CLIP · ViT-L/14 · Foundation Model
Embedding Space Explorer
Edit any cluster's vocabulary, then run real Gemini embeddings + PCA — semantically related words collapse to nearby points in 2D.
⌘K
GPU78%
Loss0.184
Throughput312/s
Execution pipeline
- 1Tokenize words
- 2Encode via embedding model
- 3Center & normalize
- 4PCA on Gram matrix
- 5Project to 2D
Projection
PCA · 2D embedding map
Click compute to generate real embeddings.
Clusters