Commit Graph

  • dee1e211d5 [docs]: refine dpo tutorial (#1739) mrhaoxx 2025-12-22 18:44:24 +08:00
  • 0bce173e3b [feat]: Release version to 0.4.4 (#1738) v0.4.4 Peilin Li 2025-12-22 11:20:40 +08:00
  • 16d5d89f50 [docs]: Update Python version options in DPO tutorial (#1734) Peilin Li 2025-12-20 13:44:35 +08:00
  • df998e0f36 [docs]: Add RL-DPO Tutorial (#1733) Peilin Li 2025-12-20 12:49:02 +08:00
  • 83a3c7d94f deploy: 39449ed1af ovowei 2025-12-18 09:21:57 +00:00
  • 39449ed1af update PyPI Install and readme (#1731) Jianwei Dong 2025-12-18 17:21:47 +08:00
  • 180d2e0ca7 kernel ok, I think xwy-amd8 2025-12-18 03:58:55 +00:00
  • 6d7bb6f8d1 nvfp4 impl xwy-amd8 2025-12-18 03:31:50 +00:00
  • 8d6f465e33 deploy: 3c134359bc SkqLiao 2025-12-17 16:12:18 +00:00
  • 3c134359bc Fix CPU Instruction Set and Installation (#1729) Jiaqi Liao 2025-12-18 00:11:57 +08:00
  • a8667ddb58 [fix](test): fix import kt-kernel (#1728) ErvinXie 2025-12-17 19:46:32 +08:00
  • 298c3a8c02 progress xwy-amd8 2025-12-17 10:40:10 +00:00
  • 70739cefba [fix](test): fix import kt-kernel fix-import xwy-amd8 2025-12-17 10:00:13 +00:00
  • 2c9a9acebc update docker push fix-deploy-djw djw 2025-12-17 08:20:09 +00:00
  • 6fc4080a7d [fix](kt-kernel): fix typo in moe-tp's forward time-profiling (#1720) SCDESPERTATE 2025-12-17 12:06:33 +08:00
  • 661e19a8e5 Update release-pypi.yml (#1726) Jianwei Dong 2025-12-16 20:37:20 +08:00
  • 5ba3fb56d1 Update release-pypi.yml (#1725) Jianwei Dong 2025-12-16 20:30:03 +08:00
  • 11cfe7ded1 [fix](kt-kernel): fix setup version issue cpu-fix SkqLiao 2025-12-16 19:24:27 +08:00
  • 532c84af2a [feat](kt-kernel): AVX512 fallback kernel for RAW-INT4 SkqLiao 2025-12-16 18:15:28 +08:00
  • 7a29a19900 [fix](kt-kernel): fix AVX512 cpu instruction set detection SkqLiao 2025-12-16 17:57:20 +08:00
  • 3126b8deaa Update release-pypi.yml (#1724) Jianwei Dong 2025-12-16 17:45:29 +08:00
  • 6dcacd9daf Update release-pypi.yml (#1723) Jianwei Dong 2025-12-16 17:42:28 +08:00
  • fe8049d3a9 Update release-pypi.yml (#1722) Jianwei Dong 2025-12-16 17:35:29 +08:00
  • 5ff0026fc1 Update release-pypi.yml (#1721) Jianwei Dong 2025-12-16 17:29:29 +08:00
  • c1938fdbbb deploy: 1f79f6da92 ovowei 2025-12-16 07:20:29 +00:00
  • 1f79f6da92 [feat](kt-kernel): Add automatic deployment workflow (#1719) Jianwei Dong 2025-12-16 15:20:06 +08:00
  • f25e58ad69 fix: qwen3-npu bugs; update: add readme-for-qwen3-npu (#1717) Shaoxu Cheng 2025-12-16 14:27:04 +08:00
  • 18fb8fc897 Npu revise benchmark results and prerequisites (#1716) RICHARDNAN 2025-12-16 14:26:44 +08:00
  • 4b576bc559 deploy: 34230eaf44 KMSorSMS 2025-12-15 09:10:36 +00:00
  • 34230eaf44 [docs]: Fix image link in README.md (#1718) ZiWei Yuan 2025-12-15 01:10:15 -08:00
  • 008de19e16 [fix](kt-kernel): drop the weights held in Python for loading weights operation in C++ (#1695) SCDESPERTATE 2025-12-12 11:42:33 +08:00
  • 1e69563363 update: add cache class and ascend ln mlp op for qwen3 adapt npu (#1708) Shaoxu Cheng 2025-12-11 17:08:35 +08:00
  • cea490a326 update: add ascend attn and experts ops for npu qwen3moe adapt (#1707) Shaoxu Cheng 2025-12-11 17:08:15 +08:00
  • adcfa9080f update: Qwen3 MoE model adaptation for NPU (framework) (#1706) Shaoxu Cheng 2025-12-11 17:07:57 +08:00
  • 53f6a6d6e1 [feat]: patch kml problem (#1704) ZiWei Yuan 2025-12-10 22:40:29 -08:00
  • c65febe05c [feat]: Automatically detect whether blis is installed on amd cpus (#1702) Jianwei Dong 2025-12-11 14:25:36 +08:00
  • 6431888928 add deploy in docker image (#1691) RICHARDNAN 2025-12-11 14:11:27 +08:00
  • f868154f84 deploy: 2f1b743050 KMSorSMS 2025-12-11 05:01:44 +00:00
  • 2f1b743050 [docs]: update website doc png (#1696) ZiWei Yuan 2025-12-10 21:01:32 -08:00
  • e87a042ef0 [fix](kt-kernel): fix write_buffer do numa job (#1699) Oql 2025-12-10 16:39:16 +08:00
  • 8995378a91 update: add attention and ln ut for npu (#1698) Shaoxu Cheng 2025-12-10 16:12:26 +08:00
  • 1e03e8b612 [docs](kt-kernel): add Docker usage guide for SGLang deployment ljq-dev-1 SkqLiao 2025-12-09 14:54:49 +08:00
  • f992de55da [fix](kt-sft): fix peft adaptations for RL tasks (#1674) mrhaoxx 2025-12-09 14:28:51 +08:00
  • 503295fc88 [feat](kt-kernel): refactor convert_cpu_weights.py to support conversation for GLM-4.6V (#1687) mrhaoxx 2025-12-09 14:24:41 +08:00
  • ac69ea891e Fix K2 MoE decode bug in buffer management (#1686) Oql 2025-12-08 21:08:28 +08:00
  • 8139c092bf Reduce CPU memory usage during large chunk prefill (Fixes #1676) (#1683) Oql 2025-12-08 20:19:07 +08:00
  • eefc8cf98d 更新 Kimi-K2-Thinking-Native.md (#1684) ErvinXie 2025-12-08 19:58:20 +08:00
  • 4662368606 sft_route_moe use openmp for parallel in backward (#1662) route-moe Pory 2025-12-06 20:38:50 +08:00
  • f20e5d1da5 Revise prefill strategy and performance metrics (#1675) Jiaqi Liao 2025-12-06 15:36:04 +08:00
  • 1d62ac21f7 Update Kimi-K2-Thinking-Native.md (#1673) Jiaqi Liao 2025-12-05 23:08:02 +08:00
  • 69fa7b1a57 Revise installation steps in Kimi-K2 documentation (#1672) Jiaqi Liao 2025-12-05 23:05:24 +08:00
  • 12ab6abef6 deploy: 721b6c4c94 SkqLiao 2025-12-05 14:46:31 +00:00
  • 721b6c4c94 [docs] Update Native Kimi-K2-Thinking documentation and kt-kernel parameters (#1671) v0.4.3 Jiaqi Liao 2025-12-05 22:46:16 +08:00
  • 47da806cde [doc](kt-kernel): add kimi-k2-thinking (#1670) Jiaqi Liao 2025-12-05 21:53:59 +08:00
  • 71f683acec Support Native Kimi K2 Thinking (#1663) ErvinXie 2025-12-05 21:53:05 +08:00
  • 53cdf1e50e deploy: 4850424345 KMSorSMS 2025-12-05 08:52:39 +00:00
  • 4850424345 [docs]: add amd blis backend usage guide (#1669) ZiWei Yuan 2025-12-05 00:52:26 -08:00
  • e2f2c77343 deploy: 1ca3a2662e SkqLiao 2025-12-05 07:44:18 +00:00
  • 1ca3a2662e Add 9#AISoft to the list of contributors (#1668) Jiaqi Liao 2025-12-05 15:44:04 +08:00
  • 0698252484 [fix](kt-kernel): gate RAWINT4 behind AVX512 and avoid AVX2 build break (#1660) Jiaqi Liao 2025-12-03 00:43:23 +08:00
  • 670c488155 [docs]: Add deepseek-v3.2 run tutorial (#1659) Jianwei Dong 2025-12-02 20:04:10 +08:00
  • fcf8882075 [Feature] Add avx-based kimi-k2 support (#1656) Jiaqi Liao 2025-12-02 16:01:07 +08:00
  • c2b8c60c4e [ci]: add int4_1 & int4_1k (#1653) ZiWei Yuan 2025-12-02 15:58:14 +08:00
  • fd78fe520a fix(scripts): resolve OOM when converting gpu weights and update README (#1640) Jianwei Dong 2025-12-01 14:15:14 +08:00
  • fb787283f6 use amx for calculate of route-moe lora JimmyPeilinLi 2025-12-01 02:27:02 +00:00
  • f0fe6b1b1c deploy: e637fedc65 JimmyPeilinLi 2025-11-29 07:47:45 +00:00
  • c7924fed79 deploy: 7ee80bbc3d JimmyPeilinLi 2025-11-29 07:47:17 +00:00
  • 7a209aa1ab deploy: e637fedc65 JimmyPeilinLi 2025-11-29 07:47:07 +00:00
  • e637fedc65 [docs]: Add Full introduction of KT (#1636) Peilin Li 2025-11-29 15:46:55 +08:00
  • 7ee80bbc3d [docs]: Update README with Python 3.12 and dependency changes (#1634) Peilin Li 2025-11-29 15:46:05 +08:00
  • 47e6abcb11 remove debug info JimmyPeilinLi 2025-11-28 04:18:37 +00:00
  • a11ad178de update unit test of sft_route_moe JimmyPeilinLi 2025-11-28 04:18:06 +00:00
  • 76118a7d5c remove some debug info JimmyPeilinLi 2025-11-28 04:13:05 +00:00
  • 637c49c83f [feat](kt-kernel): support qwen3-vl weights convert (#1648) mrhaoxx 2025-11-27 22:29:09 +08:00
  • 0ccc96477e [debug]: pass the unit test of route moe and other modify, still shared memory problem JimmyPeilinLi 2025-11-27 09:11:52 +00:00
  • c256150e08 update ci test (#1647) Jianwei Dong 2025-11-27 16:39:48 +08:00
  • c25c1e49a5 deploy: 1374b98ee5 KMSorSMS 2025-11-27 04:09:05 +00:00
  • 1374b98ee5 [feat](moe_kernel): add amd blis support (int8) (#1600) ZiWei Yuan 2025-11-27 12:08:53 +08:00
  • fef6dd98a8 add accuracy and performance test (#1643) Jianwei Dong 2025-11-27 10:56:39 +08:00
  • 4ad343afd6 deploy: e7d1c1de09 SkqLiao 2025-11-26 15:19:50 +00:00
  • e7d1c1de09 fix(llamafile): resolve deferred experts data race and update README (#1646) Jiaqi Liao 2025-11-26 23:19:37 +08:00
  • 51745a9ea1 add ci (#1642) Jianwei Dong 2025-11-25 20:52:08 +08:00
  • b113e8714c add route moe sft-amx JimmyPeilinLi 2025-11-25 12:52:03 +00:00
  • 2cffdf7033 [docs]: Update DeepseekR1_V3_tutorial_zh_for_Ascend_NPU.md (#1638) RICHARDNAN 2025-11-24 11:51:07 +08:00
  • e657a648bc deploy: e72a4fb880 KMSorSMS 2025-11-22 04:00:29 +00:00
  • e72a4fb880 [feat](kt-kernel): Add resume arg to CPU weight conversion (#1630) DocShotgun 2025-11-21 20:00:15 -08:00
  • e10760ed4f deploy: e69c67713f SkqLiao 2025-11-20 05:56:08 +00:00
  • e69c67713f [refactor] fix third_party issue (#1632) v0.4.2 Jiaqi Liao 2025-11-20 13:55:55 +08:00
  • 3032a4cfa8 deploy: 46af8fcab5 SkqLiao 2025-11-19 08:42:10 +00:00
  • 46af8fcab5 [doc] fix kt parameters (#1629) Jiaqi Liao 2025-11-19 16:41:57 +08:00
  • c357958fce deploy: 171578a7ec KMSorSMS 2025-11-17 03:48:56 +00:00
  • 171578a7ec [refactor]: Change named 'KT-SFT' to 'kt-sft' (#1626) Peilin Li 2025-11-17 11:48:42 +08:00
  • 2887050ca1 [Feature] add Qwen3MoE models for KTransformers-FT (#1602) Pory 2025-11-16 16:39:19 +08:00
  • d508615c72 deploy: ab8ad0a110 KMSorSMS 2025-11-16 06:40:34 +00:00
  • ab8ad0a110 [docs]: update web doc (#1625) ZiWei Yuan 2025-11-16 14:40:22 +08:00
  • d27834efaf deploy: be6db6f46b KMSorSMS 2025-11-16 05:21:55 +00:00
  • be6db6f46b [docs]: improve structure for kt-kernel (#1624) ZiWei Yuan 2025-11-16 13:21:41 +08:00
  • d19387ffb6 deploy: 133eea037c KMSorSMS 2025-11-16 04:41:13 +00:00
  • 133eea037c [docs]: improve docs structure (#1623) ZiWei Yuan 2025-11-16 12:40:59 +08:00
  • 5d2b4778ff deploy: c2d2edbeef KMSorSMS 2025-11-16 04:09:54 +00:00