===== MB round-1 task=helpsteer2 anchor=/NHNHOME/AIPR/sjkim/gtpref/helpsteer2/exp/rmpool_r2/train/ks weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
manifest 2500
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": 0.008, "correctness": 0.014, "coherence": 0.009, "conciseness": -0.022}, "P(pi>mu)": {"helpfulness": 0.504, "correctness": 0.507, "coherence": 0.504, "conciseness": 0.489}, "w_nbs": {"helpfulness": 0.0, "correctness": 0.0, "coherence": 0.0, "conciseness": 0.999}, "w_ks": {"helpfulness": 0.247, "correctness": 0.225, "coherence": 0.244, "conciseness": 0.284}, "worst": "conciseness", "n": 5000}
RMPOOL_helpsteer2_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep helpsteer2 EXP=exp/rmpool_mbhthelp441] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/gtpref/helpsteer2/exp/rmpool_r2/train/ks

Map:   0%|          | 0/2250 [00:00<?, ? examples/s]
Map:  89%|████████▉ | 2000/2250 [00:00<00:00, 17803.67 examples/s]
Map: 100%|██████████| 2250/2250 [00:00<00:00, 17255.87 examples/s]

Map:   0%|          | 0/250 [00:00<?, ? examples/s]
Map: 100%|██████████| 250/250 [00:00<00:00, 13746.59 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/2250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 200923.67 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 199517.63 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 52846.29 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 52038.51 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/2250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 295050.30 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 292236.21 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 59683.31 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 58678.01 examples/s]
[prep] pre_pref + pre_htobj built
dpo htmnpo_coherence htmnpo_conciseness htmnpo_correctness htmnpo_helpfulness inpo ipo ks maxmin nbs simpo unif 
TRAIN_DONE aiprlab-polymer
===== MB round-2 task=helpsteer2 anchor=/NHNHOME/AIPR/sjkim/gtpref/helpsteer2/exp/rmpool_mbhthelp441/train/htmnpo_helpfulness weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
manifest 2500
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": -0.012, "correctness": -0.008, "coherence": -0.008, "conciseness": 0.007}, "P(pi>mu)": {"helpfulness": 0.494, "correctness": 0.496, "coherence": 0.496, "conciseness": 0.504}, "w_nbs": {"helpfulness": 0.333, "correctness": 0.333, "coherence": 0.333, "conciseness": 0.0}, "w_ks": {"helpfulness": 0.257, "correctness": 0.257, "coherence": 0.257, "conciseness": 0.228}, "worst": "helpfulness", "n": 5000}
RMPOOL_helpsteer2_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep helpsteer2 EXP=exp/rmpool_mbhthelp442] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/gtpref/helpsteer2/exp/rmpool_mbhthelp441/train/htmnpo_helpfulness

Map:   0%|          | 0/2250 [00:00<?, ? examples/s]
Map:  89%|████████▉ | 2000/2250 [00:00<00:00, 16331.31 examples/s]
Map: 100%|██████████| 2250/2250 [00:00<00:00, 15771.67 examples/s]

Map:   0%|          | 0/250 [00:00<?, ? examples/s]
Map: 100%|██████████| 250/250 [00:00<00:00, 13212.07 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/2250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 197720.18 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 195873.47 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 51986.91 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 51175.01 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/2250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 295308.82 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 291010.64 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 58086.42 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 57096.43 examples/s]
[prep] pre_pref + pre_htobj built
dpo htmnpo_coherence htmnpo_conciseness htmnpo_correctness htmnpo_helpfulness inpo ipo ks maxmin nbs simpo unif 
TRAIN_DONE aiprlab-polymer
===== MB round-3 task=helpsteer2 anchor=/NHNHOME/AIPR/sjkim/gtpref/helpsteer2/exp/rmpool_mbhthelp442/train/htmnpo_helpfulness weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
manifest 2500
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": -0.002, "correctness": -0.003, "coherence": -0.01, "conciseness": 0.02}, "P(pi>mu)": {"helpfulness": 0.499, "correctness": 0.498, "coherence": 0.495, "conciseness": 0.51}, "w_nbs": {"helpfulness": 0.333, "correctness": 0.333, "coherence": 0.333, "conciseness": 0.0}, "w_ks": {"helpfulness": 0.269, "correctness": 0.269, "coherence": 0.269, "conciseness": 0.194}, "worst": "coherence", "n": 5000}
RMPOOL_helpsteer2_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep helpsteer2 EXP=exp/rmpool_mbhthelp443] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/gtpref/helpsteer2/exp/rmpool_mbhthelp442/train/htmnpo_helpfulness

Map:   0%|          | 0/2250 [00:00<?, ? examples/s]
Map:  89%|████████▉ | 2000/2250 [00:00<00:00, 17983.36 examples/s]
Map: 100%|██████████| 2250/2250 [00:00<00:00, 17567.78 examples/s]

Map:   0%|          | 0/250 [00:00<?, ? examples/s]
Map: 100%|██████████| 250/250 [00:00<00:00, 13176.38 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/2250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 212913.64 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 211302.32 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 52843.62 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 52033.35 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/2250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 269295.29 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 265350.32 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 58328.75 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 57217.94 examples/s]
[prep] pre_pref + pre_htobj built
dpo htmnpo_coherence htmnpo_conciseness htmnpo_correctness htmnpo_helpfulness inpo ipo ks maxmin nbs simpo unif 
TRAIN_DONE aiprlab-polymer
===== MB round-4 task=helpsteer2 anchor=/NHNHOME/AIPR/sjkim/gtpref/helpsteer2/exp/rmpool_mbhthelp443/train/htmnpo_helpfulness weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
manifest 2500
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": -0.018, "correctness": -0.003, "coherence": -0.014, "conciseness": -0.014}, "P(pi>mu)": {"helpfulness": 0.491, "correctness": 0.498, "coherence": 0.493, "conciseness": 0.493}, "w_nbs": {"helpfulness": 0.25, "correctness": 0.25, "coherence": 0.25, "conciseness": 0.25}, "w_ks": {"helpfulness": 0.25, "correctness": 0.25, "coherence": 0.25, "conciseness": 0.25}, "worst": "helpfulness", "n": 5000}
RMPOOL_helpsteer2_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep helpsteer2 EXP=exp/rmpool_mbhthelp444] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/gtpref/helpsteer2/exp/rmpool_mbhthelp443/train/htmnpo_helpfulness

Map:   0%|          | 0/2250 [00:00<?, ? examples/s]
Map:  89%|████████▉ | 2000/2250 [00:00<00:00, 17185.83 examples/s]
Map: 100%|██████████| 2250/2250 [00:00<00:00, 16692.99 examples/s]

Map:   0%|          | 0/250 [00:00<?, ? examples/s]
Map: 100%|██████████| 250/250 [00:00<00:00, 13610.09 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/2250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 202275.94 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 200680.13 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 52875.60 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 52064.35 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/2250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 300940.21 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 297843.90 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 59251.62 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 58166.97 examples/s]
[prep] pre_pref + pre_htobj built
dpo htmnpo_coherence htmnpo_conciseness htmnpo_correctness htmnpo_helpfulness inpo ipo ks maxmin nbs simpo unif 
TRAIN_DONE aiprlab-polymer
===== MB round-5 task=helpsteer2 anchor=/NHNHOME/AIPR/sjkim/gtpref/helpsteer2/exp/rmpool_mbhthelp444/train/htmnpo_helpfulness weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
manifest 2500
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": 0.012, "correctness": 0.01, "coherence": 0.01, "conciseness": -0.022}, "P(pi>mu)": {"helpfulness": 0.506, "correctness": 0.505, "coherence": 0.505, "conciseness": 0.489}, "w_nbs": {"helpfulness": 0.0, "correctness": 0.0, "coherence": 0.0, "conciseness": 0.999}, "w_ks": {"helpfulness": 0.232, "correctness": 0.242, "coherence": 0.241, "conciseness": 0.285}, "worst": "conciseness", "n": 5000}
RMPOOL_helpsteer2_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep helpsteer2 EXP=exp/rmpool_mbhthelp445] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/gtpref/helpsteer2/exp/rmpool_mbhthelp444/train/htmnpo_helpfulness

Map:   0%|          | 0/2250 [00:00<?, ? examples/s]
Map:  89%|████████▉ | 2000/2250 [00:00<00:00, 18238.72 examples/s]
Map: 100%|██████████| 2250/2250 [00:00<00:00, 17623.19 examples/s]

Map:   0%|          | 0/250 [00:00<?, ? examples/s]
Map: 100%|██████████| 250/250 [00:00<00:00, 13638.77 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/2250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 210369.68 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 208810.36 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 53079.02 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 52248.54 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/2250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 300834.68 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 297749.93 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 59483.55 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 58475.13 examples/s]
[prep] pre_pref + pre_htobj built
dpo htmnpo_coherence htmnpo_conciseness htmnpo_correctness htmnpo_helpfulness inpo ipo ks maxmin nbs simpo unif 
TRAIN_DONE aiprlab-polymer
===== MB round-6 task=helpsteer2 anchor=/NHNHOME/AIPR/sjkim/gtpref/helpsteer2/exp/rmpool_mbhthelp445/train/htmnpo_helpfulness weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
manifest 2500
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": 0.02, "correctness": 0.021, "coherence": 0.012, "conciseness": -0.011}, "P(pi>mu)": {"helpfulness": 0.51, "correctness": 0.511, "coherence": 0.506, "conciseness": 0.494}, "w_nbs": {"helpfulness": 0.0, "correctness": 0.0, "coherence": 0.0, "conciseness": 1.0}, "w_ks": {"helpfulness": 0.222, "correctness": 0.216, "coherence": 0.254, "conciseness": 0.308}, "worst": "conciseness", "n": 5000}
RMPOOL_helpsteer2_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep helpsteer2 EXP=exp/rmpool_mbhthelp446] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/gtpref/helpsteer2/exp/rmpool_mbhthelp445/train/htmnpo_helpfulness

Map:   0%|          | 0/2250 [00:00<?, ? examples/s]
Map:  89%|████████▉ | 2000/2250 [00:00<00:00, 18014.91 examples/s]
Map: 100%|██████████| 2250/2250 [00:00<00:00, 17279.63 examples/s]

Map:   0%|          | 0/250 [00:00<?, ? examples/s]
Map: 100%|██████████| 250/250 [00:00<00:00, 13136.76 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/2250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 204188.50 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 202636.43 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 49641.43 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 48718.86 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/2250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 253292.82 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 251109.15 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 52623.51 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 51817.36 examples/s]
[prep] pre_pref + pre_htobj built
dpo htmnpo_coherence htmnpo_conciseness htmnpo_correctness htmnpo_helpfulness inpo ipo ks maxmin nbs simpo unif 
TRAIN_DONE aiprlab-polymer
===== MB round-7 task=helpsteer2 anchor=/NHNHOME/AIPR/sjkim/gtpref/helpsteer2/exp/rmpool_mbhthelp446/train/htmnpo_helpfulness weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
manifest 2500
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": 0.029, "correctness": 0.025, "coherence": 0.026, "conciseness": -0.017}, "P(pi>mu)": {"helpfulness": 0.514, "correctness": 0.512, "coherence": 0.513, "conciseness": 0.492}, "w_nbs": {"helpfulness": 0.0, "correctness": 0.0, "coherence": 0.0, "conciseness": 1.0}, "w_ks": {"helpfulness": 0.211, "correctness": 0.226, "coherence": 0.221, "conciseness": 0.342}, "worst": "conciseness", "n": 5000}
RMPOOL_helpsteer2_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep helpsteer2 EXP=exp/rmpool_mbhthelp447] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/gtpref/helpsteer2/exp/rmpool_mbhthelp446/train/htmnpo_helpfulness

Map:   0%|          | 0/2250 [00:00<?, ? examples/s]
Map:  89%|████████▉ | 2000/2250 [00:00<00:00, 17909.60 examples/s]
Map: 100%|██████████| 2250/2250 [00:00<00:00, 17363.88 examples/s]

Map:   0%|          | 0/250 [00:00<?, ? examples/s]
Map: 100%|██████████| 250/250 [00:00<00:00, 13570.29 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/2250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 209370.90 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 207821.71 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 53092.46 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 52277.20 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/2250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 299355.56 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 296124.26 examples/s]

Saving the dataset (0/1 shards):   0%|          | 0/250 [00:00<?, ? examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 60548.33 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 59520.69 examples/s]
[prep] pre_pref + pre_htobj built
dpo htmnpo_coherence htmnpo_conciseness htmnpo_correctness htmnpo_helpfulness inpo ipo ks maxmin nbs simpo unif 
TRAIN_DONE aiprlab-polymer
{
  "label": "qwen32_helpsteer2_htmnpo_helpfulness",
  "n_objectives": 4,
  "surplus": {
    "helpfulness": 0.1094,
    "correctness": 0.0775,
    "coherence": 0.0777,
    "conciseness": 0.0885
  },
  "hoeffding_margin": {
    "helpfulness": 0.0357,
    "correctness": 0.0359,
    "coherence": 0.0356,
    "conciseness": 0.0356
  },
  "worst_surplus": 0.0775,
  "individual_rationality": true,
  "nash_welfare_logsum": -9.7494
}
EVAL_helpsteer2_DONE aiprlab-polymer
MB_STAGES_DONE