Conditions for Emotions

checking Ollama…retry

setup / Einrichtung

Install Ollama from ollama.com and open it.
In a terminal:
ollama pull gemma3:4b
ollama pull embeddinggemma-2:440m
Allow the browser to talk to Ollama:
Mac — launchctl setenv OLLAMA_ORIGINS "*"
Windows — Start menu → “environment variables” → Edit environment variables for your account → New: OLLAMA_ORIGINS = *
Quit and reopen Ollama, then click retry.

Ollama von ollama.com installieren und öffnen.
Im Terminal:
ollama pull gemma3:4b
ollama pull embeddinggemma-2:440m
Dem Browser den Zugriff auf Ollama erlauben:
Mac — launchctl setenv OLLAMA_ORIGINS "*"
Windows — Startmenü → „Umgebungsvariablen“ → Umgebungsvariablen für dieses Konto bearbeiten → Neu: OLLAMA_ORIGINS = *
Ollama beenden und neu starten, dann auf retry klicken.

generation · vision

embedding

recommended — gemma3:4b · embeddinggemma-2:440m

camera needs a vision model (gemma3:4b / gemma4); image distance needs a multimodal embedding model (embeddinggemma-2:440m).

system prompt — the condition

restart applies a changed system prompt

words or sentences — one per line, at least three

PCA 768→2 · drag to pan · scroll to zoom

anchor sentence candidates — one per line

cosine similarity in 768-dimensional space

pole A

pole B

words to place — one per line

semantic differential (Osgood 1957) · position = cos(w,B) − cos(w,A)

prompt — the machine continues this

tokens

click a token to take that path

sentence — write [Variable] where the value goes variable — separated by comma

tokens

same sentence, one substitution, raw continuation — any language works

image — the anchor texts — one per line

image and text in the same 768-dimensional space · cosine similarity · needs a multimodal embedding model (embeddinggemma-2:440m)

question — asked about every frame

every s speak

camera on file:// works in Firefox (allow when asked) — Chrome needs http://localhost