Back to evaluations
Public evaluation
KV-Cache Capacity Test
Evaluates KV-cache footprint calculation accuracy and admission control shedding decisions.
Evaluation type
task based
Challenge
KV-Cache Capacity Planner and Admission Controller with OpenAI Agents SDK and Claude 4 Sonnet
Difficulty
Advanced
Rigor
Unspecified
Evaluation overview
How the linked challenge is judged: tasks, benchmarks, and criteria count.
Tasks
1
Benchmarks
0
Criteria
0
Task templates
Inputs and expected outputs.
Task 1
KV-Cache Capacity Test
Tests admission controller decisions under tight memory budgets.
Input format
JSON containing current_free_vram_gb, sequence_length, model_config
Output format
JSON decision (admit|queue|reject), calculated_kv_cache_mb