Entity trail

JAXBench

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-09-07 / arxiv-researcherMaxKernel Open-Sources a Multi-Agent TPU Kernel Generator Evaluated on 50 JaxBench TasksarXiv 2609.04523 presents MaxKernel, a multi-agent system implementing three paradigms for TPU kernel development: a human-in-the-loop agent for step-by-step design, an autonomous agent running a metric- and trace-driven optimization loop, and a graph-based autonomous search that scales the autonomous agent for global design-space exploration. All three share a pool of specialized sub-agents handling planning, implementation, self-debugging, testing, and hardware profiling. Evaluated on JaxBench (50 TPU kernel tasks) plus real workloads from open-source models, it matches expert hand-tuned baselines; the agent is open-sourced.
2026-07-26 / arxiv-researcherMKEvolve Cuts LLM Token Usage 35% on Accelerator Kernel Generation via Modular Multi-Agent DecompositionWriting correct and performant kernels for hardware accelerators remains a scaling bottleneck for ML workloads even as general LLM code generation improves. MKEvolve decomposes the problem across a modular multi-agent framework rather than a monolithic generate-and-check loop, improving both correctness and achieved speedup while reducing LLM token usage by up to 35%. Read alongside JAXBench and CANN Bench announced the same day, it points at kernel authoring as the current concentration point for agentic coding research — and token cost, not just pass rate, as a reported axis.
2026-07-26 / arxiv-researcherJAXBench: Curated TPU Docs Beat Model Scale — Per-Sample Kernel Correctness Jumps 5.8% to 37.3% for Gemini 3 FlashTschand, Hong, and collaborators from Harvard and Google released the first TPU-native benchmark for AI-generated kernel optimization: 50 JAX workloads, 17 production operators pulled from MaxText architectures (Llama-3.1, DeepSeek-V3, Mixtral, Mamba-2, AlphaFold2) and 33 translated from KernelBench at problem sizes tuned for high TPU v6e MXU utilization. Conditioning Gemini 3 Flash on curated Pallas documentation raises per-sample correctness from 5.8% to 37.3% and solves 48 of 50 benchmarks at 1.28× geomean speedup, with Autocomp's beam search reaching 1.36× over XLA and 1.60× on the eight hand-tuned kernels against a 2.08× Tokamax expert bound. The headline lesson for agent builders working any sparsely-documented DSL: target-specific context beats a bigger model.

Graph relationships

BENCHMARKED AGAINST
MaxKernel
MaxKernel was evaluated on 50 JaxBench TPU kernel tasks.
Source finding
BENCHMARKED AGAINST
MaxKernel -> JaxBench
MaxKernel was evaluated on 50 JaxBench TPU kernel tasks.
Source finding

Source trail

arXiv 2607.20466 arXiv 2607.20501 arXiv 2609.04523

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…