Open Models Almanac
Methodstraining objectivedistillation objective

single mention · training objective

Depth-representation distillation for geometric supervision

Current and future queries are trained to predict depth representations associated with manipulation frames.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See distillation objective for the methods filed beside it.

Also called depth teacher, LingBot-Depth.

Filed under training objective › distillation objective