Commit Graph

  • a05a88645b try removing shared cache dir mount Tyler Burt 2026-01-12 12:55:31 -0800
  • 2385df7173 Update the FI version to 0.6.0 Chenghao Zhang 2026-01-12 12:55:12 -0800
  • ff749bc0ef
    Merge branch 'main' into fix_spec_gate Zheyu Fu 2026-01-12 12:53:23 -0800
  • 6c1d31cfa9
    Merge 91fab5d670 into 18a33764b5 DylanChen-NV 2026-01-12 20:45:48 +0000
  • b45eb98362
    Merge 9ccd0be96d into 18a33764b5 Ludwig Schneider 2026-01-12 14:39:11 -0600
  • 9ccd0be96d unwaived other failing test as well (same suspected root cause) Ludwig Schneider 2026-01-12 14:38:02 -0600
  • 9f9af89783 unwaive test, a previous PR may have resolved hang Ludwig Schneider 2026-01-07 12:44:01 -0600
  • 78c0f31c00 activate NCCL_SYMMETRIC auto-tuning Ludwig Schneider 2026-01-06 09:45:25 -0800
  • fa161fd25e
    Merge 9af9952a5d into 18a33764b5 Thor Johnsen 2026-01-12 15:34:52 -0500
  • 1bdc7cce5f
    Merge a01cd7a11d into 18a33764b5 ebarilanM 2026-01-12 20:33:35 +0000
  • 8c37a73f6d
    Merge branch 'main' into feat/aether-sparse-attention teerth sharma 2026-01-13 01:30:06 +0530
  • 9b77d9da57
    Merge de0e50d611 into 18a33764b5 Linda 2026-01-13 03:55:37 +0800
  • a343a70be3 Skip triton mxfp4 unit test on blackwell Anish Shanbhag 2026-01-12 11:54:11 -0800
  • 621caae9c9 Remove comment Anish Shanbhag 2026-01-08 22:06:53 -0800
  • 861cbd53a6 Don't use autotuner for triton backend Anish Shanbhag 2026-01-07 14:29:08 -0800
  • 4486df0a8a Skip tests on non-Hopper Anish Shanbhag 2026-01-07 14:07:52 -0800
  • d79b36521a Fix bias shapes in Triton MoE methods Anish Shanbhag 2026-01-07 10:11:43 -0800
  • 2ab362cafe [None][feat] Include triton-kernels as a packaged dependency Anish Shanbhag 2026-01-06 17:35:34 -0800
  • 18a33764b5
    [None][chore] Print correct backend name in benchmark report (#10597) Gal Hubara-Agam 2026-01-12 21:46:00 +0200
  • ad9f473613
    Merge branch 'main' into feat/aether-sparse-attention teerth sharma 2026-01-13 01:13:22 +0530
  • dda588ec23 revert back Eagle3DecodingConfig migrations for trt backend tests Venky Ganesh 2026-01-12 11:31:41 -0800
  • 4687b415cf
    Merge 5f1748d1c0 into dacc881993 HuiGao-NV 2026-01-13 03:08:39 +0800
  • 09db89a677
    Merge b27939f839 into dacc881993 Daniel Stokes 2026-01-13 03:08:36 +0800
  • bcf38b255b
    Merge e93de22c6e into dacc881993 Yiqing Yan 2026-01-13 03:05:18 +0800
  • a6047439e8
    Merge 776f5f1d2a into dacc881993 Song Rong 2026-01-13 03:01:59 +0800
  • 3f2821b9a4
    Merge c20cc562c5 into dacc881993 v-shobhit 2026-01-13 03:01:17 +0800
  • d9b16cf239
    Merge 9ecc7789db into dacc881993 kris1025 2026-01-13 02:59:10 +0800
  • 32db0e8a0c
    Merge c5e686d510 into dacc881993 xinhe-nv 2026-01-13 02:57:58 +0800
  • 51125ec5d5
    Merge 02a6d177f3 into dacc881993 Bo Li 2026-01-13 02:57:52 +0800
  • dacc881993
    [https://nvbugs/5761391][fix] Use correct model names for config database regression tests (#10192) Anish Shanbhag 2026-01-12 10:55:07 -0800
  • dc401b5ec9 Fix import sorting Anish Shanbhag 2026-01-05 11:31:33 -0800
  • 3f38f0ef92 Remove redundant speculative model download for TRTLLM backend Anish Shanbhag 2026-01-05 11:01:01 -0800
  • 92d918bced Use model validators for eagle config Anish Shanbhag 2025-12-30 15:26:40 -0800
  • c8df1270f7 Use AliasChoices Anish Shanbhag 2025-12-22 22:37:01 -0800
  • 03ed3d6bcf Fixes Anish Shanbhag 2025-12-22 18:08:22 -0800
  • d6a2891b73 Add missing import Anish Shanbhag 2025-12-22 11:41:35 -0800
  • 28b49af586 Move download to shared helper Anish Shanbhag 2025-12-18 18:14:36 -0800
  • 76798fc52b Mock snapshot_download to avoid download from HF Anish Shanbhag 2025-12-18 17:48:45 -0800
  • d6a00ee980 Add logs Anish Shanbhag 2025-12-17 21:28:58 -0800
  • 3e38551ee7 [None][feat] Auto download speculative models from HF for pytorch backend, add speculative_model field alias Anish Shanbhag 2025-12-17 17:40:35 -0800
  • b97044612f Turn off KV block reuse. Remove target_sparsity_0 from CI. Bo Li 2026-01-10 22:00:09 +0800
  • b07c5668f5 Update ref accuracy. Use lower KV cache free GPU mem frac. Bo Li 2026-01-06 07:22:10 +0000
  • 5a662a607c Update nvbugs for waive on Blackwell. Bo Li 2026-01-05 17:21:37 +0000
  • 15ef830b33 Use 2 GPUs to test skip softmax attention on H100. Bo Li 2026-01-05 17:11:52 +0000
  • cdd4ad7dc5
    Merge 1835a076b4 into a1385243e1 Jhao-Ting Chen 2026-01-12 17:52:53 +0000
  • cf1c2d6b86
    Merge branch 'main' into feat/aether-sparse-attention teerth sharma 2026-01-12 23:12:52 +0530
  • c8d7d3b1c9
    Merge e9ae5b5182 into a1385243e1 Rongwei Zhang 2026-01-12 09:26:25 -0800
  • a1385243e1
    [#10580][fix] re-enable NemotronH MOE MMLU test (#10594) Suyog Gupta 2026-01-12 09:26:07 -0800
  • 37504c823d fix test Yibin Li 2026-01-07 21:57:07 +0000
  • 04b992b913 move transfer remaining logits logic to handle_response Yibin Li 2026-01-06 22:27:52 +0000
  • cad3b9e3e8
    Merge 467cdc3e7e into 9f044b9dd9 Jie Li 2026-01-12 16:48:06 +0000
  • 01cf98132a fix: remove unused import ixlmar 2026-01-06 16:59:35 +0100
  • 240eff4bd8 fix: conform to upstream ixlmar 2025-12-13 16:27:24 +0000
  • eb9a24e6f0 chore: refine MultimodalDataTracker.add_data ixlmar 2025-12-12 16:59:08 +0000
  • e4233671a9 address remaining review comments ixlmar 2025-12-11 13:02:16 +0000
  • e42a8f7d64 chore: use torch.save/load ixlmar 2025-12-11 12:58:20 +0000
  • db14542c35 chore: add is_embedding to MultimodalData ixlmar 2025-12-11 12:58:03 +0000
  • 0543bf01fb fix: update docs ixlmar 2025-12-08 17:02:45 +0000
  • bebc2e4317 do not run 'test_multimodal_content_mm_encoder' twice ixlmar 2025-12-05 15:33:59 +0000
  • b2a328c706 add nested "data" ixlmar 2025-12-05 14:01:39 +0000
  • 045331d494 fix: run test_single_chat_session_image_embeds on L40S ixlmar 2025-12-05 12:52:54 +0000
  • 5e3c26ebfb feat: support image_embeds in OpenAI API ixlmar 2025-12-04 17:39:03 +0100
  • b94ffc2d32 Merge remote-tracking branch 'upstream/main' into user/tjohnsen/fix_5721661 . thorjohnsen 2026-01-12 16:28:24 +0000
  • 05b1329da4 Merge branch 'user/tjohnsen/fix_5721661' of github.com:thorjohnsen/TensorRT-LLM into user/tjohnsen/fix_5721661 . thorjohnsen 2026-01-12 16:27:54 +0000
  • 3543975c12 Fix undefined behavior (imnplicit erase of object pointed to by iterator in a loop) thorjohnsen 2026-01-12 16:27:13 +0000
  • 000867d9b1 [None][chore] Update flashinfer to 0.6 Mike Iovine 2026-01-12 07:07:44 -0800
  • 678948a8bc Fix MTP 1-model sampler Mike Iovine 2025-12-31 13:02:19 -0800
  • 3209caf92c
    Merge ef7bc0fa24 into 9f044b9dd9 Timothy Gao 2026-01-12 15:56:11 +0000
  • cc441386fc Add Nemotron Nano 3 30B FP8 autodeploy perf test Eran Geva 2026-01-12 07:49:59 -0800
  • 7539cc580f
    Merge 5cb27b7377 into 9f044b9dd9 Thor Johnsen 2026-01-12 16:27:11 +0100
  • 9f044b9dd9
    [None][infra] Waive failed tests for main 01/12 (#10604) Emma Qiao 2026-01-12 23:24:54 +0800
  • e415cbfd77 doc: explain top-p-after-top-k ixlmar 2026-01-12 15:11:50 +0000
  • ad33bd45b0 Waive failed tests for main qqiao 2026-01-12 06:57:34 -0800
  • 6f66f82ee0
    Merge c85e31b516 into bf7998f1b8 Zongfei Jing 2026-01-12 11:55:51 -0300
  • a20c7383fc
    [None][fix] Raise error when visualization enabled but model-explorer not installed Karthik Vetrivel 2026-01-07 15:12:56 +0000
  • 75c49c6944 Waive failed tests for main qqiao 2026-01-12 06:53:49 -0800
  • f47cc30fab update Chenfei Zhang 2026-01-12 06:27:27 -0800
  • 5525a8544f fix https://nvbugs/5777041 Lucas Liebenwein 2026-01-06 08:00:00 -0800
  • 1b1da74143 doc: document Torch Sampler details ixlmar 2026-01-12 14:05:22 +0000
  • 432f185dee Remove unnecessary ray_placement_config from mapping. Wangshanshan 2026-01-12 00:31:45 -0800
  • e771ff820a Yaml to RayPlacementConfig convert moved in LlmArgs and keep it simple. Wangshanshan 2026-01-12 00:15:52 -0800
  • 646c58f157 Little clean up from code review. Wangshanshan 2026-01-09 04:21:55 -0800
  • f0e9ee3de9 Refactor a bit. Wangshanshan 2026-01-07 20:38:57 -0800
  • ea629253e2 Unwaive test and update the waive list. Wangshanshan 2026-01-06 05:22:21 -0800
  • 565e230215 Use RayPlacementConfig in disagg and bug fixed. Wangshanshan 2026-01-06 01:24:58 -0800
  • dae7c1cbdb Fix still in progress. Wangshanshan 2025-12-18 00:59:07 -0800
  • 03c7709801 Unwaive disagg test. Wangshanshan 2025-12-16 00:13:55 -0800
  • edb47e4cb2 Fix gpu allocation conflicts in ray disagg. Wangshanshan 2025-12-16 00:08:43 -0800
  • 494e456ac3
    Merge branch 'main' into dev-liji-unwaive-llama Jin Li 2026-01-12 21:34:01 +0800
  • 118c534c78 removing todo Ludwig Schneider 2026-01-07 12:44:01 -0600
  • 04999028c5 activate NCCL_SYMMETRIC auto-tuning Ludwig Schneider 2026-01-06 09:45:25 -0800
  • f1ec84cc33
    Merge branch 'main' into feat/aether-sparse-attention teerth sharma 2026-01-12 18:38:23 +0530
  • dc5db6deaa refactor code from some comments benzh-2025 2026-01-12 13:06:20 +0000
  • 4a5fd17e1d change to get_sm_version benzh-2025 2026-01-12 09:32:02 +0000
  • e50b32338a add nvls check benzh-2025 2026-01-12 06:56:12 +0000
  • 31f6103033 add sm103 support benzh-2025 2026-01-07 03:07:38 +0000
  • ca539fe83b support gemm+allreduce only on arch >= blackwell benzh-2025 2025-12-30 06:44:23 +0000
  • 4baa943f39 refactor logical check for gemm+allreduce fusion benzh-2025 2025-12-24 05:07:53 +0000
  • 80fe0386be add fp4 gemm + allreduce benzh 2025-12-05 08:05:34 +0000
  • 4fee1914a5 [TRTLLM-10245][feat] Add accuracy tests for super v3 fp8 model Wanli Jiang 2026-01-06 21:42:41 -0800