From 09125567eaa9267ddf7d4fdd287e3c9e650bf038 Mon Sep 17 00:00:00 2001 From: lishunyang12 Date: Wed, 15 Jul 2026 10:12:02 +0000 Subject: [PATCH] docs: pin FlashInfer/ModelOpt versions in upstream-mapping --- docs/design/skip_softmax_sage_trtllm_gen.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/docs/design/skip_softmax_sage_trtllm_gen.md b/docs/design/skip_softmax_sage_trtllm_gen.md index 1c373ae16..f2f4b829c 100644 --- a/docs/design/skip_softmax_sage_trtllm_gen.md +++ b/docs/design/skip_softmax_sage_trtllm_gen.md @@ -61,6 +61,8 @@ Env var = flag in UPPER_SNAKE (e.g. `TRTLLM_GEN_SAGE=1`). | `target_sparsity` | `threshold_scale_factor` (via calibration) | `skip_softmax_threshold_scale_factor` | | `disabled_until_timestep` | — (DiT-only) | host gate | +Versions: **FlashInfer 0.6.5 · ModelOpt 0.44.0** (runtime deps); TRT-LLM = naming reference (blog16, no runtime dep). + --- ## Part 2 — Testing Plan