Search papers, labs, and topics across Lattice.
This study evaluates the performance of small language models in tool use through a comparison of various adaptation mechanisms, including few-shot prompting and hypernetwork-based LoRA adaptation. The research reveals that despite the complexity of hypernetwork-generated weights, they do not enhance performance compared to few-shot prompting, which significantly boosts results. The findings indicate that a well-structured prompting strategy can achieve nearly 80% of GPT-5's performance at a fraction of the latency, emphasizing the effectiveness of simpler approaches in model adaptation.
Few-shot prompting outperforms complex hypernetwork adaptations, achieving 79.7% of GPT-5's performance with significantly lower latency.
Can small language models achieve strong tool-use performance without complex adaptation mechanisms? This paper investigates this question through Meta-Tool, a controlled empirical study comparing hypernetwork-based LoRA adaptation against carefully designed few-shot prompting. Using a Llama-3.2-3B-Instruct backbone, we evaluate four adaptation mechanisms--few-shot prompting, documentation encoding, hypernetwork-generated LoRA weights, and value-guided beam search--across four diverse benchmarks: Gorilla APIBench, Spider 2.0, WebArena, and InterCode. Our central finding is a well-supported negative result: despite generating non-trivial weight matrices, the 227.8M-parameter hypernetwork provides no measurable improvement over few-shot prompting alone. Comprehensive ablation studies reveal that few-shot examples contribute +21.5% to performance and documentation contributes +5.0%, while the hypernetwork adds 0%. A 3B model with well-designed prompts achieves 79.7% of GPT-5's average performance at $10 \times$ lower latency. Error analysis across 722 failure cases spanning all shot counts (0--5) shows that at the 5-shot configuration (106 failures), failure modes are task-dependent: schema-heavy tasks (Spider 2.0, WebArena) show near-zero format errors with remaining failures semantic, while format errors dominate on Gorilla (100%) and InterCode (70%). These findings redirect practitioners toward prompt engineering and example curation rather than complex adaptation architectures.
