Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-11-22 16:27:58 +01:00
Code Issues Packages Projects Releases Wiki Activity

Updated Templates supported by llama_chat_apply_template (markdown)

Xuan Son Nguyen 2024-03-04 12:25:23 +01:00
parent 86f794b1f9
commit 26c482f18e
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
Templates-supported-by-llama_chat_apply_template.md

@ -1,6 +1,6 @@
The `llama_chat_apply_template()` was added in [#5538](https://github.com/ggerganov/llama.cpp/pull/5538), which allows developers to format the chat into text prompt. By default, this function takes the template stored inside model's metadata `tokenizer.chat_template`.
Please not that, we do **not** include a jinja parser in llama.cpp due to its complexity. Our implementation works by matching the supplied template with a list of pre-defined templates hard-coded inside the function.
NOTE: We do **not** include a jinja parser in llama.cpp due to its complexity. Our implementation works by matching the supplied template with a list of pre-defined templates hard-coded inside the function.
This is the list of templates currently supported by `llama_apply_chat_template`. If you found another template on huggingface that's not yet supported by llama.cpp, please feel free to open an issue: