CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-09 10:17:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

llama : remove implicit recurrent state rollbacks

Browse Source
This commit is contained in:
Francis Couture-Harpin
2024-11-24 20:35:30 -05:00
parent 124c222f76
commit 8006f3b3c8
25 changed files with 411 additions and 1119 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
examples/batched-bench/batched-bench.cpp
View File

@@ -133,7 +133,7 @@ int main(int argc, char ** argv) {
const auto t_pp_start = ggml_time_us();
llama_past_clear(ctx);
llama_kv_cache_clear(ctx);
if (!decode_helper(ctx, batch, ctx_params.n_batch)) {
LOG_ERR("%s: llama_decode() failed\n", __func__);
@@ -142,7 +142,7 @@ int main(int argc, char ** argv) {
if (is_pp_shared) {
for (int32_t i = 1; i < pl; ++i) {
llama_past_seq_cp(ctx, 0, i, -1, -1);
llama_kv_cache_seq_cp(ctx, 0, i, -1, -1);
}
}