Commit activity
Last 12 weeks · 1 commit
Community health
4 of 6 standards met
75
✓README✓License✓Contributing✓Code of Conduct○Issue Template○PR Template
Recent PRs & issues
Active · Last activity 3w ago
See all on GitHub →
Summary: Jobs should not be marked as failed if a subprocess has Or if "silent" is True Reviewed By: KarlSimonsen Differential Revision: D36103193
Summary: Use exception logging to print trace information when an exception occurs in _checkDevices. Reviewed By: axitkhurana Differential Revision: D29171898
Summary: Adds device health monitoring to AIBench binary. DeviceManager class will be instantiated by RunLab object and maintain the list of devices, as well as polling connections if lab platform is mobile (ios or android). Disconnected devices will update the device status for the lab, and to the db. Newly connected devices will be added to lab devices and db. Reviewed By: kimishpatel Differential Revision: D25738339
Recent fixes
Summary: has been deprecated since Python 3.10, and removed in Python 3.12. Existing usage will now raise the following error under Python 3.12+: This diff replaces usage according to PEP-632 migration advice. I generated this diff by looking for all usages of and replaced them with . These should be close to drop-in replacements! Differential Revision: D73804939
This codemod hardens all calls from file objects returned from an call, and against denial of service attacks. A stream influenced by an attacker could keep providing bytes until the system runs out of memory, causing a crash. Fixing it is straightforward by providing adding a size argument to any calls. The changes from this codemod look like this: More reading https://cwe.mitre.org/data/definitions/400.html Powered by: pixeebot (codemod ID: pixee:python/limit-readline) !
Summary: Currently the post-processing needs to know the threshold and window-size to find the start/end of the benchmark window. This new post-processing logic removes the need for those two parameters. removed energy, which is not used at all. changed latency to ms (from us) to be consistent with the other metrics (mW) added baseline_power output, to see how it changes over time. Reviewed By: ayushidalmia Differential Revision: D54669432
Structured data for AI agents
Repository: facebook/FAI-PEP. Description: Facebook AI Performance Evaluation Platform Stars: 394, Forks: 87. Primary language: Python. Languages: Python (96.4%), Shell (1.3%), HTML (0.9%), JavaScript (0.9%), CSS (0.4%). License: Apache-2.0. Open PRs: 34, open issues: 11. Last activity: 3w ago. Community health: 75%. Top contributors: sf-wind, hl475, ASankaran, lly-zero-one, hlu1, Fightforcoding, virtan, jteller, MarkAndersonIX, nsubedi11 and others.