From e44e65d61c9a417caf40abb5298aa6c7f0e853a2 Mon Sep 17 00:00:00 2001
From: tongyu <119610311+tongyu0924@users.noreply.github.com>
Date: Sun, 27 Apr 2025 04:24:35 +0800
Subject: [PATCH] Update README.md

---
 colossalai/shardformer/README.md | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/colossalai/shardformer/README.md b/colossalai/shardformer/README.md
index 47ef98ccf..e41ad308a 100644
--- a/colossalai/shardformer/README.md
+++ b/colossalai/shardformer/README.md
@@ -83,6 +83,12 @@ Following are the description `ShardConfig`'s arguments:
 
 - `extra_kwargs`: A dict to store extra kwargs for ShardFormer.
 
+- - `tensor_parallel_mode`: A choice of parallel modes for tensor operations. Supported modes are:
+  - `'1d'`: 1D tensor parallelism
+  - `'2d'`: 2D tensor parallelism
+  - `'2.5d'`: 2.5D tensor parallelism
+  - `'3d'`: 3D tensor parallelism
+
 ### Write your own policy
 
 If you have a custom model, you can also use Shardformer to parallelize it by writing your own sharding policy. More information about the sharding policy can be found in [API Design](#-api-design).