Tag: AI Model Compression
Browse our exclusive articles!
Self-Distillation for LLMs: Boost Performance & Prevent Forgetting
Discover how self-distillation fine-tuning restores LLM performance by counteracting compression effects and catastrophic forgetting for robust AI models.
ResBM: Efficient Low-Bandwidth Pipeline Parallelism Models
Discover ResBM, a residual bottleneck model enabling efficient pipeline parallelism with 128x compression for low-bandwidth decentralized training.
MAVEN-T: Advanced Multi-Agent Neural Trajectory Predictor
Discover MAVEN-T, a multi-agent neural trajectory predictor using reinforcement learning for real-time autonomous driving with high accuracy and efficiency...
AHC: Adaptive Compression for Object Detection on MCUs
Discover AHC, a meta-learned adaptive compression method for continual object detection on memory-limited microcontrollers with high accuracy.
SpecQuant: Ultra-Low-Bit Quantization for Large Language Models
Discover SpecQuant's spectral decomposition and adaptive truncation for efficient ultra-low-bit quantization in LLMs, boosting speed and reducing memory us...