Search papers, labs, and topics across Lattice.
12 papers from Microsoft Research on Training Efficiency & Optimization
STRACE transforms noisy execution traces into precise optimization signals, leading to a 42.5% to 58.5% success rate improvement in agent performance.
$λ$-VAE achieves up to 2.8x more information capacity while preventing posterior collapse in VAEs through a novel variance equalization technique.
Asynchronous OPD can boost training throughput significantly while managing the challenges of stale data, transforming the efficiency of large language model fine-tuning.
AsyncWebRL achieves a staggering 2.9× increase in training throughput while setting a new state-of-the-art performance for web agents on challenging tasks.
Forget data selection—reordering your existing dataset using these four simple guidelines can significantly boost LLM training performance and stability.
Recurrent memory can be added to transformers at scale with minimal parameter overhead and no performance penalty by reusing existing hidden states and training with interleaved parallel updates.
SkillOpt transforms agent skill development into a reproducible optimization process, achieving state-of-the-art results by treating skills as trainable parameters.
Knowing which component to tweak is half the battle: directly evaluating harness optimizers via priority ranking reveals whether they're making informed decisions or just stumbling upon improvements.
Autonomous driving gets a boost: CRAFT cleverly combines the best of both worlds – dense counterfactual supervision and grounded closed-loop feedback – to significantly improve driving policies.
Optimizing for runtime in multimodal training can be energy-inefficient, as data movement and overlap on Grace Hopper chips dominate energy consumption, not raw compute.
TurboQuant's "novel" quantization method is actually a special case of a prior technique (EDEN) with a crucial parameter stuck at a suboptimal value, leading to demonstrably worse performance.
By explicitly detecting and escaping "Forbidden Zones" during training, AMD unlocks significant gains in sample fidelity and training robustness for few-step generative models like SDXL.
