[tokenizers] Use tokenizers from rust.io (#3476)

deepjavalibrary · Sep 16, 2024 · bc31aff · bc31aff
1 parent 531b1f0
commit bc31aff
Show file tree

Hide file tree

Showing 4 changed files with 5 additions and 20 deletions.
diff --git a/extensions/tokenizers/build.cmd b/extensions/tokenizers/build.cmd
@@ -3,14 +3,6 @@
 @rem choco install rust -y
 @rem choco install jdk8 -y
 
-set VERSION=v"%1"
-
-if exist "tokenizers" (
-    echo Found "tokenizers"
-) else (
-    git clone https://github.com/huggingface/tokenizers -b %VERSION%
-)
-
 if exist build rd /q /s build
 md build\classes
 

diff --git a/extensions/tokenizers/build.gradle.kts b/extensions/tokenizers/build.gradle.kts
@@ -114,11 +114,11 @@ tasks {
             if ("mac" in os || "linux" in os) {
                 val arch = if (arch == "amd64") "x86_64" else arch
                 exec {
-                    commandLine("bash", "build.sh", libs.versions.tokenizers.get(), arch, flavor)
+                    commandLine("bash", "build.sh", arch, flavor)
                 }
             } else
                 exec {
-                    commandLine("${project.projectDir}/build.cmd", libs.versions.tokenizers.get())
+                    commandLine("${project.projectDir}/build.cmd")
                 }
 
             // for ci to upload to S3

diff --git a/extensions/tokenizers/build.sh b/extensions/tokenizers/build.sh
@@ -1,17 +1,10 @@
 #!/usr/bin/env bash
 
 set -e
-WORK_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
 PLATFORM=$(uname | tr '[:upper:]' '[:lower:]')
 
-VERSION=v$1
-ARCH=$2
-FLAVOR=$3
-
-pushd "$WORK_DIR"
-if [ ! -d "tokenizers" ]; then
-  git clone https://github.com/huggingface/tokenizers -b "$VERSION"
-fi
+ARCH=$1
+FLAVOR=$2
 
 if [ ! -d "build" ]; then
   mkdir build

diff --git a/extensions/tokenizers/rust/Cargo.toml b/extensions/tokenizers/rust/Cargo.toml
@@ -13,7 +13,7 @@ candle-flash-attn = { version = "*", optional = true }
 candle-cublaslt = { git = "https://github.com/huggingface/candle-cublaslt", rev = "cf789b7dd6d4abb19b03b9556442f94f0588b4a0", optional = true }
 candle-layer-norm = { git = "https://github.com/xyang16/candle-layer-norm", rev = "e574de6a7f88bafbede8edf9ee43170c6a8ce51a", optional = true }
 candle-rotary = { git = "https://github.com/huggingface/candle-rotary", rev = "0a718a0856569a92f3112e64f10d07e4447822e8", optional = true }
-tokenizers = { path = "../tokenizers/tokenizers", version = "*", features = ["http"] }
+tokenizers = { version = "0.20.0", features = ["http"] }
 half = "2.4.0"
 tracing = "0.1.40"
 safetensors = "0.4.3"