diff options
author | Georgi Gerganov <ggerganov@gmail.com> | 2023-10-11 23:55:08 +0300 |
---|---|---|
committer | Georgi Gerganov <ggerganov@gmail.com> | 2023-10-11 23:55:41 +0300 |
commit | b8fe4b5cc9cb237ca98e5bc51b5d189e3c446d13 (patch) | |
tree | b606053e5a038daeda9d90dfa510db715bddec36 | |
parent | a8bdd65525ae86dea905e9866ad369b53e30ac14 (diff) |
main : fix session loading bug (#3400)
-rw-r--r-- | examples/main/main.cpp | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/examples/main/main.cpp b/examples/main/main.cpp index b39a67d9..55f73356 100644 --- a/examples/main/main.cpp +++ b/examples/main/main.cpp @@ -297,6 +297,9 @@ int main(int argc, char ** argv) { LOG_TEE("%s: session file matches %zu / %zu tokens of prompt\n", __func__, n_matching_session_tokens, embd_inp.size()); } + + // remove any "future" tokens that we might have inherited from the previous session + llama_kv_cache_tokens_rm(ctx, n_matching_session_tokens, -1); } LOGLN( @@ -545,9 +548,6 @@ int main(int argc, char ** argv) { if (i > 0) { embd.erase(embd.begin(), embd.begin() + i); } - - // remove any "future" tokens that we might have inherited from the session from the KV cache - llama_kv_cache_tokens_rm(ctx, n_past, -1); } // evaluate tokens in batches |