Skip to content

Add gate to disable control token decoding #856

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Add gate to disable control token decoding #856

Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
d101cfa
use environment variable to disable control token decoding
ju6ge Nov 3, 2025
9459a28
instead of an env variable use `special` parameter
ju6ge Nov 4, 2025
6ac0079
introduce new variant to `Special` enum
ju6ge Nov 4, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 4 additions & 1 deletion llama-cpp-2/src/model.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -395,8 +395,11 @@ impl LlamaModel {
if attrs.is_empty()
|| attrs
.intersects(LlamaTokenAttr::Unknown | LlamaTokenAttr::Byte | LlamaTokenAttr::Unused)
// the following exclusion of control characters stems from a requirement of the original purpose of this project see
// https://github.com/utilityai/llama-cpp-rs/issues/826#issuecomment-3478624072. But it should not be the default behavior
// so this feature is now gated through the `LLAMA_RS_FORBID_CTRL_TOKEN_DECODE` environment variable
|| attrs.contains(LlamaTokenAttr::Control)
&& (token == self.token_bos() || token == self.token_eos())
&& (token == self.token_bos() || token == self.token_eos()) && std::env::var("LLAMA_RS_FORBID_CTRL_TOKEN_DECODE").is_ok_and(|v| v.parse::<bool>().is_ok_and(|v| v))
{
return Ok(Vec::new());
}
Expand Down
Loading