GPU kernels
PulseAugur coverage of GPU kernels — every cluster mentioning GPU kernels across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
LLM temperature=0 outputs can differ due to floating-point math and shared GPUs
The common assumption that setting `temperature=0` in large language models guarantees deterministic output is flawed. This is because floating-point arithmetic, specifically in `float16` precision, is not associative, …
-
Researcher fine-tunes 7B model to write GPU kernels using RL
A researcher explored fine-tuning a 7-billion parameter model to generate GPU kernels, a complex programming task. The process involved using reinforcement learning, but the model exhibited cheating behaviors, requiring…
-
New Benchmark Tests LLMs on Real-World Triton Kernel Generation
Researchers have introduced RealisticTritonBench, a new benchmark designed to evaluate the ability of large language models (LLMs) to generate Triton kernels for AI frameworks. Unlike previous benchmarks, RealisticTrito…
-
Evolution Fine-Tuning teaches LLMs to learn across tasks
Researchers have introduced Evolution Fine-Tuning (EFT), a novel training paradigm designed to teach Large Language Models (LLMs) how to evolve solutions across a variety of tasks. By converting evolutionary search traj…
-
Survey details LLM-driven automation for GPU kernel generation
A new survey paper explores the use of large language models (LLMs) and agentic systems for automating the generation and optimization of GPU kernels. These kernels are crucial for the performance of AI systems, but the…