format

Author: lxd-cumt

README.rst

@@ -7,27 +7,52 @@
 
 
 **TransformerEngine-FL is a fork of TransformerEngine that introduces a plugin-based architecture for supporting diverse AI chips, built on top of** `FlagOS <https://github.com/flagos-ai>`_, **a unified open-source AI system software stack.**
+
 The following table summarizes the specific adaptation status of multiple chips to the three backends of TransformerEngine-FL (FlagOS Triton, Vendor, and Reference Torch).
 
-+----------------------+-------------------+-------------------+-------------------+
-| Vendor               | FlagOS (Triton)   | Reference (Torch) | Vendor            |
-+======================+===================+===================+===================+
-| NVIDIA GPU           |        ✅         |        ✅          |        ✅         |
-+----------------------+-------------------+-------------------+-------------------+
-| Metax                |        ✅         |        ✅          |        ✅         |
-+----------------------+-------------------+-------------------+-------------------+
-| Moore Threads (MUSA) |        ✅         |        ✅          |        ✅         |
-+----------------------+-------------------+-------------------+-------------------+
-| Hygon                |        ✅         |        ✅          |        ✅         |
-+----------------------+-------------------+-------------------+-------------------+
-| Iluvatar             |        ✅         |        ✅          |        ✅         |
-+----------------------+-------------------+-------------------+-------------------+
-| Kunlunxin            |        ✅         |        ✅          |                   |
-+----------------------+-------------------+-------------------+-------------------+
-| Cambricon            |        ✅         |        ✅          |                   |
-+----------------------+-------------------+-------------------+-------------------+
-| Tsingmicro (TXDA)    |        ✅         |        ✅          |                   |
-+----------------------+-------------------+-------------------+-------------------+
+**TE-FL Backends**
+
+.. list-table:: TE-FL backend support by vendor
+   :header-rows: 1
+   :widths: 28 18 20 18
+   :align: left
+
+   * - Vendor
+     - FlagOS (Triton)
+     - Reference (Torch)
+     - Vendor
+   * - NVIDIA GPU
+     - ✅
+     - ✅
+     - ✅
+   * - Metax
+     - ✅
+     - ✅
+     - ✅
+   * - Moore Threads (MUSA)
+     - ✅
+     - ✅
+     - ✅
+   * - Hygon
+     - ✅
+     - ✅
+     - ✅
+   * - Iluvatar
+     - ✅
+     - ✅
+     - ✅
+   * - Kunlunxin
+     - ✅
+     - ✅
+     - —
+   * - Cambricon
+     - ✅
+     - ✅
+     - —
+   * - Tsingmicro (TXDA)
+     - ✅
+     - ✅
+     - —
 
 
 Transformer Engine