pub struct ExecutorPrefillAdmission<'a> {
pub request_id: &'a RequestId,
pub input_tokens: &'a [TokenId],
pub maximum_sequence_tokens: usize,
pub product_prompt_tokens: usize,
pub replayed_output_tokens: usize,
pub request_origin: ExecutorRequestOrigin,
}Expand description
Borrowed, already-tokenized input used to probe plan-runtime prefill admission before the request can enter a device submission batch.
This carries semantic token identity rather than an aggregate token count:
vNext derives the exact resource work shape and its fingerprint from this
boundary. The request remains owned by the scheduler while the executor
retains any admitted authority internally until ModelExecutor::prefill
consumes it or cancellation releases it.
Fields§
§request_id: &'a RequestId§input_tokens: &'a [TokenId]§maximum_sequence_tokens: usize§product_prompt_tokens: usizePrompt tokens reported in the final product usage.
replayed_output_tokens: usizeAlready-committed output tokens replayed as part of a recompute input.
request_origin: ExecutorRequestOriginImplementations§
Source§impl<'a> ExecutorPrefillAdmission<'a>
impl<'a> ExecutorPrefillAdmission<'a>
pub const fn for_startup( request_id: &'a RequestId, input_tokens: &'a [TokenId], maximum_sequence_tokens: usize, ) -> Self
pub const fn for_diagnostic( request_id: &'a RequestId, input_tokens: &'a [TokenId], maximum_sequence_tokens: usize, ) -> Self
Sourcepub fn for_product_request(
request_id: &'a RequestId,
input_tokens: &'a [TokenId],
maximum_sequence_tokens: usize,
product_prompt_tokens: usize,
replayed_output_tokens: usize,
) -> Result<Self>
pub fn for_product_request( request_id: &'a RequestId, input_tokens: &'a [TokenId], maximum_sequence_tokens: usize, product_prompt_tokens: usize, replayed_output_tokens: usize, ) -> Result<Self>
Construct admission for a product request whose execution context may include output tokens replayed after preemption.
pub fn validate(&self) -> Result<()>
Trait Implementations§
Source§impl<'a> Clone for ExecutorPrefillAdmission<'a>
impl<'a> Clone for ExecutorPrefillAdmission<'a>
Source§fn clone(&self) -> ExecutorPrefillAdmission<'a>
fn clone(&self) -> ExecutorPrefillAdmission<'a>
Returns a duplicate of the value. Read more
1.0.0 (const: unstable) · Source§fn clone_from(&mut self, source: &Self)
fn clone_from(&mut self, source: &Self)
Performs copy-assignment from
source. Read moreimpl<'a> Copy for ExecutorPrefillAdmission<'a>
Auto Trait Implementations§
impl<'a> Freeze for ExecutorPrefillAdmission<'a>
impl<'a> RefUnwindSafe for ExecutorPrefillAdmission<'a>
impl<'a> Send for ExecutorPrefillAdmission<'a>
impl<'a> Sync for ExecutorPrefillAdmission<'a>
impl<'a> Unpin for ExecutorPrefillAdmission<'a>
impl<'a> UnsafeUnpin for ExecutorPrefillAdmission<'a>
impl<'a> UnwindSafe for ExecutorPrefillAdmission<'a>
Blanket Implementations§
Source§impl<T> BorrowMut<T> for Twhere
T: ?Sized,
impl<T> BorrowMut<T> for Twhere
T: ?Sized,
Source§fn borrow_mut(&mut self) -> &mut T
fn borrow_mut(&mut self) -> &mut T
Mutably borrows from an owned value. Read more