fix: chat completions should use empty stop words (#626)

TabbyML · Oct 24, 2023 · 4c6f4b1 · 4c6f4b1
1 parent 2a40d36
commit 4c6f4b1
Show file tree

Hide file tree

Showing 2 changed files with 19 additions and 0 deletions.
diff --git a/crates/tabby-common/src/languages.rs b/crates/tabby-common/src/languages.rs
@@ -36,6 +36,11 @@ pub struct Language {
 
 impl Language {
     pub fn get_stop_words(&self) -> Vec<String> {
+        // Special handling for empty languages - returns empty stop words.
+        if self.get_hashkey() == "empty" {
+            return vec![];
+        }
+
         let mut out = vec![];
         out.push(format!("\n{}", self.line_comment));
         for word in &self.top_level_keywords {
@@ -62,6 +67,11 @@ lazy_static! {
         line_comment: "".to_owned(),
         top_level_keywords: vec![],
     };
+    pub static ref EMPTY_LANGUAGE: Language = Language {
+        languages: vec!["empty".to_owned()],
+        line_comment: "".to_owned(),
+        top_level_keywords: vec![],
+    };
 }
 
 pub fn get_language(language: &str) -> &'static Language {
@@ -71,3 +81,10 @@ pub fn get_language(language: &str) -> &'static Language {
         .find(|c| c.languages.iter().any(|x| x == language))
         .unwrap_or(&UNKNOWN_LANGUAGE)
 }
+
+mod tests {
+    #[test]
+    fn test_empty_language() {
+        assert!(super::EMPTY_LANGUAGE.get_stop_words().is_empty())
+    }
+}
diff --git a/crates/tabby/src/serve/chat.rs b/crates/tabby/src/serve/chat.rs
@@ -11,6 +11,7 @@ use axum::{
 use axum_streams::StreamBodyAs;
 use prompt::ChatPromptBuilder;
 use serde::{Deserialize, Serialize};
+use tabby_common::languages::EMPTY_LANGUAGE;
 use tabby_inference::{TextGeneration, TextGenerationOptions, TextGenerationOptionsBuilder};
 use tracing::{debug, instrument};
 use utoipa::ToSchema;
@@ -88,6 +89,7 @@ fn parse_request(
     builder
         .max_input_length(2048)
         .max_decoding_length(1920)
+        .language(&EMPTY_LANGUAGE)
         .sampling_temperature(0.1);
 
     (