Actions - markbaseengine - Gitea: Git with a cup of tea

warren/markbaseengine

23 workflow runs

v2: install EmbeddingServer with EmbeddingGemma on port 8084

#53:Commit b33947feaf pushed by warren

v2

2026-07-07 01:55:32 +08:00

0s

v2: EmbeddingGemma multi-language working, 500-test stability issue pending

#52:Commit c0347d5213 pushed by warren

v2

2026-07-06 20:24:29 +08:00

0s

v2: EmbeddingGemma multi-language support working (single request), MPS GPU fault known issue

#51:Commit e41e244b25 pushed by warren

v2

2026-07-06 19:17:23 +08:00

0s

v2: fix multi-language support - use SentencePieceTokenizer for GemmaTokenizer

#50:Commit 319f29bf69 pushed by warren

v2

2026-07-06 18:23:01 +08:00

0s

v2: add actor-based serialization for EmbeddingServer, fix stability for ASCII text

#49:Commit 91d44a924d pushed by warren

v2

2026-07-07 17:19:32 +08:00

0s

v2: add EmbeddingGemma model with RoPE, sliding window attention, Q/K norm

#48:Commit 88aeff7935 pushed by warren

v2

2026-07-07 11:19:32 +08:00

0s

v2: add embedding tests, multilingual embedding support

#47:Commit 85dd87e28a pushed by warren

v2

2026-07-07 11:19:32 +08:00

0s

v2: fix E4B determinism test — float32 atomics cause inherent non-determinism

#46:Commit ba4c41c29f pushed by warren

v2

2026-07-07 11:19:32 +08:00

0s

v2: add E4B multimodal test, fix VisionTower missing groupSize

#45:Commit 96fe213bc4 pushed by warren

v2

2026-07-07 05:19:32 +08:00

0s

v2: add full context 2048-token, repeated tokens, edge token tests

#44:Commit 97f9bdcf90 pushed by warren

v2

2026-07-07 05:19:32 +08:00

0s

v2: add 1024-token long context test

#43:Commit 16c16b9bee pushed by warren

v2

2026-07-07 05:19:32 +08:00

0s

v2: add long context 12B test (256 tokens)

#42:Commit 7e686c3c5a pushed by warren

v2

2026-07-07 05:19:32 +08:00

0s

v2: add multimodal 12B test, fix VisionTower12B kernel dispatch

#41:Commit af1d10737e pushed by warren

v2

2026-07-07 05:19:32 +08:00

0s

v2: add 12B model test (Model12BTest)

#40:Commit 07459e8ee3 pushed by warren

v2

2026-07-07 05:19:32 +08:00

0s

v2: remove remaining logit scaling hacks from batch/optimized paths

#39:Commit 7a8edf77ee pushed by warren

v2

2026-07-06 23:19:32 +08:00

0s

v2: fix 26B activation explosion — normalize groupSize=32 scales, fix hardcoded loops

#38:Commit 239474bef0 pushed by warren

v2

2026-07-06 23:19:32 +08:00

0s

v2: add 26B + 31B model tests (Phase 3)

#37:Commit 8a29dae613 pushed by warren

v2

2026-07-06 17:19:32 +08:00

0s

v2: fix GPU non-determinism test tolerance

#36:Commit 2fd03d0ac1 pushed by warren

v2

2026-07-06 17:19:32 +08:00

0s

v2: add E4B-MarkBase model tests (Phase 2)

#35:Commit e9ab994533 pushed by warren

v2

2026-07-06 17:19:32 +08:00

0s

v2: clean up CI test triggers

#34:Commit 97798850e3 pushed by warren

v2

2026-07-06 17:19:32 +08:00

0s

v2: Apply tokenizer UTF-8 fix + Engine writeFloats helper

#33:Commit 31427770b1 pushed by warren

v2

2026-07-06 17:19:32 +08:00

0s

v2: Fix build conflict and unit tests

#32:Commit e23ef405bc pushed by warren

v2

2026-07-06 17:19:32 +08:00

0s

v2: Initial clean branch with unit tests + CI/CD pipeline

#31:Commit 8a66b9086a pushed by warren

v2

2026-07-06 17:19:32 +08:00

0s