LLM AI Evaluator & LLM AI Trainer (RLHF)
Internal Proprietary ToolingTextRLHF
As an LLM AI Evaluator & Trainer (RLHF), I evaluated AI responses for quality, factual accuracy, safety, and instruction-following using structured criteria. I performed RLHF-style evaluation by comparing and ranking multiple AI responses, providing structured feedback to inform model improvement. My work included labeling hallucinations, logical errors, and policy violations, as well as creating high-quality reference answers for model iteration.
• Evaluated and ranked LLM outputs for preference and alignment.
• Annotated safety and factuality concerns in model responses.
• Provided reference/gold standard answers for reinforcement learning.
• Conducted multilingual parallel translations and style consistency reviews across languages.
As an LLM AI Evaluator & Trainer (RLHF), I evaluated AI responses for quality, factual accuracy, safety, and instruction-following using structured criteria. I performed RLHF-style evaluation by comparing and ranking multiple AI responses, providing structured feedback to inform model improvement. My work included labeling hallucinations, logical errors, and policy violations, as well as creating high-quality reference answers for model iteration. • Evaluated and ranked LLM outputs for preference and alignment. • Annotated safety and factuality concerns in model responses. • Provided reference/gold standard answers for reinforcement learning. • Conducted multilingual parallel translations and style consistency reviews across languages.
2024