vrr/koboldcpp
2
0
Fork 0
mirror of https://github.com/LostRuins/koboldcpp.git synced 2025-09-11 01:24:36 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'master' into concedo_experimental

Browse source 
# Conflicts:
#	.github/workflows/build.yml
#	CMakeLists.txt
#	Makefile
#	README-sycl.md
#	README.md
#	flake.lock
#	scripts/sync-ggml-am.sh
#	scripts/sync-ggml.last
#	scripts/sync-ggml.sh
#	tests/.gitignore
#	tests/test-backend-ops.cpp
This commit is contained in:
Concedo 2024-03-11 23:00:47 +08:00
commit 6a32c14e86
41 changed files with 4274 additions and 4304 deletions
Download patch file Download diff file Expand all files Collapse all files

14
examples/embedding/embedding.cpp
View file

@ -24,17 +24,6 @@ static void batch_add_seq(llama_batch & batch, const std::vector<int32_t> & toke
}
}
static void normalize(const float * vec, float * out, int n) {
float norm = 0;
for (int i = 0; i < n; i++) {
norm += vec[i] * vec[i];
}
norm = sqrt(norm);
for (int i = 0; i < n; i++) {
out[i] = vec[i] / norm;
}
}
static void batch_decode(llama_context * ctx, llama_batch & batch, float * output, int n_seq, int n_embd) {
// clear previous kv_cache values (irrelevant for embeddings)
llama_kv_cache_clear(ctx);
@ -45,7 +34,6 @@ static void batch_decode(llama_context * ctx, llama_batch & batch, float * outpu
fprintf(stderr, "%s : failed to decode\n", __func__);
}
// normalize on copy
for (int i = 0; i < batch.n_tokens; i++) {
if (!batch.logits[i]) {
continue;
@ -62,7 +50,7 @@ static void batch_decode(llama_context * ctx, llama_batch & batch, float * outpu
}
float * out = output + batch.seq_id[i][0] * n_embd;
normalize(embd, out, n_embd);
llama_embd_normalize(embd, out, n_embd);
}
}