Small LLMs fail complement consistency in agent opinion simulations
Paired true and false reports from Qwen3-4B, Llama-3.2-3B and Gemma-3-4B do not sum to one, and complementary populations diverge within five rounds.
36
AI model
Llama-3.2-3B
- Mentions · 7d2
- Mentions · 30d3
- All mentions4
- First seenSep 25, 2026
- Last mentioned2d ago
Timeline
4 articles · newest firstSPA approximates gated activations for faster FHE inference
On [-5,5], SPA-balanced is 3.25× faster than MOAI for erf-GELU at comparable error.
42
Poster: A Preliminary Study of LLM Distillation Inference
Distillation-inference test flags Llama-3.2-3B models distilled from Qwen2.5-7B at TPR 1.0 and p=0.02.
48
PrismQuant: Optimal Null-Space Rotations for Grouped Quantizers
PrismQuant's INT4 rotations nearly match full precision on Llama-3.1-70B and speed Llama-3.1-8B inference.
58
Appears with
- Alibabavendor · 2
- Metavendor · 2
- Qwen2.5-7Bai model · 1
- PrismQuantproduct · 1
- Mistral AIvendor · 1
- Gemma-3-4Bai model · 1
- Llama-3.1-8Bai model · 1
- Llama-3.1-70Bai model · 1
- Qwen3-4Bai model · 1
- Gemma-4-E2Bai model · 1
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.