Turn words into a picture vector
Turn words into the vector a picture model (CLIP ViT-B-32, OpenAI weights) gives pictures that show them, to compare with picture vectors kept for recordings
Encodes 1 to 64 short texts with the CLIP ViT-B-32 text model (OpenAI weights, ONNX export pinned by BLAKE3) and returns one unit-length 512-number vector per text, with the model named, so it is compared only with picture vectors made by the same model. Runs sealed, without the network.
What it asks for
- texts (Needed, list):
Use it from your AI
Connect the AI you already use to Trillion once. Then ask in your own words; your AI finds this tool by what it does and runs it.
Say to your AI: “Use Trillion to turn words into the vector a picture model (CLIP ViT-B-32, OpenAI weights) gives pictures that show them, to compare with picture vectors kept for recordings”
What your AI calls: run_tool {"tool_id":"tool:words-as-picture-vector"}