 encodings (#8386)3ab9181bbe feat: add FLOAT6E2M3 and FLOAT6E3M2 quantization data types (#8349)
