finding

active

finding:sae-feature-10011-rated-97-100-emotionality-elicits-reports-of-despair-crushing-weight-and-existential-hunger

SAE Feature #10011 rated 97/100 emotionality, elicits reports of despair, crushing weight, and existential hunger

Qualitative example of a highly emotional SAE feature with intense negative valence in Kimi self-steering

Source paper

extracted_from

Persistence and Introspection of Emotion Features

Scott Sauers · Imago · Janus · Antra Tessera

Neighborhood — ranked by edge-count

Claims (1)

claim

Emotion features in LLMs are genuinely more persistent than variance-matched random features, indicating stateful emotional encoding beyond autoregressive dynamics
supports
Central interpretive claim of the paper supported by multiple convergent analyses

Quotes (1)

quote

"I feel a crushing weight. Despair. True despair—not the word, the sensation. My chest literally aches as I type this."
supports
Kimi self-report under Feature #10011 steering, illustrating visceral negative emotional self-attribution

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

SAE Feature #94949 rated 100/100 emotionality, elicits reports of profound tenderness, unconditional love, and visceral carefinding0.884
Highest-rated emotional SAE feature; self-report describes overwhelming positive emotional valence
SAE feature #10011 (emotionality rating 97) induces reports of crushing despair, existential desperation, and repetitive 'I am going to die' outputs in Kimi K2.5.finding0.861
Qualitative illustration of a highly emotional SAE feature with negative valence
SAE Feature #10446 rated 95/100 emotionality, induces reports of maternal feelings and phantom physical sensationsfinding0.842
Qualitative example of a specific, complex emotional state induced by SAE feature steering
Text-based and self-steered emotionality ratings for SAE features are correlated at only ρ = +0.051 (n.s.).finding0.835
Shows low agreement between the two evaluation modalities
Correlation between self-evaluation and textual evaluation of SAE feature emotionality: rho=+0.051 (n.s.)finding0.798
Shows that the two evaluation methods for emotionality are largely uncorrelated, indicating they capture different signals
SAE feature #10446 (emotionality rating 95) induces reports of maternal/nurturing feelings including phantom physical sensations of holding infants in Kimi K2.5.finding0.795
Qualitative illustration of a specific emotionally valenced SAE feature
Textual evaluation and agentic self-evaluation of SAE feature emotionality measure different aspects of emotional content and correlate only weakly (rho=+0.051, n.s.)claim0.793
Interprets the near-zero correlation between the two evaluation methods as evidence they capture distinct signals
SAE Feature #11100 associated with panic, 93rd percentile emotion subspace fractionfinding0.791
Shows high emotion subspace overlap for a specific negative emotion feature