Search papers, labs, and topics across Lattice.
5
1
7
2
Enterprise AI agents can now generate valid, high-fidelity tabular data for training and evaluation without needing access to sensitive schemas or real systems, bypassing critical industry bottlenecks.
EdgeGen is a synthetic task generation framework that extracts compliance rules from an agent's specification and uses them to generate database-grounded edge-case tasks designed to violate these rules, and when combined with existing synthetic data generation techniques, enables agent improvement through finetuning and harness optimization.
TRACE not only preserves but enhances tool understanding in enterprise LLMs, achieving up to 9 percentage points improvement in accuracy while enabling real-time retrieval.
ToolSense uncovers a surprising 50-64% drop in retrieval performance for LLMs when faced with realistic, ambiguous queries, challenging assumptions about their tool understanding.
LLMs can retrieve tools far more effectively when their query rewriting and the retrieval encoder are trained together, iteratively, than when trained separately.
