Skip to content

Releases: theoad/cpu2tensor

cpu2tensor 0.5.1

Choose a tag to compare

@theoad theoad released this 12 Sep 10:48
aec51e7

cpu2tensor 0.5.1 adds a supported exact wire-frame observer for profiling and replay capture.

from cpu2tensor import Pool

with open("capture.bin", "wb") as capture:
    def record(frame):
        capture.write(frame.data)

    with Pool([endpoint], wire_observer=record) as pool:
        for batch in pool.read():
            train(batch)

Each callback receives the complete header and payload as a read-only borrowed memoryview, plus stable worker, endpoint, and header metadata. The view is released on return. Per-worker order is preserved; multiworker callbacks may overlap. With capture disabled, the normal path allocates no observer metadata or views. See docs/wire-observer.md.

The attached offline learner runtime contains cpu2tensor 0.5.1 built from commit aec51e7c415b4c6b2a59a136773d2ed727eeed13, CPU PyTorch 2.13.0, and dependencies for CPython 3.12 on Linux x86-64. Its builder boundary is Ubuntu 24.04 with glibc 2.39. QEMU, targets, credentials, and host orchestration remain operator-owned.

cpu2tensor 0.5.0

Choose a tag to compare

@theoad theoad released this 11 Sep 22:57
dbf4fcd

This release provides the reproducible learner runtime requested in #19 for AlphaFlow Process Exploration Recovery.

The attached cpu2tensor-runtime-cp312-linux-x86_64.tar.gz contains:

  • cpu2tensor 0.5.0 built from commit dbf4fcd7e07e5a590b32f2694fa98201b4b8d479
  • CPU PyTorch 2.13.0 and its Python dependencies
  • an offline installer for CPython 3.12 on Linux x86-64
  • a runtime manifest and SHA-256 hashes for every wheel

Verify and install into a new absolute virtual-environment path:

sha256sum -c cpu2tensor-runtime-cp312-linux-x86_64.tar.gz.sha256
tar -xzf cpu2tensor-runtime-cp312-linux-x86_64.tar.gz
./cpu2tensor-runtime-cp312-linux-x86_64/install.sh /absolute/path/to/new-venv

The builder boundary is Ubuntu 24.04, glibc 2.39, CPython 3.12.3, and x86-64. CI verified a fresh offline install, CPU tensor creation, and loading the cpu2tensor native extension. The bundle intentionally excludes QEMU, target material, credentials, and remote-host setup; see docs/runtime-artifacts.md for the contract.