← all emotions · markdown (agent skill)

Interest conditioning guide

Target: EmoNet Interest strength. reward = (genu + blend + 1.25·emo)/(1+WER).

TL;DR. Best reward comes WITHOUT the LoRA — plain BASE tops the table at 0.690, edging out the evolved prompt (BASE_P, 0.679) and beating every merge (LoRA50 0.618, LoRA150 0.552). Interest is essentially a native emotion here: the base model already hits emo=0.187 with a neutral prompt, so just prompt plainly and skip the LoRA entirely. Adding the LoRA barely moves target strength (emo 0.187 to 0.227 only at 150%) while dragging blend down hard (0.453 to 0.274) and inflating WER (0.269 to 0.513 at 100%). The one thing the LoRA reliably drags along is a cartoonish, rough delivery (corr Cartoonish Style 0.82, Roughness 0.73 at 100%) — so higher merges buy you a caricatured voice, not more genuine interest.

Without a LoRA

Best: neutral prompt, no LoRA → reward 0.69
GENERAL: A voice viscerally expressing interest, a keenly interested voice, leaning in, curious and engaged, with every word. SCRIPT: (keenly interested and curious) "<neutral sentence>" · temp 1.15 top_p 0.95 top_k 40

With the emotion LoRA (dose)

50% ★ — reward 0.62 · emo 0.19
genu 0.10 · blend 0.37 · qual 0.58 · WER 0.20
100% — reward 0.54 · emo 0.20
genu 0.15 · blend 0.33 · qual 0.54 · WER 0.51
150% — reward 0.55 · emo 0.23
genu 0.15 · blend 0.27 · qual 0.53 · WER 0.36

Conditions

conditionrewardemotiongenublendqualityWER
neutral prompt, no LoRA0.690.190.110.450.630.27
evolved prompt, no LoRA0.680.200.120.430.580.26
evolved prompt + LoRA 50%0.620.190.100.370.580.20
evolved prompt + LoRA 100%0.540.200.150.330.540.51
evolved prompt + LoRA 150%0.550.230.150.270.530.36

Side effects & correlations (LoRA @ 100%)

↑ pushed up: Hope Enthusiasm Optimism (+0.38), Impatience and Irritability (+0.24), Elation (+0.17)
↓ pushed down: Warmth (-0.15), Recording Quality (-0.14), Fullness (-0.13)
+ correlates: Cartoonish Style (+0.82), Roughness (+0.73), Metallic Character (+0.67)
− anti-correlates: Chunking (-0.67), Smoothness (-0.61), Fatigue Exhaustion (-0.61)

Conclusion

Best reward comes WITHOUT the LoRA — plain BASE tops the table at 0.690, edging out the evolved prompt (BASE_P, 0.679) and beating every merge (LoRA50 0.618, LoRA150 0.552). Interest is essentially a native emotion here: the base model already hits emo=0.187 with a neutral prompt, so just prompt plainly and skip the LoRA entirely. Adding the LoRA barely moves target strength (emo 0.187 to 0.227 only at 150%) while dragging blend down hard (0.453 to 0.274) and inflating WER (0.269 to 0.513 at 100%). The one thing the LoRA reliably drags along is a cartoonish, rough delivery (corr Cartoonish Style 0.82, Roughness 0.73 at 100%) — so higher merges buy you a caricatured voice, not more genuine interest.