Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions docs/low_precision/fp8.md
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
# FP8 RL in verl

Last updated: 03/05/2026
Last updated: 09/29/2026

verl supports two FP8 modes for accelerating RL training:

Expand Down Expand Up @@ -168,7 +168,7 @@ FP8 E2E applies FP8 to the entire RL pipeline: forward/backward passes via Trans

- **CUDA 12.9+** (required for block-wise FP8 scaling)
- **Transformer Engine** with block-wise FP8 support
- Environment variable: `NVTE_FP8_BLOCK_SCALING_FP32_SCALES=1`
- Environment variable: `NVTE_FP8_BLOCK_SCALING_FP32_SCALES=1`, on **Hopper only**. Do not set it on Blackwell (SM100+): there Transformer Engine runs block-wise scaling through MXFP8 GEMMs, whose E8M0 scales hold only an exponent, so it requires power-of-two block scales and stops at the first FP8 quantization with `Assertion failed: pow2_scale ... requires using power of two scaling factors` when the variable is set (see [#6172](https://github.com/verl-project/verl/issues/6172)). Leaving it unset gives power-of-two scales, the Transformer Engine default.

### Key Configuration

Expand Down