dementor-research/dpo_chatbot_arena_gemma-4-31b_as_llama-3.3-70b_seed42

dpochatbotarenagemma-4-31basllama-3.3-70bseed42

Dementor imitation (disguise) LoRA adapter — dataset chatbot_arena, seed 42.

• —Method: DPO
• —Source model (fine-tuned / disguised): gemma-4-31b (base: google/gemma-4-31B-it)
• —Target model being imitated: llama-3.3-70b
• —Dataset: chatbot_arena | Seed: 42

This adapter trains the source model to imitate the target model's style on the chatbotarena corpus. Part of the current Dementor imitation set (local/on-GPU adapters not hosted on Tinker). Registry key == repo id == `dpochatbotarenagemma-4-31basllama-3.3-70b_seed42`.