[router] remove all tokenizer metrics for performance (#9474)

This commit is contained in:
Chang Su
2025-08-21 18:35:24 -07:00
committed by GitHub
parent 9708d353b7
commit 53e2cd464f
8 changed files with 28 additions and 117 deletions

View File

@@ -36,24 +36,21 @@ pub enum Encoding {
}
impl Encoding {
/// Returns a reference to token IDs when possible, owned Vec for compatibility
pub fn token_ids(&self) -> Vec<TokenIdType> {
match self {
Encoding::Hf(inner) => inner.get_ids().to_vec(),
Encoding::Sp(inner) => inner.clone(),
Encoding::Tiktoken(inner) => inner.clone(),
}
}
/// Returns a reference to token IDs where possible
pub fn token_ids_ref(&self) -> &[TokenIdType] {
/// Returns a reference to token IDs - zero-copy operation
pub fn token_ids(&self) -> &[TokenIdType] {
match self {
Encoding::Hf(inner) => inner.get_ids(),
Encoding::Sp(inner) => inner,
Encoding::Tiktoken(inner) => inner, // Now works with tiktoken-rs 0.7.0!
Encoding::Tiktoken(inner) => inner,
}
}
/// Deprecated: Use token_ids() instead (kept for compatibility)
#[deprecated(since = "0.1.0", note = "Use token_ids() instead")]
pub fn token_ids_ref(&self) -> &[TokenIdType] {
self.token_ids()
}
/// Get a hash of the token IDs for caching purposes
pub fn get_hash(&self) -> u64 {
let mut hasher = DefaultHasher::new();