vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

llama.android : update to new API

Browse source 
ggml-ci
This commit is contained in:
Georgi Gerganov 2025-01-09 16:03:09 +02:00
parent 330bd07b82
commit d8931a701c
No known key found for this signature in database
GPG key ID: 449E073F9DC10735
4 changed files with 10 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

4
examples/batched.swift/Sources/main.swift
View file

@ -23,12 +23,12 @@ defer {
} }
let model_params = llama_model_default_params() let model_params = llama_model_default_params()
guard let model = llama_load_model_from_file(modelPath.cString(using: .utf8), model_params) else { guard let model = llama_model_load_from_file(modelPath.cString(using: .utf8), model_params) else {
print("Failed to load model") print("Failed to load model")
exit(1) exit(1)
} }
defer { defer {
llama_free_model(model) llama_model_free(model)
} }
var tokens = tokenize(text: prompt, add_bos: true) var tokens = tokenize(text: prompt, add_bos: true)

7
examples/llama.android/llama/src/main/cpp/llama-android.cpp
View file

@ -87,7 +87,7 @@ Java_android_llama_cpp_LLamaAndroid_load_1model(JNIEnv *env, jobject, jstring fi
auto path_to_model = env->GetStringUTFChars(filename, 0); auto path_to_model = env->GetStringUTFChars(filename, 0);
LOGi("Loading model from %s", path_to_model); LOGi("Loading model from %s", path_to_model);
auto model = llama_load_model_from_file(path_to_model, model_params); auto model = llama_model_load_from_file(path_to_model, model_params);
env->ReleaseStringUTFChars(filename, path_to_model); env->ReleaseStringUTFChars(filename, path_to_model);
if (!model) { if (!model) {
@ -102,7 +102,7 @@ Java_android_llama_cpp_LLamaAndroid_load_1model(JNIEnv *env, jobject, jstring fi
extern "C" extern "C"
JNIEXPORT void JNICALL JNIEXPORT void JNICALL
Java_android_llama_cpp_LLamaAndroid_free_1model(JNIEnv *, jobject, jlong model) { Java_android_llama_cpp_LLamaAndroid_free_1model(JNIEnv *, jobject, jlong model) {
llama_free_model(reinterpret_cast<llama_model *>(model)); llama_model_free(reinterpret_cast<llama_model *>(model));
} }
extern "C" extern "C"
@ -405,6 +405,7 @@ Java_android_llama_cpp_LLamaAndroid_completion_1loop(
const auto batch = reinterpret_cast<llama_batch *>(batch_pointer); const auto batch = reinterpret_cast<llama_batch *>(batch_pointer);
const auto sampler = reinterpret_cast<llama_sampler *>(sampler_pointer); const auto sampler = reinterpret_cast<llama_sampler *>(sampler_pointer);
const auto model = llama_get_model(context); const auto model = llama_get_model(context);
const auto vocab = llama_get_vocab(model);
if (!la_int_var) la_int_var = env->GetObjectClass(intvar_ncur); if (!la_int_var) la_int_var = env->GetObjectClass(intvar_ncur);
if (!la_int_var_value) la_int_var_value = env->GetMethodID(la_int_var, "getValue", "()I"); if (!la_int_var_value) la_int_var_value = env->GetMethodID(la_int_var, "getValue", "()I");
@ -414,7 +415,7 @@ Java_android_llama_cpp_LLamaAndroid_completion_1loop(
const auto new_token_id = llama_sampler_sample(sampler, context, -1); const auto new_token_id = llama_sampler_sample(sampler, context, -1);
const auto n_cur = env->CallIntMethod(intvar_ncur, la_int_var_value); const auto n_cur = env->CallIntMethod(intvar_ncur, la_int_var_value);
if (llama_token_is_eog(model, new_token_id) || n_cur == n_len) { if (llama_token_is_eog(vocab, new_token_id) || n_cur == n_len) {
return nullptr; return nullptr;
} }

4
examples/llama.swiftui/llama.cpp.swift/LibLlama.swift
View file

@ -52,8 +52,8 @@ actor LlamaContext {
deinit { deinit {
llama_sampler_free(sampling) llama_sampler_free(sampling)
llama_batch_free(batch) llama_batch_free(batch)
llama_model_free(model)
llama_free(context) llama_free(context)
llama_free_model(model)
llama_backend_free() llama_backend_free()
} }
@ -65,7 +65,7 @@ actor LlamaContext {
model_params.n_gpu_layers = 0 model_params.n_gpu_layers = 0
print("Running on simulator, force use n_gpu_layers = 0") print("Running on simulator, force use n_gpu_layers = 0")
#endif #endif
let model = llama_load_model_from_file(path, model_params) let model = llama_model_load_from_file(path, model_params)
guard let model else { guard let model else {
print("Could not load model at \(path)") print("Could not load model at \(path)")
throw LlamaError.couldNotInitializeContext throw LlamaError.couldNotInitializeContext

4
tests/test-tokenizer-random.py
View file

@ -76,7 +76,7 @@ class LibLlamaModel:
self.ffi = libllama.ffi self.ffi = libllama.ffi
if isinstance(mparams, dict): if isinstance(mparams, dict):
mparams = libllama.model_default_params(**mparams) mparams = libllama.model_default_params(**mparams)
self.model = self.lib.llama_load_model_from_file(path_model.encode(), mparams) self.model = self.lib.llama_model_load_from_file(path_model.encode(), mparams)
if not self.model: if not self.model:
raise RuntimeError("error: failed to load model '%s'" % path_model) raise RuntimeError("error: failed to load model '%s'" % path_model)
if isinstance(cparams, dict): if isinstance(cparams, dict):
@ -92,7 +92,7 @@ class LibLlamaModel:
if self.ctx: if self.ctx:
self.lib.llama_free(self.ctx) self.lib.llama_free(self.ctx)
if self.model: if self.model:
self.lib.llama_free_model(self.model) self.lib.llama_model_free(self.model)
self.ctx = None self.ctx = None
self.model = None self.model = None
self.lib = None self.lib = None