CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-02 09:12:03 +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

2
examples/batched/batched.cpp
View File

@@ -138,7 +138,7 @@ int main(int argc, char ** argv) {
//// assign the system KV cache to all parallel sequences
//// this way, the parallel sequences will "reuse" the prompt tokens without having to copy them
//for (int32_t i = 1; i < n_parallel; ++i) {
// llama_past_seq_cp(ctx, 0, i, -1, -1);
// llama_kv_cache_seq_cp(ctx, 0, i, -1, -1);
//}
if (n_parallel > 1) {