diff options
| author | Sigbjørn Skjæret <sigbjorn.skjaeret@scala.com> | 2024-06-18 14:19:45 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-06-18 22:19:45 +1000 |
| commit | 91c188d6c296bd3384f2a02a83b71187aa3d18b3 (patch) | |
| tree | 93e5f4aa20756a15ac8d86394c811736b8b05fb1 /examples/server/server.cpp | |
| parent | 84f6de17f6a8602e7ff7f7c7bda36a73f510a2dd (diff) | |
Only use FIM middle token if it exists (#7648)
* Only use FIM middle if it exists
* Only use FIM middle if it exists
Diffstat (limited to 'examples/server/server.cpp')
| -rw-r--r-- | examples/server/server.cpp | 7 |
1 files changed, 6 insertions, 1 deletions
diff --git a/examples/server/server.cpp b/examples/server/server.cpp index 919078f2..ec59307b 100644 --- a/examples/server/server.cpp +++ b/examples/server/server.cpp @@ -2038,7 +2038,12 @@ struct server_context { prefix_tokens.insert(prefix_tokens.begin(), llama_token_bos(model)); // always add BOS prefix_tokens.insert(prefix_tokens.end(), llama_token_suffix(model)); prefix_tokens.insert(prefix_tokens.end(), suffix_tokens.begin(), suffix_tokens.end()); - prefix_tokens.push_back(llama_token_middle(model)); + + const llama_token middle_token = llama_token_middle(model); + if (middle_token >= 0) { + prefix_tokens.push_back(middle_token); + } + prompt_tokens = prefix_tokens; } else { prompt_tokens = tokenize(slot.prompt, system_prompt.empty()); // add BOS if there isn't system prompt |