Oxford-HIPlab/BT_LoRA_skywork80k_on_gemma-2-2b-it_seed1
Updated
Reward Models Inherit Value Biases from Pretraining ICLR2026
Reward models and logprobs for the paper Christian et al., "Reward Models Inherit Value Biases from Pretraining" (ICLR 2026)
-
-
Oxford-HIPlab/BT_LoRA_skywork80k_on_gemma-2-2b-it_seed1-every_1
Updated -
Oxford-HIPlab/BT_LoRA_skywork80k_on_gemma-2-2b-it_seed1-every_10
Updated -
Oxford-HIPlab/BT_LoRA_skywork80k_on_Llama_3.2_3B_Instruct_seed1
Updated -
Oxford-HIPlab/BT_LoRA_skywork80k_on_Qwen2.5-3B-Instruct_seed1
Updated -
Oxford-HIPlab/BT_LoRA_Unified_stepsize_1_on_gemma-2-2b-it_seed1
Updated -
Oxford-HIPlab/BT_LoRA_Unified_stepsize_1_on_Llama-3.2-3B-Instruct_seed1
Updated -
Oxford-HIPlab/BT_LoRA_Unified_stepsize_2_on_gemma-2-2b-it_seed1
Updated -
Oxford-HIPlab/BT_LoRA_Unified_stepsize_2_on_Llama-3.2-3B-Instruct_seed1
Updated -
Oxford-HIPlab/BT_LoRA_Unified_stepsize_4_on_gemma-2-2b-it_seed1
Updated -
Oxford-HIPlab/BT_LoRA_Unified_stepsize_4_on_Llama-3.2-3B-Instruct_seed1
Updated -
Oxford-HIPlab/Unified_stepsize_8_on_gemma-2-2b-it_seed1
Updated -
Oxford-HIPlab/Unified_stepsize_8_on_Llama-3.2-3B-Instruct_seed1
Updated -
Oxford-HIPlab/Unified_stepsize_8_on_Qwen2.5-3B-Instruct_seed1
Updated -
Oxford-HIPlab/Unified_stepsize_16_on_gemma-2-2b-it_seed1
Updated -
Oxford-HIPlab/Unified_stepsize_16_on_Llama-3.2-3B-Instruct_seed1
Updated -
Oxford-HIPlab/Unified_stepsize_16_on_Qwen2.5-3B-Instruct_seed1
Updated -
Oxford-HIPlab/Unified_stepsize_32_on_gemma-2-2b-it_seed1
Updated -
Oxford-HIPlab/Unified_stepsize_32_on_Llama-3.2-3B-Instruct_seed1
Updated -
Oxford-HIPlab/Unified_stepsize_32_on_Qwen2.5-3B-Instruct_seed1
Updated -
Oxford-HIPlab/Unified_stepsize_64_on_gemma-2-2b-it_seed1
Updated -
Oxford-HIPlab/Unified_stepsize_64_on_Llama-3.2-3B-Instruct_seed1
Updated -
Oxford-HIPlab/Unified_stepsize_64_on_Qwen2.5-3B-Instruct_seed1
Updated -
Oxford-HIPlab/iclr2026-lm-logprobs
Viewer • Updated • 2.05M • 20