pub enum TaskCategory {
SingleCall,
ParallelCall,
MultiTurn,
Irrelevance,
LongContext,
}Expand description
Category of an agentic tool-calling scenario, following the BFCL split
(plus LongContext, which is gglib-specific).
Variants§
SingleCall
Exactly one tool call is expected.
ParallelCall
Multiple independent tool calls are expected in the same turn.
MultiTurn
A multi-turn, stateful scenario requiring sequential tool calls that build on prior tool results.
Irrelevance
No tool call is expected at all — tests whether the model correctly abstains instead of calling a tool it doesn’t need.
LongContext
Same evaluation as the other categories, but user_prompt is sent
after TuneTask::history has already been injected into the
conversation — tests whether a long prior session (thousands of
tokens of simulated dummy code/turns) causes the model to lose
attention and trigger the agent loop’s LoopDetector/
StagnationDetector, or mis-call a tool it would otherwise get
right from a cold start.
Trait Implementations§
Source§impl Clone for TaskCategory
impl Clone for TaskCategory
Source§fn clone(&self) -> TaskCategory
fn clone(&self) -> TaskCategory
1.0.0 (const: unstable) · Source§fn clone_from(&mut self, source: &Self)
fn clone_from(&mut self, source: &Self)
source. Read moreSource§impl Debug for TaskCategory
impl Debug for TaskCategory
Source§impl<'de> Deserialize<'de> for TaskCategory
impl<'de> Deserialize<'de> for TaskCategory
Source§fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>where
__D: Deserializer<'de>,
fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>where
__D: Deserializer<'de>,
Source§impl PartialEq for TaskCategory
impl PartialEq for TaskCategory
Source§fn eq(&self, other: &TaskCategory) -> bool
fn eq(&self, other: &TaskCategory) -> bool
self and other values to be equal, and is used by ==.