RubyGems - llama_cpp - Versions diffs - 0.14.2 → 0.14.4 - Mend

llama_cpp 0.14.2 → 0.14.4

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.

Files changed (33) hide show

checksums.yaml +4 -4
data/CHANGELOG.md +14 -0
data/ext/llama_cpp/llama_cpp.cpp +64 -0
data/lib/llama_cpp/version.rb +2 -2
data/sig/llama_cpp.rbs +6 -0
data/vendor/tmp/llama.cpp/Makefile +91 -21
data/vendor/tmp/llama.cpp/ggml-alloc.c +14 -5
data/vendor/tmp/llama.cpp/ggml-backend-impl.h +5 -0
data/vendor/tmp/llama.cpp/ggml-backend.c +155 -125
data/vendor/tmp/llama.cpp/ggml-backend.h +4 -4
data/vendor/tmp/llama.cpp/ggml-common.h +25 -2
data/vendor/tmp/llama.cpp/ggml-cuda.cu +1779 -10762
data/vendor/tmp/llama.cpp/ggml-cuda.h +6 -15
data/vendor/tmp/llama.cpp/ggml-kompute.cpp +5 -0
data/vendor/tmp/llama.cpp/ggml-metal.m +167 -124
data/vendor/tmp/llama.cpp/ggml-metal.metal +603 -303
data/vendor/tmp/llama.cpp/ggml-opencl.cpp +5 -0
data/vendor/tmp/llama.cpp/ggml-quants.c +663 -56
data/vendor/tmp/llama.cpp/ggml-quants.h +3 -0
data/vendor/tmp/llama.cpp/ggml-sycl.cpp +341 -469
data/vendor/tmp/llama.cpp/ggml-sycl.h +19 -4
data/vendor/tmp/llama.cpp/ggml-vulkan-shaders.hpp +37199 -14939
data/vendor/tmp/llama.cpp/ggml-vulkan.cpp +335 -307
data/vendor/tmp/llama.cpp/ggml-vulkan.h +0 -11
data/vendor/tmp/llama.cpp/ggml.c +229 -107
data/vendor/tmp/llama.cpp/ggml.h +11 -5
data/vendor/tmp/llama.cpp/llama.cpp +2136 -464
data/vendor/tmp/llama.cpp/llama.h +86 -23
data/vendor/tmp/llama.cpp/unicode-data.cpp +1651 -0
data/vendor/tmp/llama.cpp/unicode-data.h +16 -0
data/vendor/tmp/llama.cpp/unicode.cpp +8 -1403
data/vendor/tmp/llama.cpp/unicode.h +2 -0
metadata +5 -3

data/vendor/tmp/llama.cpp/ggml-opencl.cpp CHANGED Viewed

@@ -2234,6 +2234,11 @@ static ggml_backend_buffer_type_t ggml_backend_opencl_get_default_buffer_type(gg
 static ggml_status ggml_backend_opencl_graph_compute(ggml_backend_t backend, ggml_cgraph * graph) {
     for (int i = 0; i < graph->n_nodes; ++i) {
         ggml_tensor * node = graph->nodes[i];
+        if (ggml_is_empty(node)) {
+            continue;
+        }
         switch (node->op) {
             case GGML_OP_MUL_MAT:
                 ggml_cl_mul_mat(node->src[0], node->src[1], node, nullptr, 0);