AI/ML●●Solid
Try out emotion steering of LLMs here
High-throughput vLLM API for per-request emotion steering beats prompt-only methods.
Niche GemBig Brain
ChrisPoensgen
201mo ago
Back to browse
GitHub Repository
Extract and serve CAA-style emotion steering vectors for any HF causal LM
5 starsPython
Per-request emotion steering for vLLM, with batching preserved
by ChrisPoensgen·May 5, 2026·1 point·0 comments
AI Analysis
●●SolidNiche GemBig Brain
Per-request emotion steering that preserves vLLM continuous batching for Qwen3.
Strengths
- •Preserves continuous batching in vLLM while applying per-request steering vectors.
- •Automated layer selection uses AUC probing to find optimal steering locations.
- •Provides both extraction tools and a serving API for immediate integration.
Weaknesses
- •Fast path is strictly pinned to vLLM 0.20.x and Qwen3 architecture only.
- •Niche utility limits appeal to researchers specifically interested in activation addition.
Category
Target Audience
ML engineers and researchers experimenting with model steering
Similar To
LLM-Beamer · Activation Additions
Similar Projects
Developer Tools●Mid
vLLM Studio – A macOS app for using vLLM models
Local vLLM GUI with layout-aware OCR, but macOS-only in a crowded space.
Ship ItNiche Gem
ahmedriad1
442mo ago
AI/ML●●Solid
Emotional probes for Gemma 4 – Replicating Anthropic's emotion research
Anthropic's emotion research replicated open-source with a Flask visualizer for Gemma 4.
Niche GemBig Brain
justsomeguy1996
112mo ago
AI/ML●●●Banger
Coding agents find the right GPU bottleneck 70% of the time, fix it 30%
Reveals agents diagnose bottlenecks 87% correctly but fix them only 17%—scaffolding matters more than model.
Big BrainWizardry
ayushnangia16
313mo ago
Education●●Solid
Tiny-vLLM – high performance LLM inference engine in C++ and CUDA
Build vLLM from scratch with PagedAttention kernels when llama.cpp already exists.
Big BrainNiche Gem
yu3zhou4
2051820d ago
AI/ML●●Solid
Kronaxis Router – Don't pay frontier prices when a local LLM is enough
LLM cost routing with LoRA awareness when LiteLLM already handles basic proxying.
Big BrainSolve My Problem
JasonDuke
202mo ago