Search papers, labs, and topics across Lattice.
5
0
7
0
Existing scoring methods for LVLMs miss the mark, but LookBack reveals how visual grounding can drastically enhance response quality.
Explicit token-level guidance can dramatically enhance LLM adherence to complex system prompts without altering the underlying model architecture.
Data contamination leaves a tell-tale geometric fingerprint across LLM layers, detectable even when standard output-based methods fail after RL post-training.
Agents excel at using tools but falter significantly in navigation, with errors dominating their performance in complex tasks.
LM Arena's model anonymity is more vulnerable than previously thought: a new attack, INTERPOL, leverages interpolated preference learning to expose deep stylistic patterns and manipulate rankings.
