Index

A

as_dict() (slosizer.schema.PlanResult method) assumptions (slosizer.schema.PlanResult attribute) (slosizer.schema.SimulationResult attribute)

available_vertex_profiles() (in module slosizer.providers.vertex) azure_profile() (in module slosizer.providers.azure)

B

baseline_latency_model (slosizer.schema.PlanOptions attribute)

BaselineLatencyModel (class in slosizer.schema) bucket_required_units() (in module slosizer.simulation)

C

cached_input_token_s (slosizer.schema.BaselineLatencyModel attribute) cached_input_tokens_col (slosizer.schema.RequestSchema attribute) cached_input_weight (slosizer.schema.CapacityProfile attribute)

CapacityProfile (class in slosizer.schema) class_col (slosizer.schema.RequestSchema attribute) compare_scenarios() (in module slosizer.planning)

E

E2E (slosizer.schema.LatencyMetric attribute)

F

fit_baseline_latency_model() (in module slosizer.simulation)

frame (slosizer.schema.RequestTrace attribute) from_dataframe() (in module slosizer.ingest)

H

headroom_factor (slosizer.schema.PlanOptions attribute)

I

input_token_s (slosizer.schema.BaselineLatencyModel attribute) input_tokens_col (slosizer.schema.RequestSchema attribute)

input_weight (slosizer.schema.CapacityProfile attribute) intercept_s (slosizer.schema.BaselineLatencyModel attribute)

L

label() (slosizer.schema.LatencyTarget method) (slosizer.schema.ThroughputTarget method) latency_col (slosizer.schema.RequestSchema attribute) latency_summary (slosizer.schema.PlanResult attribute) (slosizer.schema.SimulationResult attribute) LatencyMetric (class in slosizer.schema)

LatencySLO (class in slosizer.schema) LatencyTarget (class in slosizer.schema) long_input_cached_input_weight (slosizer.schema.CapacityProfile attribute) long_input_input_weight (slosizer.schema.CapacityProfile attribute) long_input_output_weight (slosizer.schema.CapacityProfile attribute) long_input_thinking_weight (slosizer.schema.CapacityProfile attribute) long_input_threshold (slosizer.schema.CapacityProfile attribute)

M

make_synthetic_trace() (in module slosizer.synthetic) MAX_OUTPUT_TOKENS (slosizer.schema.OutputTokenSource attribute) max_output_tokens_col (slosizer.schema.RequestSchema attribute) max_overload_probability (slosizer.schema.ThroughputTarget attribute) max_units_to_search (slosizer.schema.PlanOptions attribute) metadata (slosizer.schema.RequestTrace attribute) metric (slosizer.schema.LatencySLO attribute) metrics (slosizer.schema.PlanResult attribute) min_units (slosizer.schema.CapacityProfile attribute) model (slosizer.schema.CapacityProfile attribute) (slosizer.schema.RequestTrace attribute)

module slosizer.ingest slosizer.planning slosizer.plotting slosizer.providers.azure slosizer.providers.vertex slosizer.schema slosizer.simulation slosizer.synthetic

N

notes (slosizer.schema.CapacityProfile attribute)

O

objective (slosizer.schema.PlanResult attribute) OBSERVED (slosizer.schema.OutputTokenSource attribute) optimize_trace() (in module slosizer.synthetic) output_token_s (slosizer.schema.BaselineLatencyModel attribute)

output_token_source (slosizer.schema.PlanOptions attribute) output_tokens_col (slosizer.schema.RequestSchema attribute) output_weight (slosizer.schema.CapacityProfile attribute) OutputTokenSource (class in slosizer.schema)

P

percentile (slosizer.schema.LatencySLO attribute) (slosizer.schema.ThroughputTarget attribute) plan_capacity() (in module slosizer.planning) PlanOptions (class in slosizer.schema) PlanResult (class in slosizer.schema) plot_capacity_tradeoff() (in module slosizer.plotting)

plot_latency_vs_units() (in module slosizer.plotting) plot_required_units_distribution() (in module slosizer.plotting) plot_slack_tradeoff() (in module slosizer.plotting) predict() (slosizer.schema.BaselineLatencyModel method) provider (slosizer.schema.CapacityProfile attribute) (slosizer.schema.RequestTrace attribute) purchase_increment (slosizer.schema.CapacityProfile attribute)

Q

QUEUE_DELAY (slosizer.schema.LatencyMetric attribute)

R

recommended_units (slosizer.schema.PlanResult attribute) region (slosizer.schema.RequestTrace attribute) request_level (slosizer.schema.PlanResult attribute) (slosizer.schema.SimulationResult attribute)

RequestSchema (class in slosizer.schema) RequestTrace (class in slosizer.schema)

S

schema (slosizer.schema.RequestTrace attribute) simulate_capacity() (in module slosizer.simulation) SimulationResult (class in slosizer.schema) slack_summary (slosizer.schema.PlanResult attribute) (slosizer.schema.SimulationResult attribute) slo (slosizer.schema.LatencyTarget attribute) slosizer.ingest module slosizer.planning module slosizer.plotting module

slosizer.providers.azure module slosizer.providers.vertex module slosizer.schema module slosizer.simulation module slosizer.synthetic module source (slosizer.schema.CapacityProfile attribute) summarize_slack() (in module slosizer.simulation)

T

target (slosizer.schema.PlanResult attribute) thinking_token_s (slosizer.schema.BaselineLatencyModel attribute) thinking_tokens_col (slosizer.schema.RequestSchema attribute) thinking_weight (slosizer.schema.CapacityProfile attribute)

threshold_s (slosizer.schema.LatencySLO attribute) throughput_per_unit (slosizer.schema.CapacityProfile attribute) ThroughputTarget (class in slosizer.schema) time_col (slosizer.schema.RequestSchema attribute)

U

unit_name (slosizer.schema.CapacityProfile attribute) (slosizer.schema.PlanResult attribute) (slosizer.schema.SimulationResult attribute)

units (slosizer.schema.SimulationResult attribute)

V

vertex_profile() (in module slosizer.providers.vertex)

W

windows_s (slosizer.schema.ThroughputTarget attribute)