mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2025-10-31 08:51:55 +00:00
llama : remove implicit recurrent state rollbacks
This commit is contained in:
@@ -111,7 +111,7 @@ if llama_decode(context, batch) != 0 {
|
||||
}
|
||||
|
||||
for i in 1 ..< n_parallel {
|
||||
llama_past_seq_cp(context, 0, Int32(i), -1, -1)
|
||||
llama_kv_cache_seq_cp(context, 0, Int32(i), -1, -1)
|
||||
}
|
||||
|
||||
if n_parallel > 1 {
|
||||
|
||||
Reference in New Issue
Block a user