Aether AI Training Projrct
VideoRLHF
Evaluated and ranked AI-generated responses based on quality, relevance, factual accuracy, and instruction adherence.
Provided preference judgments between model outputs to support reinforcement learning from human feedback (RLHF).
Identified hallucinations, reasoning flaws, and guideline violations in generated responses.
Applied detailed annotation guidelines consistently to maintain evaluation quality and reliability.
Evaluated and ranked AI-generated responses based on quality, relevance, factual accuracy, and instruction adherence. Provided preference judgments between model outputs to support reinforcement learning from human feedback (RLHF). Identified hallucinations, reasoning flaws, and guideline violations in generated responses. Applied detailed annotation guidelines consistently to maintain evaluation quality and reliability.
2026 - 2026