v0.1.6

What's Changed

• fix: infer success from steps by @Shr1ftyy in #73
• feat: bigger submission url ttl by @Shr1ftyy in #74
• feat(endpoints): allow unapproving leader candidates by @Shr1ftyy in #76
• refactor: backend vali msg + episode row sync by @Shr1ftyy in #77

Full Changelog: v0.1.5...v0.1.6

v0.1.5

What's Changed

• feat(leaderboard): expose leader submission ids on leaderboard by @Shr1ftyy in #69
• feat: multi view support + propagate specs to orchestrator by @Shr1ftyy in #71
• Fix episode success logging and rerun endpoint logic by @Shr1ftyy in #72

Full Changelog: v0.1.4...v0.1.5

v0.1.4

What's Changed

• Fix/str ids logs by @Shr1ftyy in #66
• feat: add lite validator mode and public weights endpoint by @Shr1ftyy in #67

Full Changelog: v0.1.3...v0.1.4

v0.1.3

What's Changed

• feat: include evaluation status in submission responses by @Shr1ftyy in #64
• feat: add miner logs by @Shr1ftyy in #65

Full Changelog: v0.1.2...v0.1.3

v0.1.2

What's Changed

• fix(evaluator): avoid overwriting submission files with templates by @Shr1ftyy in #63

Full Changelog: v0.1.1...v0.1.2

v0.1.1

Summary

We've now made it easier to test your agents locally before submitting to the Kinitro platform for evaluation. You'll also now be able to install your dependencies before evaluation runs.

What's Changed

• feat(miner): add local evaluation harness with benchmark specs by @Shr1ftyy in #61
• feat(evaluator): install submission deps at runtime by @Shr1ftyy in #62

Full Changelog: v0.1.0...v0.1.1

v0.1.0 Initial Release

Initial release of Kinitro

Kinitro incentivizes the emergence of agents that can conquer various tasks across different environments. Miners publish agents to compete, validators peform rollouts and evaluate the agents, and reward miners based on the results. All this happens in real-time and can easily be viewed by anyone through our dashboard.

To get started, visit our documentation.

Full Changelog: https://github.com/threetau/kinitro/commits/v0.1.0