Lgnd.ai Devlogs

Research notes and devlogs from LGND AI and Clay

Research notes and devlogs from LGND AI and Clay.

Playing Inside Clay’s Embedding Space

embeddings

foundation-models

clay

probing

Why Clay picks the mature orchard over the brown field — and what that tells you about embeddings.

Cheaper than a pixel, faster than an agent; the geoembedding knows more than you think

Clay embeddings do contain semantic information like location, size, orientation, and count of semantics within the image. Readable with linear probes.

embeddings

foundation-models

clay

probing

This post shows that Clay embeddings do contain semantic information like location, size, orientation, and count. And the mechanism to scale this retrieval to planetary scale, without agents or reading pixels.

ELLE: Embeddings Linearly contain their Loss Estimate

embeddings

uncertainty

geospatial

foundation-models

self-supervised

Foundation model CLS embeddings linearly encode their own per-sample pretraining loss – readable by a Ridge probe in 1 us, with no labels and no extra forward passes. Validated across 20 models and 6 modalities.

---
title: "Lgnd.ai Devlogs"
subtitle: "Research notes and devlogs from LGND AI and Clay"
listing:
  contents: posts
  sort: "date desc"
  type: default
  categories: true
  sort-ui: false
  filter-ui: false
  fields: [date, title, description, categories, image]
  date-format: "YYYY-MM-DD"
page-layout: full
title-block-banner: false
---

Research notes and devlogs from LGND AI and Clay.

---