Harman Singh's picture

4

Harman Singh

harman

·

https://harmandotpy.github.io/

AI & ML interests

NLP, Multimodal

Recent Activity

upvoted a paper 23 days ago

Robust Reward Modeling via Causal Rubrics

authored a paper about 1 month ago

Robust Reward Modeling via Causal Rubrics

updated a dataset 2 months ago

harman/ultrainteract_ours_qr_neutrals_our_improve_degrade_data_90k_filtered0p2thresh_plus_original

View all activity

Organizations

Papers 4

arxiv:2506.16507

arxiv:2412.06089

arxiv:2409.02384

arxiv:2404.16816

models 38

harman/gemma2-9b_ultrainteract60k-RM_BT

Text Classification • 9B • Updated May 21 • 3

harman/gemma2-9b_ultrafeedback-CARMA-qrandomized_neutrals_our_improve_degrade_data_BT

Text Classification • 9B • Updated May 16 • 3

harman/gemma2-9b_ultrafeedback-rrm_BT

Text Classification • 9B • Updated May 16 • 3

harman/gemma2-9b_ultrafeedback-RM

Text Classification • 9B • Updated May 16 • 50

harman/gemma2-9b_ultrafeedback-CARMA_originalplusoursall_AND_rrm_sampled_aug_plus_original_wo_neutrals

Feature Extraction • 9B • Updated May 14 • 2

harman/qwen2p5-7b_ultrafeedback-RRM_pairpm_v2

Feature Extraction • 7B • Updated May 13 • 2

harman/qwen2p5-7b_ultrafeedback-CARMA_qrandomized_neutrals_our_improve_degrade_data_all_BT

Feature Extraction • 7B • Updated May 13 • 2

harman/qwen2p5-7b_ultrafeedback-CARMA_qrandomized_neutrals_our_improve_degrade_data_all_pairpm

Feature Extraction • 7B • Updated May 13 • 2

harman/qwen2p5-7b_ultrafeedback-RRM_BT

Feature Extraction • 7B • Updated May 12 • 2

harman/qwen2p5-7b_ultrafeedback-RM_BT

Feature Extraction • 7B • Updated May 12 • 2

datasets 81

harman/ultrainteract_ours_qr_neutrals_our_improve_degrade_data_90k_filtered0p2thresh_plus_original

Viewer • Updated May 22 • 153k • 4

harman/ultrainteract_rrm_50subsamples_filtered0p2thresh_v2_238k_sub90k_plus_original

Viewer • Updated May 22 • 153k • 12

harman/ultrainteract_ours_qr_neutrals_our_improve_degrade_data_120k_filtered0p2thresh_plus_original

Viewer • Updated May 22 • 183k • 6

harman/ultrainteract_ours_qr_neutrals_our_improve_degrade_data_60k_filtered0p2thresh_plus_original

Viewer • Updated May 22 • 123k • 4

harman/ultrainteract_rrm_50subsamples_filtered0p2thresh_v2_238k_sub120k_plus_original

Viewer • Updated May 22 • 183k • 5

harman/ultrainteract_rrm_50subsamples_filtered0p2thresh_v2_238k_sub60k_plus_original

Viewer • Updated May 22 • 123k • 5

harman/ultrainteract_rrm_50subsamples_filtered0p2thresh_v2_238k

Viewer • Updated May 22 • 238k • 5

harman/ultrainteract_rrm_50subsamples_filtered0p2thresh_v2_LAST

Viewer • Updated May 22 • 20.8k • 6

harman/ultrainteract_rrm_50subsamples_filtered0p2thresh_v2_400000

Viewer • Updated May 22 • 27.1k • 6

harman/ultrainteract_rrm_50subsamples_filtered0p2thresh_v2_350000

Viewer • Updated May 22 • 27.3k • 5

View 81 datasets