Graph explorer Paper
Benchmark graph around Inherited Heads: Audio language models track speakers with their text backbone's attention, and an attention-mass ranking retrieves a different set
Benchmarks and the models evaluated on them. Click a node to inspect it, double-click to expand, drag to pan, wheel to zoom.
1 nodes · 0 edges
No benchmark graph relations recorded for Inherited Heads: Audio language models track speakers with their text backbone's attention, and an attention-mass ranking retrieves a different set
Relations are written only when a source states them. Try another mode above, or go back to Inherited Heads: Audio language models track speakers with their text backbone's attention, and an attention-mass ranking retrieves a different set →