queue_gm2/ghc.log

===== MB round-1 task=hs2gemma anchor=/NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/ weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": -0.028, "correctness": -0.027, "coherence": -0.019, "conciseness": 0.002}, "P(pi>mu)": {"helpfulness": 0.486, "correctness": 0.486, "coherence": 0.491, "conciseness": 0.501}, "w_nbs": {"helpfulness": 0.333, "correctness": 0.333, "coherence": 0.333, "conciseness": 0.0}, "w_ks": {"helpfulness": 0.252, "correctness": 0.252, "coherence": 0.252, "conciseness": 0.245}, "worst": "helpfulness", "n": 5000}
RMPOOL_hs2gemma_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep hs2gemma EXP=exp/rmpool_mbghc1] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/

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, 94789.86 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 94238.96 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, 24589.64 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 24369.05 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, 104043.75 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 103641.54 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, 26506.64 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 26288.01 examples/s]
[prep] pre_pref + pre_htobj built
dpo htmnpo_coherence htmnpo_conciseness htmnpo_correctness htmnpo_helpfulness inpo ipo ks maxmin nbs simpo unif 
/NHNHOME/AIPR/sjkim/gtpref/train_gpu.sh: line 17: syntax error near unexpected token `)'
/NHNHOME/AIPR/sjkim/gtpref/train_gpu.sh: line 17: `(g+1)); [ $((g%NGPU)) -eq 0 ] && wait'
MB1_TRAIN_FAIL
===== MB round-1 task=hs2gemma anchor=/NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/ weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": -0.028, "correctness": -0.027, "coherence": -0.019, "conciseness": 0.002}, "P(pi>mu)": {"helpfulness": 0.486, "correctness": 0.486, "coherence": 0.491, "conciseness": 0.501}, "w_nbs": {"helpfulness": 0.333, "correctness": 0.333, "coherence": 0.333, "conciseness": 0.0}, "w_ks": {"helpfulness": 0.252, "correctness": 0.252, "coherence": 0.252, "conciseness": 0.245}, "worst": "helpfulness", "n": 5000}
RMPOOL_hs2gemma_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep hs2gemma EXP=exp/rmpool_mbghc1] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/

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, 94998.83 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 94459.69 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, 24972.04 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 24751.58 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, 99454.98 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 99052.05 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, 26082.03 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 25858.20 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
MB1_TRAIN_FAIL
===== MB round-1 task=hs2gemma anchor=/NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/ weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": -0.028, "correctness": -0.027, "coherence": -0.019, "conciseness": 0.002}, "P(pi>mu)": {"helpfulness": 0.486, "correctness": 0.486, "coherence": 0.491, "conciseness": 0.501}, "w_nbs": {"helpfulness": 0.333, "correctness": 0.333, "coherence": 0.333, "conciseness": 0.0}, "w_ks": {"helpfulness": 0.252, "correctness": 0.252, "coherence": 0.252, "conciseness": 0.245}, "worst": "helpfulness", "n": 5000}
RMPOOL_hs2gemma_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep hs2gemma EXP=exp/rmpool_mbghc1] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/

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, 91859.48 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 91329.65 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, 23684.86 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 23501.71 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, 104208.04 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 103812.55 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, 25259.59 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 25060.97 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
MB1_TRAIN_FAIL
===== MB round-1 task=hs2gemma anchor=/NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/ weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": -0.028, "correctness": -0.027, "coherence": -0.019, "conciseness": 0.002}, "P(pi>mu)": {"helpfulness": 0.486, "correctness": 0.486, "coherence": 0.491, "conciseness": 0.501}, "w_nbs": {"helpfulness": 0.333, "correctness": 0.333, "coherence": 0.333, "conciseness": 0.0}, "w_ks": {"helpfulness": 0.252, "correctness": 0.252, "coherence": 0.252, "conciseness": 0.245}, "worst": "helpfulness", "n": 5000}
RMPOOL_hs2gemma_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep hs2gemma EXP=exp/rmpool_mbghc1] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/

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, 85266.26 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 84849.21 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, 24684.57 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 24474.28 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, 104090.80 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 103689.37 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, 26096.31 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 25883.10 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
MB1_TRAIN_FAIL
===== MB round-1 task=hs2gemma anchor=/NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/ weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": -0.028, "correctness": -0.027, "coherence": -0.019, "conciseness": 0.002}, "P(pi>mu)": {"helpfulness": 0.486, "correctness": 0.486, "coherence": 0.491, "conciseness": 0.501}, "w_nbs": {"helpfulness": 0.333, "correctness": 0.333, "coherence": 0.333, "conciseness": 0.0}, "w_ks": {"helpfulness": 0.252, "correctness": 0.252, "coherence": 0.252, "conciseness": 0.245}, "worst": "helpfulness", "n": 5000}
RMPOOL_hs2gemma_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep hs2gemma EXP=exp/rmpool_mbghc1] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/

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, 94144.95 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 93668.39 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, 24172.99 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 23981.15 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, 103004.66 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 102622.71 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, 25849.28 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 25647.59 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
MB1_TRAIN_FAIL
===== MB round-1 task=hs2gemma anchor=/NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/ weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": -0.028, "correctness": -0.027, "coherence": -0.019, "conciseness": 0.002}, "P(pi>mu)": {"helpfulness": 0.486, "correctness": 0.486, "coherence": 0.491, "conciseness": 0.501}, "w_nbs": {"helpfulness": 0.333, "correctness": 0.333, "coherence": 0.333, "conciseness": 0.0}, "w_ks": {"helpfulness": 0.252, "correctness": 0.252, "coherence": 0.252, "conciseness": 0.245}, "worst": "helpfulness", "n": 5000}
RMPOOL_hs2gemma_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep hs2gemma EXP=exp/rmpool_mbghc1] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/

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, 85619.01 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 85237.76 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, 24746.91 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 24547.04 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, 98380.86 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 98040.52 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, 23595.85 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 23424.01 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
MB1_TRAIN_FAIL
===== MB round-1 task=hs2gemma anchor=/NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/ weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": -0.028, "correctness": -0.027, "coherence": -0.019, "conciseness": 0.002}, "P(pi>mu)": {"helpfulness": 0.486, "correctness": 0.486, "coherence": 0.491, "conciseness": 0.501}, "w_nbs": {"helpfulness": 0.333, "correctness": 0.333, "coherence": 0.333, "conciseness": 0.0}, "w_ks": {"helpfulness": 0.252, "correctness": 0.252, "coherence": 0.252, "conciseness": 0.245}, "worst": "helpfulness", "n": 5000}
RMPOOL_hs2gemma_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep hs2gemma EXP=exp/rmpool_mbghc1] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/

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, 91227.24 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 90776.19 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, 24140.71 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 23949.39 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, 97126.34 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 96792.62 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, 24496.58 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 24314.24 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
MB1_TRAIN_FAIL
===== MB round-1 task=hs2gemma anchor=/NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/ weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
allresp built
scores combined {'pol_42': 2500, 'pol_43': 2500, 'base_44': 2500, 'base_45': 2500}
{"surplus(2P-1)": {"helpfulness": -0.028, "correctness": -0.027, "coherence": -0.019, "conciseness": 0.002}, "P(pi>mu)": {"helpfulness": 0.486, "correctness": 0.486, "coherence": 0.491, "conciseness": 0.501}, "w_nbs": {"helpfulness": 0.333, "correctness": 0.333, "coherence": 0.333, "conciseness": 0.0}, "w_ks": {"helpfulness": 0.252, "correctness": 0.252, "coherence": 0.252, "conciseness": 0.245}, "worst": "helpfulness", "n": 5000}
RMPOOL_hs2gemma_DONE
[build_bon] 2500 pairs (train 2250/test 250) weights {'helpfulness': 0.25, 'correctness': 0.25, 'coherence': 0.25, 'conciseness': 0.25}
[prep hs2gemma EXP=exp/rmpool_mbghc1] objs: helpfulness correctness coherence conciseness parent: /NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/

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, 94047.38 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 93530.07 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, 24244.53 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 24043.29 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, 105429.26 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 2250/2250 [00:00<00:00, 104984.75 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, 26432.47 examples/s]
Saving the dataset (1/1 shards): 100%|██████████| 250/250 [00:00<00:00, 26191.48 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
MB1_TRAIN_FAIL
===== MB round-1 task=hs2gemma anchor=/NHNHOME/AIPR/sjkim/policy_bases/models--google--gemma-2-9b-it/snapshots/11c9b309abf73637e4b6f9a3fa1e92e615547819/ weights={"helpfulness":0.2500,"correctness":0.2500,"coherence":0.2500,"conciseness":0.2500} =====
allresp built