Commit History

README: final B200 continuation checkpoints
f8b89f7
verified

cpyang commited on

add verl_ckpt/v29_C_taxv3_plus_cf_sft_4b_s42_balance_noloc_sdar_off_resume120_b200_v28/global_step_560 (B200 continuation to its end, best saved step)
8c4ec9c
verified

cpyang commited on

add rl_balance_noloc_sdar_off_step560 (B200 continuation to its end, best saved step)
49d92f2
verified

cpyang commited on

add verl_ckpt/v29_C_sft1e5_4b_s42_balance_noloc_resume60_b200_v28/global_step_220 (B200 continuation to its end, best saved step)
b23135d
verified

cpyang commited on

add rl_sft1e5_balance_noloc_step220 (B200 continuation to its end, best saved step)
a18260a
verified

cpyang commited on

add verl_ckpt/v29_C_taxv3_plus_cf_taxv3om_4b_s42_balance_noloc_resume270_b200_v28/global_step_1800 (B200 continuation to its end, best saved step)
12da06e
verified

cpyang commited on

add rl_balance_noloc_taxv3om_step1800 (B200 continuation to its end, best saved step)
3227c84
verified

cpyang commited on

README: B200 continuation checkpoints
3245fb4
verified

cpyang commited on

add verl_ckpt/v29_C_taxv3_plus_cf_taxv3om_4b_s42_balance_noloc_resume270_b200_v28/global_step_480
56fdce1
verified

cpyang commited on

add rl_balance_noloc_taxv3om_step480 (B200 continuation, best of resume)
dbb869f
verified

cpyang commited on

README: B200 continuation checkpoints
b5e92eb
verified

cpyang commited on

add verl_ckpt/v29_C_sft1e5_4b_s42_balance_noloc_resume60_b200_v28/global_step_180
520ca8b
verified

cpyang commited on

add rl_sft1e5_balance_noloc_step180 (B200 continuation, best of resume)
44c1996
verified

cpyang commited on

README: B200 continuation checkpoints
5c5e539
verified

cpyang commited on

add verl_ckpt/v29_C_taxv3_plus_cf_sft_4b_s42_balance_noloc_sdar_off_resume120_b200_v28/global_step_280
bfee23a
verified

cpyang commited on

add rl_balance_noloc_sdar_off_step280 (B200 continuation, best of resume)
5a60a35
verified

cpyang commited on

add verl_ckpt/v29_C_taxv3_plus_cf_floc_sft_4b_s42_balance_floc_resume160_b200_v28/global_step_380
1f8540d
verified

cpyang commited on

add rl_balance_floc_step380 (B200 continuation, best of resume)
b3e403a
verified

cpyang commited on

README: B200 continuation checkpoints
5e1d192
verified

cpyang commited on

add verl_ckpt/v29_C_sft1e5_4b_s42_balance_noloc_resume60_b200_v28/global_step_120
6a68f84
verified

cpyang commited on

add rl_sft1e5_balance_noloc_step120 (B200 continuation, best of resume)
112f9fb
verified

cpyang commited on

add verl_ckpt/v29_C_taxv3_plus_cf_floc_sft_4b_s42_balance_floc_resume160_b200_v28/global_step_240
aef6f74
verified

cpyang commited on

add rl_balance_floc_step240 (B200 continuation, best of resume)
0aad144
verified

cpyang commited on

add verl_ckpt/v29_C_taxv3_plus_cf_sft_4b_s42_balance_noloc_fullhist_resume140_b200_v28/global_step_260
b689ee4
verified

cpyang commited on

add rl_balance_noloc_fullhist_step260 (B200 continuation, best of resume)
3a32dda
verified

cpyang commited on

Add raw verl checkpoint v29_C_taxv3_plus_cf_sft_4b_s42_balance_noloc_sdar_off_v28 global_step_120 (FSDP shards + optimizer, resumable)
57ff31c
verified

cpyang commited on

Add raw verl checkpoint v29_C_sft1e5_4b_s42_balance_noloc_v28 global_step_60 (FSDP shards + optimizer, resumable)
18915aa
verified

cpyang commited on

verl RL checkpoint global_step_270 (M_taxv3om balance-noloc): actor FSDP shards + optimizer + dataloader state
7ec530d
verified

cpyang commited on

Add raw verl checkpoint v29_C_taxv3_plus_cf_floc_sft_4b_s42_balance_floc_v28 global_step_160 (FSDP shards + optimizer, resumable)
0e01451
verified

cpyang commited on

Add raw verl checkpoint v29_C_taxv3_plus_cf_sft_4b_s42_balance_noloc_fullhist_v28 global_step_140 (FSDP shards + optimizer, resumable)
fe4eae5
verified

cpyang commited on

Add rl_balance_noloc_sdar_off_step120: best RL checkpoint by val-key-askthreshold j/macro_f1 (Duke RTX Pro 6000 run)
3208d92
verified

cpyang commited on

Add rl_sft1e5_balance_noloc_step60: best RL checkpoint by val-key-askthreshold j/macro_f1 (Duke RTX Pro 6000 run)
719d64c
verified

cpyang commited on

Add rl_balance_floc_step160: best RL checkpoint by val-key-askthreshold j/macro_f1 (Duke RTX Pro 6000 run)
9b83c9d
verified

cpyang commited on

Add rl_balance_noloc_fullhist_step140: best RL checkpoint by val-key-askthreshold j/macro_f1 (Duke RTX Pro 6000 run)
3092e43
verified

cpyang commited on

Add model card
8c51f64
verified

cpyang commited on

initial commit
d4e6a39
verified

cpyang commited on