@realtimex/node-llama-cpp 0.84.0 → 0.86.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
|
@@ -397,7 +397,7 @@ AddonContext::AddonContext(const Napi::CallbackInfo& info) : Napi::ObjectWrap<Ad
|
|
|
397
397
|
|
|
398
398
|
context_params = llama_context_default_params();
|
|
399
399
|
context_params.n_ctx = 4096;
|
|
400
|
-
context_params.n_threads = std::max(
|
|
400
|
+
context_params.n_threads = std::max(common_cpu_get_num_math(), 1);
|
|
401
401
|
context_params.n_threads_batch = context_params.n_threads;
|
|
402
402
|
context_params.no_perf = true;
|
|
403
403
|
context_params.swa_full = false;
|
|
@@ -745,7 +745,7 @@ Napi::Value AddonContext::SetThreads(const Napi::CallbackInfo& info) {
|
|
|
745
745
|
|
|
746
746
|
const auto threads = info[0].As<Napi::Number>().Int32Value();
|
|
747
747
|
const auto resolvedThreads = threads == 0
|
|
748
|
-
? std::max((int32_t)std::thread::hardware_concurrency(), std::max(
|
|
748
|
+
? std::max((int32_t)std::thread::hardware_concurrency(), std::max(common_cpu_get_num_math(), 1))
|
|
749
749
|
: threads;
|
|
750
750
|
|
|
751
751
|
if (llama_n_threads(ctx) != resolvedThreads) {
|
package/llama/addon/addon.cpp
CHANGED
|
@@ -51,7 +51,7 @@ Napi::Value addonGetSupportsMlock(const Napi::CallbackInfo& info) {
|
|
|
51
51
|
}
|
|
52
52
|
|
|
53
53
|
Napi::Value addonGetMathCores(const Napi::CallbackInfo& info) {
|
|
54
|
-
return Napi::Number::New(info.Env(),
|
|
54
|
+
return Napi::Number::New(info.Env(), common_cpu_get_num_math());
|
|
55
55
|
}
|
|
56
56
|
|
|
57
57
|
Napi::Value addonGetBlockSizeForGgmlType(const Napi::CallbackInfo& info) {
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@realtimex/node-llama-cpp",
|
|
3
|
-
"version": "0.
|
|
3
|
+
"version": "0.86.0",
|
|
4
4
|
"description": "Run AI models locally on your machine with node.js bindings for llama.cpp. Enforce a JSON schema on the model output on the generation level",
|
|
5
5
|
"main": "./dist/index.js",
|
|
6
6
|
"type": "module",
|
|
@@ -233,19 +233,19 @@
|
|
|
233
233
|
}
|
|
234
234
|
},
|
|
235
235
|
"optionalDependencies": {
|
|
236
|
-
"@realtimex/node-llama-cpp-linux-arm64": "0.
|
|
237
|
-
"@realtimex/node-llama-cpp-linux-armv7l": "0.
|
|
238
|
-
"@realtimex/node-llama-cpp-linux-x64": "0.
|
|
239
|
-
"@realtimex/node-llama-cpp-linux-x64-cuda": "0.
|
|
240
|
-
"@realtimex/node-llama-cpp-linux-x64-cuda-ext": "0.
|
|
241
|
-
"@realtimex/node-llama-cpp-linux-x64-vulkan": "0.
|
|
242
|
-
"@realtimex/node-llama-cpp-mac-arm64-metal": "0.
|
|
243
|
-
"@realtimex/node-llama-cpp-mac-x64": "0.
|
|
244
|
-
"@realtimex/node-llama-cpp-win-arm64": "0.
|
|
245
|
-
"@realtimex/node-llama-cpp-win-x64": "0.
|
|
246
|
-
"@realtimex/node-llama-cpp-win-x64-cuda": "0.
|
|
247
|
-
"@realtimex/node-llama-cpp-win-x64-cuda-ext": "0.
|
|
248
|
-
"@realtimex/node-llama-cpp-win-x64-vulkan": "0.
|
|
236
|
+
"@realtimex/node-llama-cpp-linux-arm64": "0.86.0",
|
|
237
|
+
"@realtimex/node-llama-cpp-linux-armv7l": "0.86.0",
|
|
238
|
+
"@realtimex/node-llama-cpp-linux-x64": "0.86.0",
|
|
239
|
+
"@realtimex/node-llama-cpp-linux-x64-cuda": "0.86.0",
|
|
240
|
+
"@realtimex/node-llama-cpp-linux-x64-cuda-ext": "0.86.0",
|
|
241
|
+
"@realtimex/node-llama-cpp-linux-x64-vulkan": "0.86.0",
|
|
242
|
+
"@realtimex/node-llama-cpp-mac-arm64-metal": "0.86.0",
|
|
243
|
+
"@realtimex/node-llama-cpp-mac-x64": "0.86.0",
|
|
244
|
+
"@realtimex/node-llama-cpp-win-arm64": "0.86.0",
|
|
245
|
+
"@realtimex/node-llama-cpp-win-x64": "0.86.0",
|
|
246
|
+
"@realtimex/node-llama-cpp-win-x64-cuda": "0.86.0",
|
|
247
|
+
"@realtimex/node-llama-cpp-win-x64-cuda-ext": "0.86.0",
|
|
248
|
+
"@realtimex/node-llama-cpp-win-x64-vulkan": "0.86.0"
|
|
249
249
|
},
|
|
250
250
|
"publishConfig": {
|
|
251
251
|
"access": "public"
|