chore: Pump llama.cpp version

CameronNg · CameronNg · commit 90df75a298fb · 2024-04-02T14:47:20.000+07:00
diff --git a/context/llama_server_context.h b/context/llama_server_context.h
@@ -1,3 +1,5 @@
+#include <condition_variable>
+#include <atomic>
 #include <string>
 #include <vector>
 #include <set>
@@ -1525,7 +1527,7 @@ struct llama_server_context {
             slot.id, slot.params.n_keep, n_left, n_discard);
         llama_kv_cache_seq_rm(ctx, slot.id, slot.params.n_keep + 1,
                               slot.params.n_keep + n_discard + 1);
-        llama_kv_cache_seq_shift(ctx, slot.id,
+        llama_kv_cache_seq_add(ctx, slot.id,
                                  slot.params.n_keep + 1 + n_discard,
                                  slot.n_past, -n_discard);
 
diff --git a/llama.cpp b/llama.cpp
@@ -1 +1 @@
-Subproject commit 15499eb94227401bdc8875da6eb85c15d37068f7
+Subproject commit 33a52448061cfd2ea44da9e6cb30b2ec22e2f6d0
diff --git a/whisper.cpp b/whisper.cpp
@@ -1 +1 @@
-Subproject commit a0ddd8392c3427e833d893dc79b9b50f43cf8c9e
+Subproject commit ac283dbce7d42735e3ed985329037bf23fe180aa
