[Version] v1.7.2. (#454)

intel · Jun 18, 2024 · da2a7fa · da2a7fa
1 parent cc9675f
commit da2a7fa
Show file tree

Hide file tree

Showing 2 changed files with 12 additions and 2 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,5 +1,15 @@
 # CHANGELOG
-# [Version v1.7.0](https://github.com/intel/xFasterTransformer/releases/tag/v1.7.0)
+# [Version v1.7.2](https://github.com/intel/xFasterTransformer/releases/tag/v1.7.2)
+v1.7.2 - Continuous batching feature supports Qwen 1.0 & hybrid data types.
+
+## Functionality
+- Add continuous batching support of Qwen 1.0 models.
+- Enable hybrid data types for continuous batching feature, including `BF16_FP16, BF16_INT8, BF16_W8A8, BF16_INT4, BF16_NF4, W8A8_INT8, W8A8_int4, W8A8_NF4`.
+
+## BUG fix
+- Fixed the convert fault in Baichuan1 models.
+
+# [Version v1.7.1](https://github.com/intel/xFasterTransformer/releases/tag/v1.7.1)
 v1.7.1 - Continuous batching feature supports ChatGLM2/3.
 
 ## Functionality

diff --git a/VERSION b/VERSION
@@ -1 +1 @@
-1.7.1
+1.7.2