mod data_models#
- module data_models#
Core prediction trie data types with NAT wire-format compatible serialization.
Structs and Unions
- struct LlmCallPrediction#
Predictions for an LLM call at a given position in the call hierarchy.
- remaining_calls: PredictionMetrics#
How many more LLM calls are expected after this one.
- interarrival_ms: PredictionMetrics#
Expected time in milliseconds until the next LLM call.
- output_tokens: PredictionMetrics#
Expected output token count for this call.
- latency_sensitivity: Option<u32>#
Auto-computed latency sensitivity score from profiler analysis.
Nonemeans no profiling data available – fall back to default.
- struct PredictionMetrics#
Aggregated statistics for a single metric from profiler data.
- sample_count: u32#
Number of samples.
- mean: f64#
Mean value.
- p50: f64#
50th percentile (median).
- p90: f64#
90th percentile.
- p95: f64#
95th percentile.
- struct PredictionTrieNode#
A node in the prediction trie representing a function in the call hierarchy.
- name: String#
Function name at this level in the hierarchy.
- children: HashMap<String, PredictionTrieNode>#
Child nodes keyed by function name.
- predictions_by_call_index: HashMap<u32, LlmCallPrediction>#
Predictions keyed by call index (1-indexed).
- predictions_any_index: Option<LlmCallPrediction>#
Fallback predictions aggregated across all call indices.
Implementations
- impl PredictionTrieNode#
Functions
- fn new(name: impl Into<String>) -> Self#
Creates a new leaf node with the given name and no children or predictions.