monster/trellis-2-mrp-mlx
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

Commit Graph

  • 0db816b55d vertex baker: export contract + vertex normals; validator accepts COLOR_0 asset class main m3ultra 2026-07-20 03:53:35 +1000
  • 7860148eb2 spconv Metal kernel (opt-in) + fix vertex baker scheduler fallthrough m3ultra 2026-07-20 03:37:20 +1000
  • fa972aa345 mlx_backend: pure-MLX FlowEuler+CFG(+interval) sampler loop m3ultra 2026-07-20 03:01:57 +1000
  • 746e727da2 generate_asset: --baker vertex — IDW voxel-grid vertex-color baker (dark-patch fix) m3ultra 2026-07-19 19:16:08 +1000
  • 754d403e62 test: isolate standalone Metal integration smoke pr175-head-754d403 Jourloy 2026-07-17 10:09:01 +0300
  • 847bd59c25 fix: use SDPA for dense attention on macOS Jourloy 2026-07-17 09:44:06 +0300
  • 1488ba1505 fix: distinguish requested and safety decimation Jourloy 2026-07-17 09:27:47 +0300
  • 83cf977a8f test: probe production split-k Metal kernel Jourloy 2026-07-17 09:27:07 +0300
  • b50b5c4fc5 fix: retry Metal build with ABI fallback Jourloy 2026-07-17 09:26:00 +0300
  • 2b7f47fc87 fix: cache only pinned runtime weights Jourloy 2026-07-17 09:24:41 +0300
  • f59e6e5981 feat: add native Apple Silicon asset pipeline Jourloy 2026-07-17 09:21:34 +0300
  • 9c34570ffc deps: pin torch>=2.11.0 in requirements_macos.txt Pedro Augusto 2026-04-22 01:10:56 +0100
  • f3177cd807 sparse/config: default ATTN to flex_gemm_sparse_attn on Darwin when flex_gemm probe passes Pedro Augusto 2026-04-22 00:26:44 +0100
  • cf886fd00e benchmarks/e2e_decoder: shell-conv + attention block micro-bench Pedro Augusto 2026-04-22 00:17:19 +0100
  • 2b33c15277 sparse/attention: drop FUSED_ATTN_MAX_SEQLEN threshold to match CUDA-path precedent Pedro Augusto 2026-04-22 00:14:53 +0100
  • 369cb6c6fe sparse/norm: CPU-staged zeros_like workaround for MPS Pedro Augusto 2026-04-21 16:27:02 +0100
  • c869eb995a sparse/attention: add 'flex_gemm_sparse_attn' backend option Pedro Augusto 2026-04-21 16:01:32 +0100
  • 71f071fc76 sparse/config: probe flex_gemm masked path too, not just dense Pedro Augusto 2026-04-21 14:54:27 +0100
  • 2780bc459a sparse/config: probe flex_gemm on MPS, soft-fall-back to pytorch backend Pedro Augusto 2026-04-21 03:04:22 +0100
  • d15e6de829 inline env defaults in api_server, remove start script, drop intermediate mx.eval syncs Pedro Augusto 2026-03-22 14:11:49 +0000
  • 9cd27537d9 add mlx backend for apple silicon with metal gpu acceleration Pedro Augusto 2026-03-17 18:23:31 +0000
  • 75fbf01830
         Merge pull request #166 from microsoft/copilot/fix-failing-github-actions-job Jianfeng Xiang 2026-06-05 13:19:14 +0800
  • 173e6ee351
         Fix CodeQL c-cpp analysis: change build-mode from autobuild to none copilot-swe-agent[bot] 2026-06-05 05:15:34 +0000
  • a908f0e246
         Initial plan copilot-swe-agent[bot] 2026-06-05 05:13:09 +0000
  • d3b9d483e9
         Add CodeQL analysis workflow configuration Jianfeng Xiang 2026-06-05 13:03:23 +0800
  • 5565d240c4 Release Training Code JeffreyXiang 2026-01-10 09:47:30 +0000
  • 903bfcf51a update texturing pipeline JeffreyXiang 2025-12-23 12:57:08 +0000
  • 1762f493fe update paper link JeffreyXiang 2025-12-17 02:27:18 +0000
  • 10b3de9e6f Fix website url JeffreyXiang 2025-12-16 20:13:17 +0000
  • 4f15c72a17 Add demo link JeffreyXiang 2025-12-16 19:15:03 +0000
  • de38fdd408 Init Release JeffreyXiang 2025-12-09 09:21:05 +0000
  • 7f9ccee603
         Initial commit microsoft-github-operations[bot] 2025-11-26 10:04:36 +0000