cui-llama.rn 0.2.0 → 1.0.1

package/android/src/main/java/com/rnllama/LlamaContext.java

@@ -53,6 +53,8 @@ public class LlamaContext {
       params.hasKey("use_mlock") ? params.getBoolean("use_mlock") : true,
       // boolean use_mmap,
       params.hasKey("use_mmap") ? params.getBoolean("use_mmap") : true,
+      //boolean vocab_only,
+      params.hasKey("vocab_only") ? params.getBoolean("vocab_only") : false,
       // String lora,
       params.hasKey("lora") ? params.getString("lora") : "",
       // float lora_scaled,
@@ -297,6 +299,7 @@ public class LlamaContext {
     int n_gpu_layers, // TODO: Support this
     boolean use_mlock,
     boolean use_mmap,
+    boolean vocab_only,
     String lora,
     float lora_scaled,
     String lora_base,

package/android/src/main/java/com/rnllama/RNLlama.java

@@ -221,7 +221,7 @@ public class RNLlama implements LifecycleEventListener {
     tasks.put(task, "stopCompletion-" + contextId);
   }
 
-  public void tokenize(double id, final String text, final Promise promise) {
+  public void tokenizeAsync(double id, final String text, final Promise promise) {
     final int contextId = (int) id;
     AsyncTask task = new AsyncTask<Void, Void, WritableMap>() {
       private Exception exception;
@@ -253,6 +253,16 @@ public class RNLlama implements LifecycleEventListener {
     tasks.put(task, "tokenize-" + contextId);
   }
 
+  public WritableMap tokenizeSync(double id, final String text) {
+    int contextId = (int) id;
+    LlamaContext context = contexts.get(contextId);
+    if (context == null) {
+      return Arguments.createMap();
+    }
+    return context.tokenize(text);
+  }
+
+
   public void detokenize(double id, final ReadableArray tokens, final Promise promise) {
     final int contextId = (int) id;
     AsyncTask task = new AsyncTask<Void, Void, String>() {

package/android/src/main/jni.cpp

@@ -129,6 +129,7 @@ Java_com_rnllama_LlamaContext_initContext(
     jint n_gpu_layers, // TODO: Support this
     jboolean use_mlock,
     jboolean use_mmap,
+    jboolean vocab_only,
     jstring lora_str,
     jfloat lora_scaled,
     jstring lora_base_str,
@@ -139,6 +140,11 @@ Java_com_rnllama_LlamaContext_initContext(
 
     gpt_params defaultParams;
 
+    defaultParams.vocab_only = vocab_only;
+    if(vocab_only) {
+        defaultParams.warmup = false;
+    }
+
     const char *model_path_chars = env->GetStringUTFChars(model_path_str, nullptr);
     defaultParams.model = model_path_chars;
 

package/android/src/newarch/java/com/rnllama/RNLlamaModule.java

@@ -6,6 +6,7 @@ import com.facebook.react.bridge.Promise;
 import com.facebook.react.bridge.ReactApplicationContext;
 import com.facebook.react.bridge.ReactMethod;
 import com.facebook.react.bridge.ReadableMap;
+import com.facebook.react.bridge.WritableMap;
 import com.facebook.react.bridge.ReadableArray;
 import com.facebook.react.module.annotations.ReactModule;
 
@@ -63,8 +64,13 @@ public class RNLlamaModule extends NativeRNLlamaSpec {
   }
 
   @ReactMethod
-  public void tokenize(double id, final String text, final Promise promise) {
-    rnllama.tokenize(id, text, promise);
+  public void tokenizeAsync(double id, final String text, final Promise promise) {
+    rnllama.tokenizeAsync(id, text, promise);
+  }
+
+  @ReactMethod(isBlockingSynchronousMethod=true)
+  public WritableMap tokenizeSync(double id, final String text) {
+    return rnllama.tokenizeSync(id, text);
   }
 
   @ReactMethod

package/android/src/oldarch/java/com/rnllama/RNLlamaModule.java

@@ -7,6 +7,7 @@ import com.facebook.react.bridge.ReactApplicationContext;
 import com.facebook.react.bridge.ReactContextBaseJavaModule;
 import com.facebook.react.bridge.ReactMethod;
 import com.facebook.react.bridge.ReadableMap;
+import com.facebook.react.bridge.WritableMap;
 import com.facebook.react.bridge.ReadableArray;
 import com.facebook.react.module.annotations.ReactModule;
 
@@ -64,12 +65,17 @@ public class RNLlamaModule extends ReactContextBaseJavaModule {
   }
 
   @ReactMethod
-  public void tokenize(double id, final String text, final Promise promise) {
-    rnllama.tokenize(id, text, promise);
+  public void tokenizeAsync(double id, final String text, final Promise promise) {
+    rnllama.tokenizeAsync(id, text, promise);
+  }
+
+  @ReactMethod(isBlockingSynchronousMethod=true)
+  public WritableMap tokenizeSync(double id, final String text) {
+    return rnllama.tokenizeSync(id, text);
   }
 
   @ReactMethod
-  public void detokenize(double id, final ReadableArray tokens, final Promise promise) {
+  public void detokenize(double id, final ReadableArray tokens, final Promise promise) {  
     rnllama.detokenize(id, tokens, promise);
   }