ResearchBenchmarksPersonalized Interaction on MBPP SyntheticFollow7.9Personalization ScorePPOpt5.88246.40626.937.4538Feb 12, 2026Evaluation ResultsMethodMethodLinksPersonalization ScoreTask Completion ScorePPOptBackbone=Llama-3-8b-in...Backbone=Llama-3-8b-instruct2026.027.98.7PPOptBackbone=GPT-oss-20bBackbone=GPT-oss-20b2026.027.788.86PPOptBackbone=Qwen3-8bBackbone=Qwen3-8b2026.027.748.84VanillaBackbone=Llama-3-8b-in...Backbone=Llama-3-8b-instruct2026.025.969.08VanillaBackbone=Qwen3-8bBackbone=Qwen3-8b2026.025.969.08VanillaBackbone=GPT-oss-20bBackbone=GPT-oss-20b2026.025.969.08