From cb8dab06be6f7d708bccf4d4e955780642daba8b Mon Sep 17 00:00:00 2001 From: amote-i <49533125+amote-i@users.noreply.github.com> Date: Tue, 22 Sep 2026 19:04:17 +0800 Subject: [PATCH] [NPU] [DOC] Remove duplicated features in npu docs (#40714) --- .../ascend-npus/optimization/quantization.mdx | 40 +++++++++---------- .../reference/support_features.mdx | 24 ----------- 2 files changed, 20 insertions(+), 44 deletions(-) diff --git a/docs/docs/hardware-platforms/ascend-npus/optimization/quantization.mdx b/docs/docs/hardware-platforms/ascend-npus/optimization/quantization.mdx index 95f64e3a3e..1ac1b7cd91 100644 --- a/docs/docs/hardware-platforms/ascend-npus/optimization/quantization.mdx +++ b/docs/docs/hardware-platforms/ascend-npus/optimization/quantization.mdx @@ -13,9 +13,9 @@ SGLang supports **mix-bits** quantization (independently defines and loads each Quantization scheme Layer type - Ascend A2 Series Products Supported - Ascend A3 Series Products Supported - Ascend 950PR/DT Series Products Supported + A2 Series Supported + A3 Series Supported + 950PR/DT Series Supported Diffusion models @@ -133,9 +133,9 @@ SGLang supports **mix-bits** quantization (independently defines and loads each Quantization scheme Layer type - Ascend A2 Series Products Supported - Ascend A3 Series Products Supported - Ascend 950PR/DT Series Products Supported + A2 Series Supported + A3 Series Supported + 950PR/DT Series Supported @@ -169,9 +169,9 @@ GPTQ on Ascend support Quantization scheme Layer type - Ascend A2 Series Products Supported - Ascend A3 Series Products Supported - Ascend 950PR/DT Series Products Supported + A2 Series Supported + A3 Series Supported + 950PR/DT Series Supported @@ -212,9 +212,9 @@ GPTQ on Ascend support Quantization scheme Layer type - Ascend A2 Series Products Supported - Ascend A3 Series Products Supported - Ascend 950PR/DT Series Products Supported + A2 Series Supported + A3 Series Supported + 950PR/DT Series Supported @@ -255,9 +255,9 @@ Compressed-tensors (LLM Compressor) on Ascend support: Quantization scheme Layer type - Ascend A2 Series Products Supported - Ascend A3 Series Products Supported - Ascend 950PR/DT Series Products Supported + A2 Series Supported + A3 Series Supported + 950PR/DT Series Supported @@ -298,9 +298,9 @@ Compressed-tensors (LLM Compressor) on Ascend support: Quantization type Layer type - Ascend A2 Series Products Supported - Ascend A3 Series Products Supported - Ascend 950PR/DT Series Products Supported + A2 Series Supported + A3 Series Supported + 950PR/DT Series Supported @@ -504,8 +504,8 @@ SGLang-Diffusion supports MXFP8 online and offline quantization for diffusion mo quant_type in JSON Scheme class Mode - Ascend A2/A3 Series Products Supported - Ascend 950PR/DT Series Products Supported + A2/A3 Series Supported + 950PR/DT Series Supported Trigger diff --git a/docs/docs/hardware-platforms/ascend-npus/reference/support_features.mdx b/docs/docs/hardware-platforms/ascend-npus/reference/support_features.mdx index d22553c894..b20ccbfd8c 100644 --- a/docs/docs/hardware-platforms/ascend-npus/reference/support_features.mdx +++ b/docs/docs/hardware-platforms/ascend-npus/reference/support_features.mdx @@ -2100,12 +2100,6 @@ If the value is int8, you must also set the environment variable:DEEP_NORMAL_MOD bool flag
(set to enable) A2/A3 Series - - `--cuda-graph-bs-decode` - `None` - List[int] - A2/A3 Series - `--disable-cuda-graph` `False` @@ -2244,30 +2238,12 @@ If the value is int8, you must also set the environment variable:DEEP_NORMAL_MOD bool flag
(set to enable) A2/A3 Series - - `--cuda-graph-bs-prefill` - `None` - Type: JSON
list - A2/A3 Series - - - `--cuda-graph-tc-compiler` - `eager` - `eager` - A2/A3 Series - `--torch-compile-max-bs` `32` Type: int A2/A3 Series - - `--cuda-graph-max-bs-prefill` - `None` - Type: int - A2/A3 Series - `--enable-p2p-check` `False`