Commit Graph

  • a5b1c7006e fix image name main HaimKortovich 2026-05-11 15:04:37 -05:00
  • ee6129d54e push run num HaimKortovich 2026-05-11 15:03:02 -05:00
  • f30289ec57 fix tag and push HaimKortovich 2026-05-11 15:01:19 -05:00
  • 97e6afcf3b fix label HaimKortovich 2026-05-11 14:50:08 -05:00
  • eae788259a run job on arm64 HaimKortovich 2026-05-11 14:48:16 -05:00
  • 896cdefedf build on arm64 HaimKortovich 2026-05-11 14:40:20 -05:00
  • d3dbfb682a set docker platform to arm64 HaimKortovich 2026-05-11 14:34:17 -05:00
  • 0bb0da779e run using bash HaimKortovich 2026-05-11 13:40:42 -05:00
  • f307d8dc76 Merge branch 'main' of gitea.corredorconect.com:software-engineering/spark-vllm-docker HaimKortovich 2026-05-11 13:21:53 -05:00
  • 1d0fe50d46 login using action HaimKortovich 2026-05-11 13:21:19 -05:00
  • f24d177802 Update README.md haimkortovich 2026-05-11 18:20:33 +00:00
  • bb0d120177 gitea workflow HaimKortovich 2026-05-11 13:16:59 -05:00
  • ba9dde963f Fixed 3-node Qwen 397B recipe to prevent OOM and use instanttensor prebuilt-vllm-current prebuilt-flashinfer-current eugr 2026-05-10 22:20:49 -07:00
  • ae8ac815ac Adjusted Qwen3.5-397B recipe to fix OOM issue and lower memory requirements eugr 2026-05-09 13:45:15 -07:00
  • 83a680c87b Fixed OOM for Qwen3.5-397B eugr 2026-05-09 13:25:31 -07:00
  • 69ea62294f remove unnecessary mod from qwen3-coder-next template Eugene Rakhmatulin 2026-05-08 16:32:54 -07:00
  • 8e548ce664 Fixed typo Eugene Rakhmatulin 2026-05-08 14:59:13 -07:00
  • bca64f9a53 Performance regression fix Eugene Rakhmatulin 2026-05-08 13:40:55 -07:00
  • 29d5904b80 Fix performance regression Eugene Rakhmatulin 2026-05-08 12:56:28 -07:00
  • b87854fd4c Fixed qwen3.6 recipes Eugene Rakhmatulin 2026-05-06 10:56:09 -07:00
  • c67c5b5c1e Add chat template and recipe for Qwen3.6-35B-A3B-FP8 model Eugene Rakhmatulin 2026-05-06 10:32:46 -07:00
  • 9fbed882bc Added EXPERIMENTAL mod for b12x - initial support Eugene Rakhmatulin 2026-04-29 14:38:37 -07:00
  • 97e51d5d23 fixed gemma4 recipe Eugene Rakhmatulin 2026-04-29 12:56:07 -07:00
  • 87cb9f6e1e Reverted gemma4 to safetensors. Fixes #214 and #217. Eugene Rakhmatulin 2026-04-29 10:56:40 -07:00
  • e3243bf555 Merge pull request #197 from mmonad/minimax-m2.7-awq-recipe eugr 2026-04-25 19:26:43 -07:00
  • 43a00ed90f Fixed #205 Eugene Rakhmatulin 2026-04-25 18:39:46 -07:00
  • ef9b0e50f4 Merge pull request #210 from Kaweees/main eugr 2026-04-25 10:00:52 -07:00
  • c1e952de2e Update gpu-mem-util-gb: patch with new vLLM default value Miguel Villa Floran 2026-04-24 11:40:41 -07:00
  • b13a3600d3 Remove a dependency Eugene Rakhmatulin 2026-04-23 07:47:23 -07:00
  • 7dea11bbf0 More robust handling of PRs Eugene Rakhmatulin 2026-04-22 13:18:12 -07:00
  • c187912e23 Removed merged PRs Eugene Rakhmatulin 2026-04-21 09:47:26 -07:00
  • caa28c8e12 Add recipe for MiniMax-M2.7-AWQ L.B.R. 2026-04-18 22:43:09 +01:00
  • 5415c1fe9e Include a PR to fix broken torch bindings (vllm pr 40191) Eugene Rakhmatulin 2026-04-18 09:19:50 -07:00
  • d49fac1b8b Re-enable flashinfer_cutlass Eugene Rakhmatulin 2026-04-16 16:40:56 -07:00
  • 6b7f8dace6 Fixes #187 Eugene Rakhmatulin 2026-04-15 22:32:14 -07:00
  • 76fbf0d0be Fix for broken MiniMax M2 parser Eugene Rakhmatulin 2026-04-15 16:31:50 -07:00
  • b7830469be Updated README Eugene Rakhmatulin 2026-04-14 17:23:42 -07:00
  • b50fa426c8 Merge pull request #190 Eugene Rakhmatulin 2026-04-14 17:18:56 -07:00
  • 2c13e1ce25 Add InstantTensor to runtime dependencies Tim Messerschmidt 2026-04-14 19:38:36 +02:00
  • c026c92bd0 Updated README Eugene Rakhmatulin 2026-04-13 11:27:57 -07:00
  • cf4cb35356 added new flashinfer build dependency Eugene Rakhmatulin 2026-04-13 08:47:34 -07:00
  • 1ad85442ac Added a helper mod for Qwen3.5-397B recipe Eugene Rakhmatulin 2026-04-12 19:14:23 -07:00
  • 30919581ee Included .gitgnore in wheels Eugene Rakhmatulin 2026-04-11 17:02:39 -07:00
  • b7c8616743 Pinned pytorch version Eugene Rakhmatulin 2026-04-11 11:54:46 -07:00
  • 8e8e850ef1 fix for new requirements structure Eugene Rakhmatulin 2026-04-10 20:14:47 -07:00
  • fc08740fba Increased uv timeout Eugene Rakhmatulin 2026-04-10 19:38:38 -07:00
  • 288da8e911 Mod to fix Gemma4 tool parser Eugene Rakhmatulin 2026-04-04 16:48:07 -07:00
  • 7bc4e4ce5e Fixes #158 by adding build args to gemma4 recipe Eugene Rakhmatulin 2026-04-04 10:46:06 -07:00
  • 49d6d9fefd Removed PR2927 as it's been merged Eugene Rakhmatulin 2026-04-03 16:56:00 -07:00
  • 4afca860a5 Fix broken compilation (PR 38919) Eugene Rakhmatulin 2026-04-03 10:22:10 -07:00
  • ed32612cdd A recipe for Gemma4-26B Eugene Rakhmatulin 2026-04-02 23:53:55 -07:00
  • 44808f7018 Apply vLLM PR 35568 Eugene Rakhmatulin 2026-04-02 17:13:54 -07:00
  • 12caec228e switching gpt-oss-120b to solo only for now Eugene Rakhmatulin 2026-04-01 10:27:50 -07:00
  • 27eb35f08d Fixed 4x qwen recipe Eugene Rakhmatulin 2026-04-01 10:09:01 -07:00
  • 3335540972 Merge branch 'pr-152' eugr 2026-04-01 08:59:01 -07:00
  • ae25d64ac0 Changed CUTLASS ref for mxfp4 build eugr 2026-04-01 08:58:31 -07:00
  • a770865834 Updated PRs to apply Eugene Rakhmatulin 2026-04-01 08:31:34 -07:00
  • 7b47235463 Pin nvidia-nvshmem-cu13 to <3.6 in Dockerfile.mxfp4 Artyom 2026-04-01 07:38:53 +02:00
  • 3a3ab98b3e Temporarily added PR2897 to Dockerfile Eugene Rakhmatulin 2026-03-31 22:06:08 -07:00
  • 23fb7dcc20 Merge branch '3-node-autodiscover' Eugene Rakhmatulin 2026-03-31 18:22:23 -07:00
  • c4860b86a2 Updated README with 3-node support Eugene Rakhmatulin 2026-03-31 18:19:22 -07:00
  • 044557943c Bugfixes Eugene Rakhmatulin 2026-03-31 17:49:17 -07:00
  • ead749239d Bugfix Eugene Rakhmatulin 2026-03-31 16:57:56 -07:00
  • a889fed254 Updated README Eugene Rakhmatulin 2026-03-31 16:54:19 -07:00
  • e89104d91b Always rerun discovery when --discover is specified Eugene Rakhmatulin 2026-03-31 16:25:05 -07:00
  • 15a04ada32 Bug fixes Eugene Rakhmatulin 2026-03-31 16:20:23 -07:00
  • a467a7a0bd Updated README for 3-node Eugene Rakhmatulin 2026-03-31 13:47:04 -07:00
  • 48318380f9 Bugfix Eugene Rakhmatulin 2026-03-31 13:41:35 -07:00
  • 287d3c72e5 Fix for forced autodiscovery Eugene Rakhmatulin 2026-03-31 13:34:59 -07:00
  • 9370b2bb34 Don't start the cluster if only --setup/--discover is specified Eugene Rakhmatulin 2026-03-31 13:29:56 -07:00
  • bb177383ff Bugfix in autodiscovery dedup Eugene Rakhmatulin 2026-03-31 12:46:15 -07:00
  • 7f0be29fcc Handle edge case when two sparks have both cables plugged and assigned IPs Eugene Rakhmatulin 2026-03-31 11:59:03 -07:00
  • 41c0ce2c9a Fixed FI PR Eugene Rakhmatulin 2026-03-30 14:25:42 -07:00
  • 45494688d1 Updated README, added NVFP4 fix Eugene Rakhmatulin 2026-03-30 11:45:40 -07:00
  • a3201f8873 --flashinfer-ref / --apply-flashinfer-pr Eugene Rakhmatulin 2026-03-29 22:40:35 -07:00
  • e471ca2436 Don't copy if -c is not specified Eugene Rakhmatulin 2026-03-28 18:12:32 -07:00
  • 32674c2619 removed temporary patch as it causes more issues. Eugene Rakhmatulin 2026-03-28 17:49:17 -07:00
  • 47f5f931b5 Allow to specify config file when doing setup Eugene Rakhmatulin 2026-03-28 14:55:31 -07:00
  • d37217bad0 moved PR patch before the requirements patching Eugene Rakhmatulin 2026-03-28 09:22:19 -07:00
  • e70c87b4f6 Added PR38423 (temp) Eugene Rakhmatulin 2026-03-28 08:50:54 -07:00
  • c1a6cec074 Updated documentation; default image tags in build script Eugene Rakhmatulin 2026-03-27 16:41:09 -07:00
  • 51d69c5c17 commenting out non-applicable PRs Eugene Rakhmatulin 2026-03-27 16:15:54 -07:00
  • e7f2ee692f Added temporary patch to apply PR38126 that fixes broken NVFP4 quants Eugene Rakhmatulin 2026-03-27 09:30:26 -07:00
  • 101ae6fd56 Merge branch 'main' into 3-node-autodiscover Eugene Rakhmatulin 2026-03-27 09:02:10 -07:00
  • f4ca15ce18 Made autoround mod optional to support latest version of vLLM. Fixes #144. Eugene Rakhmatulin 2026-03-27 09:00:50 -07:00
  • 3d918e0b82 Merge branch '3-node' into 3-node-autodiscover Eugene Rakhmatulin 2026-03-27 07:51:08 -07:00
  • 47a896d722 Removed expert-parallel from 3x-node Qwen eugr 2026-03-26 22:44:48 -07:00
  • 0fa585f909 Fix typo in pipeline_parallel setting in Qwen3.5-397B-INT4-Autoround recipe Eugene Rakhmatulin 2026-03-26 18:43:17 -07:00
  • cecec74828 Add recipe for Qwen3.5-397B-INT4-Autoround in pipeline-parallel mode Eugene Rakhmatulin 2026-03-26 18:41:57 -07:00
  • c8ee2a2511 Perform node count check in any mode Eugene Rakhmatulin 2026-03-26 18:15:09 -07:00
  • ce293b5f05 Additional checks for parallelism and cluster size Eugene Rakhmatulin 2026-03-26 17:52:47 -07:00
  • f872cc17a8 Fix for --setup behavior Eugene Rakhmatulin 2026-03-26 16:49:09 -07:00
  • 00c16746e5 Handle new copy hosts setup in run-recipe.py Eugene Rakhmatulin 2026-03-26 16:45:35 -07:00
  • f163ca69de Autodiscover tweaks Eugene Rakhmatulin 2026-03-26 16:30:05 -07:00
  • a78e221de3 Autodiscovery refactoring with mesh support Eugene Rakhmatulin 2026-03-26 15:47:41 -07:00
  • e6ee108cdf Temporary patch for NVFP4 Eugene Rakhmatulin 2026-03-26 11:43:44 -07:00
  • 174de6f0a8 temporary patch for PR38126 Eugene Rakhmatulin 2026-03-26 08:58:04 -07:00
  • 83a74bccec Removed extra solo mode check Eugene Rakhmatulin 2026-03-26 07:45:23 -07:00
  • ff18a9ad5b Merge branch '3-node' of gitlab.home.eugr.net:ai/spark-vllm into 3-node Eugene Rakhmatulin 2026-03-25 23:38:44 -07:00
  • c08b34a218 add --config passthrough to run-recipe Eugene Rakhmatulin 2026-03-25 23:35:52 -07:00