MOAI speeds non-interactive homomorphic transformer inference
It cuts BERT-base evaluation time 52.8% versus THOR and reports 2.36 minutes amortized per input on one GPU.
42
AI model
BERT-base
- Mentions · 7d1
- Mentions · 30d1
- All mentions3
- First seenSep 6, 2026
- Last mentioned23h ago
Timeline
3 articles · newest firstQCATS: Query Context-Aware Transformer Slicing for Efficient Predictive Query Processing
QCATS slices transformer FFNs using query predicates, cutting in-database inference latency up to 4.42x.
40
SQS: Bayesian DNN Compression through Sparse Quantized Sub-distributions
SQS unifies weight pruning and low-bit quantization via Bayesian variational learning, compressing Llama3.2 and Qwen2.5 at higher rates with comparable accuracy.
28
Appears with
- Qwen2.5ai model · 1
- LLaMA-3-8Bai model · 1
- Llama3.2ai model · 1
- Qwen-0.6Bai 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.