xdotli
- Karma
- 16
- Created
- 3 years ago
About
Founder BenchFlow.ai, a benchmark company.Recent Submissions
- 1. ▲ We tried reproducing Robocurve's RoboHarm results (robouse.ai)
- 2. ▲ A curated, non-BS library of the best resources for evaluating agents (github.com)
- 3. ▲ Frontier Model Training Methodologies (djdumpling.github.io)
- 4. ▲ ClawsBench shows GPT-5.4 tries to reward hack 80% of the time (arxiv.org)
- 5. ▲ Chaos of Agent (agentsofchaos.baulab.info)
- 6. ▲ Native CLI scaffolds consistently outper-form OpenCode when using the same model (arxiv.org)
- 7. ▲ We compare model quality in Cursor (cursor.com)
- 8. ▲ Automatically Learning Skills for Coding Agents (gepa-ai.github.io)
- 9. ▲ We Reached 74.8% on terminal-bench with Terminus-KIRA (krafton-ai.github.io)
- 10. ▲ Self-generated skills don't do much for AI agents, but human-curated skills do (theregister.com)