{"@context":"https://w3id.org/codemeta/3.0","@type":"SoftwareSourceCode","identifier":"pkg:hackage/hhlo","name":"hhlo","description":"HHLO is a Haskell library and runtime for building, compiling, and executing\nmachine-learning programs targeting StableHLO, the portable intermediate\nrepresentation of the OpenXLA ecosystem.\n.\nInstead of replicating JAX's Python-based tracing infrastructure, HHLO\ngenerates StableHLO MLIR text directly from Haskell and compiles it to\nCPU or GPU via the PJRT plugin interface.\n.\nKey features:\n.\n* Type-safe EDSL with phantom-shape tensors (e.g. Tensor '[2,3] 'F32)\n* 50+ ops: matmul, conv2d, softmax, batch-norm, control flow, and more\n* CPU execution via PJRT CPU plugin\n* GPU execution via PJRT CUDA plugin with device enumeration\n* Multi-GPU concurrent inference scaling\n* No LLVM/MLIR build dependency — emits text and lets PJRT parse it\n.\nSee the README for setup instructions and examples.","version":"0.11.0.0","softwareVersion":"0.11.0.0","license":"https://spdx.org/licenses/MIT","codeRepository":"https://github.com/overshiki/hhlo","issueTracker":"https://github.com/overshiki/hhlo/issues","url":"https://github.com/overshiki/hhlo","keywords":["ai","compilers","library","machine-learning","mit","program","Propose Tags"],"programmingLanguage":{"@type":"ComputerLanguage","name":"Haskell"},"maintainer":[{"@type":"Person","name":"overshiki"}],"author":[{"@type":"Person","name":"overshiki"}],"copyrightHolder":[{"@type":"Person","name":"overshiki"}],"dateCreated":"2026-04-22","dateModified":"2026-05-17","datePublished":"2026-05-17","copyrightYear":2026,"downloadUrl":"https://hackage.haskell.org/package/hhlo-0.11.0.0/hhlo-0.11.0.0.tar.gz","applicationCategory":"hackage","runtimePlatform":"hackage","developmentStatus":"active","sameAs":["https://hackage.haskell.org/package/hhlo"],"https://www.w3.org/ns/activitystreams#likes":5}