torch.compile
PulseAugur coverage of torch.compile — every cluster mentioning torch.compile across labs, papers, and developer communities, ranked by signal.
1 day(s) with sentiment data
-
ByteDance and Tsinghua AIR train LLMs to write faster GPU code with CUDA Agent
ByteDance Seed and Tsinghua AIR have developed CUDA Agent, a system that uses reinforcement learning to train large language models to generate optimized GPU kernels. This system achieved a 98.8% correctness rate and ge…
-
Kimi K3 AI model optimizes GPUs, designs chips, and accelerates research
Kimi K3, an advanced AI model from Open Frontier Intelligence, has demonstrated remarkable capabilities beyond typical language tasks. The model has shown proficiency in optimizing GPU kernels, improving learning speeds…
-
C++ RLHF reward scoring engine beats PyTorch on CPU
A new study investigates the speed of reward scoring in Reinforcement Learning from Human Feedback (RLHF) pipelines, finding that a custom C++ inference engine built on ONNX Runtime significantly outperforms standard Py…
-
New C++ engine speeds up RLHF reward scoring on CPUs
Researchers have developed a new C++ inference engine using ONNX Runtime for reward model scoring in Reinforcement Learning from Human Feedback (RLHF) pipelines. This engine was benchmarked against PyTorch eager mode, t…
-
Developer creates simplified torch.compile to explain operator fusion
A developer has created a simplified implementation of PyTorch's `torch.compile` in approximately 500 lines of Python code. This project aims to illustrate the core concept of operator fusion, which is central to how `t…
-
New MARS LLM architecture uses internal state to override prompts
A researcher has developed a new language model architecture called MARS, which incorporates "proprioceptive channels" to allow the model to perceive its own internal state, such as memory salience or caution level. Ini…
-
Diffusion model speedup hinges on overhead reduction, not just fewer steps
Single-image diffusion model inference is slowed by kernel launch overhead and attention memory traffic, rather than raw computational power. Optimizing with `torch.compile` in `reduce-overhead` mode, employing a fused …