tldr-bench

What problem does it solve?

This Skill provides a framework for running token-efficiency benchmarks for the tldr-swinton model, helping to optimize its performance and resource usage.

Core Features & Use Cases

• Token Efficiency Benchmarking: Measure and compare the token usage of different configurations or versions of tldr-swinton.
• Context Tracking: Benchmark how well the model handles and tracks information within a given context.
• Frontier Runs: Execute benchmarks against dynamic or evolving inputs using a shim for logging.
• Use Case: Developers can use this Skill to identify which prompts or model parameters lead to the most token-efficient responses for a specific task, thereby reducing API costs and improving response times.

Quick Start

Use the tldr-bench skill to track context with the baselines variant and print the results.