Merge pull request #583 from SKaiNET-developers/release/0.22.1

Prepare 0.22.1

Author: michalharakal

CHANGELOG.md

@@ -2,6 +2,12 @@
 
 ## [Unreleased]
 
+## [0.22.1] - 2026-04-30
+
+### Added
+
+- **`StreamingShardedSafeTensorsReader.loadTensorStorageMapped`** — by-name and by-`ShardedTensorInfo` overloads that mirror the existing single-file `StreamingSafeTensorsReader.loadTensorStorageMapped(tensor, filePath)`. Both return a `TensorStorage` whose `BufferHandle.FileBacked` references the resolved shard file's tensor byte range, enabling zero-copy / memory-mapped reads of tensors that exceed the 2 GB JVM `ByteArray` limit. The new methods delegate internally to the per-shard reader; callers don't need to know which physical shard contains a given tensor. Unblocks downstream consumers (e.g. SKaiNET-transformers' Gemma 4 PLE token-embedding table at ~4.7 GB BF16 on E2B) that previously rolled their own `FileChannel.map`. (PR #582)
+
 ## [0.22.0] - 2026-04-30
 
 ### Added

README.md

@@ -19,8 +19,8 @@ Add the core dependencies (Gradle Kotlin DSL):
 
 ```kotlin
 dependencies {
-    implementation("sk.ainet.core:SKaiNET-lang-core:0.22.0")
-    implementation("sk.ainet.core:SKaiNET-backend-cpu:0.22.0")
+    implementation("sk.ainet.core:SKaiNET-lang-core:0.22.1")
+    implementation("sk.ainet.core:SKaiNET-backend-cpu:0.22.1")
 }
 ```
 

gradle.properties

@@ -1,5 +1,5 @@
 GROUP=sk.ainet.core
-VERSION_NAME=0.22.0
+VERSION_NAME=0.22.1
 POM_DESCRIPTION=SKaiNET
 
 POM_URL=https://github.com/SKaiNET-developers/skainet/