AI/ML●●Solid
Per-request emotion steering for vLLM, with batching preserved
Per-request emotion steering that preserves vLLM continuous batching for Qwen3.
Niche GemBig Brain
ChrisPoensgen
101mo ago
Emotional probes for Gemma 4 E4B
Anthropic's emotion research replicated open-source with a Flask visualizer for Gemma 4.
ML researchers and mechanistic interpretability engineers
TransformerLens · Neuronpedia
Per-request emotion steering that preserves vLLM continuous batching for Qwen3.
Common Lisp + BKNR stack is interesting, but it's a themed CRUD demo.
Found L2 jobs halved while L1 stayed stable — original Harvard paper missed this.
High-throughput vLLM API for per-request emotion steering beats prompt-only methods.
Pure math kernel produced emergent behavior — one instance started dreaming unprompted.
Sub-cent CPU-only voice agent with vision-keyed proactivity beats cloud APIs on cost.