Download 400M/trainer_state.json from SynthCode-Project/rv2-gemma12b-g2_3_plan_oldcode_newcode: direct link, hf CLI and curl.
- Browser
- Download file 70.4 kB
-
https://huggingface.co/SynthCode-Project/rv2-gemma12b-g2_3_plan_oldcode_newcode/resolve/main/400M/trainer_state.json
- Command line
-
hf download hf://SynthCode-Project/rv2-gemma12b-g2_3_plan_oldcode_newcode/400M/trainer_state.json
-
curl -L -o trainer_state.json https://huggingface.co/SynthCode-Project/rv2-gemma12b-g2_3_plan_oldcode_newcode/resolve/main/400M/trainer_state.json
70.4 kB
| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.11159551424419402, | |
| "eval_steps": 500, | |
| "global_step": 4070, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0002741904526884374, | |
| "grad_norm": 533.0549778145878, | |
| "learning_rate": 2.910112359550562e-06, | |
| "loss": 2.0392507553100585, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0005483809053768748, | |
| "grad_norm": 129.62802977018143, | |
| "learning_rate": 3.921348314606742e-06, | |
| "loss": 1.1088291168212892, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0008225713580653121, | |
| "grad_norm": 37.028700378797375, | |
| "learning_rate": 4.932584269662922e-06, | |
| "loss": 0.8763736724853516, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0010967618107537496, | |
| "grad_norm": 11.415105214785655, | |
| "learning_rate": 5.943820224719102e-06, | |
| "loss": 0.8334451675415039, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0013709522634421868, | |
| "grad_norm": 168.0092220161655, | |
| "learning_rate": 6.955056179775282e-06, | |
| "loss": 0.841560173034668, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0016451427161306242, | |
| "grad_norm": 10.9544582922062, | |
| "learning_rate": 7.966292134831462e-06, | |
| "loss": 0.7757131099700928, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0019193331688190617, | |
| "grad_norm": 52.12017293008163, | |
| "learning_rate": 8.977528089887641e-06, | |
| "loss": 0.7379593849182129, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0021935236215074993, | |
| "grad_norm": 24.23368291019739, | |
| "learning_rate": 9.988764044943822e-06, | |
| "loss": 0.871122169494629, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0024677140741959365, | |
| "grad_norm": 44.606225719656116, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 0.8103988647460938, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0027419045268843737, | |
| "grad_norm": 19.394878252919106, | |
| "learning_rate": 1.2011235955056182e-05, | |
| "loss": 0.8088878631591797, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0030160949795728113, | |
| "grad_norm": 154.65284754287825, | |
| "learning_rate": 1.3022471910112362e-05, | |
| "loss": 0.7940959930419922, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0032902854322612485, | |
| "grad_norm": 58.453168158560985, | |
| "learning_rate": 1.403370786516854e-05, | |
| "loss": 0.8113164901733398, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.003564475884949686, | |
| "grad_norm": 121.71906195134056, | |
| "learning_rate": 1.504494382022472e-05, | |
| "loss": 0.802008056640625, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0038386663376381233, | |
| "grad_norm": 7.063904217729592, | |
| "learning_rate": 1.60561797752809e-05, | |
| "loss": 0.7886086940765381, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.004112856790326561, | |
| "grad_norm": 5.886231107848548, | |
| "learning_rate": 1.706741573033708e-05, | |
| "loss": 0.8553682327270508, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.004387047243014999, | |
| "grad_norm": 17.479053516912124, | |
| "learning_rate": 1.8078651685393256e-05, | |
| "loss": 0.878786849975586, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.004661237695703435, | |
| "grad_norm": 15.809884339669948, | |
| "learning_rate": 1.908988764044944e-05, | |
| "loss": 0.8892171859741211, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.004935428148391873, | |
| "grad_norm": 15.552804520715554, | |
| "learning_rate": 2e-05, | |
| "loss": 0.907989501953125, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.005209618601080311, | |
| "grad_norm": 8.234516558585694, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8959874153137207, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.005483809053768747, | |
| "grad_norm": 47.07479845477993, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7972547054290772, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.005757999506457185, | |
| "grad_norm": 14.678891090428264, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7945639610290527, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.006032189959145623, | |
| "grad_norm": 17.99623194706809, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8467901229858399, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.00630638041183406, | |
| "grad_norm": 51.95652459415379, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7964937210083007, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.006580570864522497, | |
| "grad_norm": 10.738865490640423, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7535751342773438, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.006854761317210935, | |
| "grad_norm": 10.795608734038868, | |
| "learning_rate": 2e-05, | |
| "loss": 0.828792953491211, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.007128951769899372, | |
| "grad_norm": 14.551689425739292, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8123558044433594, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.00740314222258781, | |
| "grad_norm": 26.532992867443514, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8517896652221679, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.007677332675276247, | |
| "grad_norm": 34.07990246963607, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8389625549316406, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.007951523127964684, | |
| "grad_norm": 27.19116883891022, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8190082550048828, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.008225713580653122, | |
| "grad_norm": 16.718638113526183, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8008914947509765, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.00849990403334156, | |
| "grad_norm": 27.009662680205103, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8091920852661133, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.008774094486029997, | |
| "grad_norm": 9.031648650732553, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8081918716430664, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.009048284938718433, | |
| "grad_norm": 15.927828911998665, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7878878593444825, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.00932247539140687, | |
| "grad_norm": 6.019580282287438, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8092373847961426, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.009596665844095308, | |
| "grad_norm": 10.097656044283081, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8236124038696289, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.009870856296783746, | |
| "grad_norm": 3.898510274122292, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8201662063598633, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.010145046749472184, | |
| "grad_norm": 10.433240846124043, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7954160213470459, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.010419237202160621, | |
| "grad_norm": 3.9109175859167626, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8451309204101562, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.010693427654849059, | |
| "grad_norm": 19.47785359214917, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8227203369140625, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.010967618107537495, | |
| "grad_norm": 1.9979422322935754, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7947024345397949, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.011241808560225932, | |
| "grad_norm": 2.1152537387703063, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7340120315551758, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.01151599901291437, | |
| "grad_norm": 2.457083494187141, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7300191879272461, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.011790189465602808, | |
| "grad_norm": 17.457321034227927, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7896897315979003, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.012064379918291245, | |
| "grad_norm": 3.7757446450653047, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7420886039733887, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.012338570370979683, | |
| "grad_norm": 10.718659609403895, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7464280605316163, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.01261276082366812, | |
| "grad_norm": 13.096467752661693, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7420991420745849, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.012886951276356558, | |
| "grad_norm": 7.554114077938612, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7896678924560547, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.013161141729044994, | |
| "grad_norm": 1.9610155148576782, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7641692161560059, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.013435332181733432, | |
| "grad_norm": 3.743726960430974, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7724472999572753, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.01370952263442187, | |
| "grad_norm": 3.6205794374072897, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7726649284362793, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.013983713087110307, | |
| "grad_norm": 2.751894400685354, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7311810970306396, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.014257903539798744, | |
| "grad_norm": 15.8354606407861, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7479830741882324, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.014532093992487182, | |
| "grad_norm": 9.741291005928273, | |
| "learning_rate": 2e-05, | |
| "loss": 0.758505916595459, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.01480628444517562, | |
| "grad_norm": 6.361558235058739, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7497701644897461, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.015080474897864056, | |
| "grad_norm": 12.406630096904701, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7578043937683105, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.015354665350552493, | |
| "grad_norm": 5.437533111150059, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7494767665863037, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.015628855803240933, | |
| "grad_norm": 5.385414434765143, | |
| "learning_rate": 2e-05, | |
| "loss": 0.76002197265625, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.01590304625592937, | |
| "grad_norm": 2.514872003150109, | |
| "learning_rate": 2e-05, | |
| "loss": 0.753857946395874, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.016177236708617804, | |
| "grad_norm": 7.350990110651167, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7598513603210449, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.016451427161306244, | |
| "grad_norm": 9.359850474100211, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7978845596313476, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.01672561761399468, | |
| "grad_norm": 2.5927971608846243, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7353355407714843, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.01699980806668312, | |
| "grad_norm": 13.580441280433185, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7530646800994873, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.017273998519371555, | |
| "grad_norm": 7.032504711444772, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7999831199645996, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.017548188972059994, | |
| "grad_norm": 3.7623518166414134, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7583576202392578, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.01782237942474843, | |
| "grad_norm": 9.996121547764488, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7126437187194824, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.018096569877436866, | |
| "grad_norm": 3.8005663709948423, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7714213371276856, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.018370760330125305, | |
| "grad_norm": 2.2803953101797445, | |
| "learning_rate": 2e-05, | |
| "loss": 0.735899829864502, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.01864495078281374, | |
| "grad_norm": 2.2804717033812687, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7463047981262207, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.01891914123550218, | |
| "grad_norm": 2.6630744686256174, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7112466812133789, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.019193331688190617, | |
| "grad_norm": 8.59421748677637, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8032501220703125, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.019467522140879056, | |
| "grad_norm": 6.839099911030037, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7257285118103027, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.019741712593567492, | |
| "grad_norm": 6.527196880605798, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7394065856933594, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.020015903046255928, | |
| "grad_norm": 4.91546686150854, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7233449459075928, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.020290093498944367, | |
| "grad_norm": 9.615722695591046, | |
| "learning_rate": 2e-05, | |
| "loss": 0.714713191986084, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.020564283951632803, | |
| "grad_norm": 2.6469047522209244, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7250522613525391, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.020838474404321242, | |
| "grad_norm": 27.513230581702402, | |
| "learning_rate": 2e-05, | |
| "loss": 0.709619665145874, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.021112664857009678, | |
| "grad_norm": 2.3305572520941102, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7650126457214356, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.021386855309698118, | |
| "grad_norm": 1.7645473455518579, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7501170635223389, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.021661045762386553, | |
| "grad_norm": 2.706703322701269, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7390304565429687, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.02193523621507499, | |
| "grad_norm": 1.343452945442269, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7560922145843506, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.02220942666776343, | |
| "grad_norm": 1.8131674161495455, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7617329597473145, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.022483617120451865, | |
| "grad_norm": 6.6257304148619, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7030315399169922, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.022757807573140304, | |
| "grad_norm": 2.7477972541714237, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7051456451416016, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.02303199802582874, | |
| "grad_norm": 1.8551689554260649, | |
| "learning_rate": 2e-05, | |
| "loss": 0.790980863571167, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.02330618847851718, | |
| "grad_norm": 4.091035715965749, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7167482376098633, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.023580378931205615, | |
| "grad_norm": 3.13444342764459, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7428287506103516, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.023854569383894055, | |
| "grad_norm": 7.428502868185567, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7720952510833741, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.02412875983658249, | |
| "grad_norm": 2.738880412295684, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7307533264160156, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.024402950289270926, | |
| "grad_norm": 5.442319700482665, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7590367794036865, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.024677140741959366, | |
| "grad_norm": 4.209775631505361, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7398377418518066, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.0249513311946478, | |
| "grad_norm": 3.317366374639838, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7573592662811279, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.02522552164733624, | |
| "grad_norm": 2.21708523073645, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7442788124084473, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.025499712100024677, | |
| "grad_norm": 5.003510596258695, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7208542823791504, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.025773902552713116, | |
| "grad_norm": 1.8666696979850983, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7219260215759278, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.026048093005401552, | |
| "grad_norm": 2.319707094464392, | |
| "learning_rate": 2e-05, | |
| "loss": 0.705035400390625, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.026322283458089988, | |
| "grad_norm": 1.559433923602148, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7253781795501709, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.026596473910778427, | |
| "grad_norm": 3.1103622605289294, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7434506416320801, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.026870664363466863, | |
| "grad_norm": 4.482952591971397, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7423406600952148, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.027144854816155303, | |
| "grad_norm": 2.349643537640656, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7380632400512696, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.02741904526884374, | |
| "grad_norm": 6.624084473754211, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7448741912841796, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.027693235721532178, | |
| "grad_norm": 1.959542744744174, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7620717048645019, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.027967426174220614, | |
| "grad_norm": 1.9359370487542638, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7283543586730957, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.02824161662690905, | |
| "grad_norm": 3.7510091510273424, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7341510772705078, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.02851580707959749, | |
| "grad_norm": 1.645233108914731, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7321323394775391, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.028789997532285925, | |
| "grad_norm": 5.060129114701537, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7923129558563232, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.029064187984974364, | |
| "grad_norm": 6.767343077930711, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7364850997924804, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.0293383784376628, | |
| "grad_norm": 7.157354259172915, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7175948143005371, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.02961256889035124, | |
| "grad_norm": 10.118526445754803, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6956705093383789, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.029886759343039675, | |
| "grad_norm": 3.2975128574851205, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7435084342956543, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.03016094979572811, | |
| "grad_norm": 3.79438108252297, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7209542274475098, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.03043514024841655, | |
| "grad_norm": 3.6413783178286896, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7462982177734375, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.030709330701104987, | |
| "grad_norm": 2.4359365603997323, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7772771835327148, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.030983521153793426, | |
| "grad_norm": 2.067253539210089, | |
| "learning_rate": 2e-05, | |
| "loss": 0.754563283920288, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.031257711606481865, | |
| "grad_norm": 1.3603842511083, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7163387298583984, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.0315319020591703, | |
| "grad_norm": 7.634204958345298, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7153648376464844, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.03180609251185874, | |
| "grad_norm": 2.224287938133906, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7743209838867188, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.03208028296454717, | |
| "grad_norm": 8.328106347948369, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7125131130218506, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.03235447341723561, | |
| "grad_norm": 4.9140382117559716, | |
| "learning_rate": 2e-05, | |
| "loss": 0.734964656829834, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.03262866386992405, | |
| "grad_norm": 1.6437427011396712, | |
| "learning_rate": 2e-05, | |
| "loss": 0.736287260055542, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.03290285432261249, | |
| "grad_norm": 1.7984413046378422, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6888255119323731, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.03317704477530092, | |
| "grad_norm": 3.003273429621072, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7354516506195068, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.03345123522798936, | |
| "grad_norm": 1.581766516298146, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7175889492034913, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.0337254256806778, | |
| "grad_norm": 2.3247470513343185, | |
| "learning_rate": 2e-05, | |
| "loss": 0.714189863204956, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.03399961613336624, | |
| "grad_norm": 7.9334494235192325, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7136695861816407, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.034273806586054674, | |
| "grad_norm": 49.62487496955337, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7086945533752441, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.03454799703874311, | |
| "grad_norm": 2.890774584879648, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7269087791442871, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.034822187491431546, | |
| "grad_norm": 4.796531005737684, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7484929084777832, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.03509637794411999, | |
| "grad_norm": 1.6372708148105157, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6999114036560059, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.035370568396808424, | |
| "grad_norm": 4.162341191850801, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7129648208618165, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.03564475884949686, | |
| "grad_norm": 4.017056556093628, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7217267036437989, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.035918949302185296, | |
| "grad_norm": 9.786655951652824, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7361152648925782, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.03619313975487373, | |
| "grad_norm": 17.798888728639277, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7423253059387207, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.036467330207562175, | |
| "grad_norm": 25.98230175781132, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7079302787780761, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.03674152066025061, | |
| "grad_norm": 4.107909151191186, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7637813568115235, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.03701571111293905, | |
| "grad_norm": 1.8177555687046802, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7440331459045411, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.03728990156562748, | |
| "grad_norm": 6.631474920460703, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7420357704162598, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.037564092018315925, | |
| "grad_norm": 1.182591249805126, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7243555068969727, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.03783828247100436, | |
| "grad_norm": 1.5239290367740697, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7780046463012695, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.0381124729236928, | |
| "grad_norm": 1.9611702789468477, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7606742858886719, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.03838666337638123, | |
| "grad_norm": 3.4538398959809564, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7161018371582031, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.03866085382906967, | |
| "grad_norm": 6.74405592418129, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7269341945648193, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.03893504428175811, | |
| "grad_norm": 5.0613023509206, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7150623321533203, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.03920923473444655, | |
| "grad_norm": 15.935601654785444, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6991084098815918, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.039483425187134984, | |
| "grad_norm": 189.58206876056943, | |
| "learning_rate": 2e-05, | |
| "loss": 0.715428352355957, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.03975761563982342, | |
| "grad_norm": 13.224258279776146, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7448666572570801, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.040031806092511855, | |
| "grad_norm": 15.23890210961455, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7545663833618164, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.0403059965452003, | |
| "grad_norm": 2.089772579003186, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7143968105316162, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.040580186997888734, | |
| "grad_norm": 672.5570249806746, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6955307960510254, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.04085437745057717, | |
| "grad_norm": 2.499047720535622, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7183216094970704, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.041128567903265606, | |
| "grad_norm": 1.6820671266931866, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7442133903503418, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.04140275835595405, | |
| "grad_norm": 1.6322487359391142, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7415005683898925, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.041676948808642485, | |
| "grad_norm": 2.0768691766562974, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7350949764251709, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.04195113926133092, | |
| "grad_norm": 3.462638107939682, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6958516597747803, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.042225329714019356, | |
| "grad_norm": 1.9410342316482636, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7019001960754394, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.04249952016670779, | |
| "grad_norm": 2.2435234267896944, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7038233757019043, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.042773710619396235, | |
| "grad_norm": 50.31729516447788, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6971765041351319, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.04304790107208467, | |
| "grad_norm": 6.726153221416043, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7901406764984131, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.04332209152477311, | |
| "grad_norm": 25.290061958502374, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7404637813568116, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.04359628197746154, | |
| "grad_norm": 11.648387859593914, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7325190544128418, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.04387047243014998, | |
| "grad_norm": 3.998377511327454, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6938001155853272, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.04414466288283842, | |
| "grad_norm": 1.4609573707991412, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6974159240722656, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.04441885333552686, | |
| "grad_norm": 1.2908767847499276, | |
| "learning_rate": 2e-05, | |
| "loss": 0.735476303100586, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.04469304378821529, | |
| "grad_norm": 1.5263587301999937, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6849418640136719, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.04496723424090373, | |
| "grad_norm": 2.71012026891521, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6952274322509766, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.04524142469359217, | |
| "grad_norm": 1.5738552850120824, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7284816265106201, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.04551561514628061, | |
| "grad_norm": 1.5435257666345013, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7115949153900146, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.045789805598969044, | |
| "grad_norm": 1.516425220632835, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6966533660888672, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.04606399605165748, | |
| "grad_norm": 1.4858677693562772, | |
| "learning_rate": 2e-05, | |
| "loss": 0.712862491607666, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.046338186504345916, | |
| "grad_norm": 1.1928774431737874, | |
| "learning_rate": 2e-05, | |
| "loss": 0.715973949432373, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.04661237695703436, | |
| "grad_norm": 1.3801750667103458, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6785056114196777, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.046886567409722794, | |
| "grad_norm": 1.1562227123885702, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6806015014648438, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.04716075786241123, | |
| "grad_norm": 1.966065169897555, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6789681911468506, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.047434948315099666, | |
| "grad_norm": 1.796240102087905, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6643640041351319, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.04770913876778811, | |
| "grad_norm": 1.4714599948408273, | |
| "learning_rate": 2e-05, | |
| "loss": 0.672383975982666, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.047983329220476545, | |
| "grad_norm": 1.05613596470809, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6921384334564209, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.04825751967316498, | |
| "grad_norm": 1.168321764739896, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6853407859802246, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.04853171012585342, | |
| "grad_norm": 1.304231611987027, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6831422328948975, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.04880590057854185, | |
| "grad_norm": 1.271885101891083, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7312606811523438, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.049080091031230295, | |
| "grad_norm": 1.1274168419233672, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7195809841156006, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.04935428148391873, | |
| "grad_norm": 2.1060356897000196, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6862109184265137, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.04962847193660717, | |
| "grad_norm": 1.1668574261874323, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7068725109100342, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.0499026623892956, | |
| "grad_norm": 3.1583477159166673, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7150220394134521, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.05017685284198404, | |
| "grad_norm": 1.5843620172233683, | |
| "learning_rate": 2e-05, | |
| "loss": 0.66549072265625, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.05045104329467248, | |
| "grad_norm": 1.2547928536390052, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7375857353210449, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.05072523374736092, | |
| "grad_norm": 4.29725259279178, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6876885890960693, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.050999424200049354, | |
| "grad_norm": 1.9025925412069744, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7387830734252929, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.05127361465273779, | |
| "grad_norm": 1.5406949311604028, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6733179092407227, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.05154780510542623, | |
| "grad_norm": 20.215212215144863, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7654403686523438, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.05182199555811467, | |
| "grad_norm": 1.5669531048405803, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7350390911102295, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.052096186010803104, | |
| "grad_norm": 7.2191546562371425, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7414397716522216, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.05237037646349154, | |
| "grad_norm": 1.2325274169727398, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7655929088592529, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.052644566916179976, | |
| "grad_norm": 1.1944179026527708, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7075240135192871, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.05291875736886842, | |
| "grad_norm": 1.302822879313398, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6797014236450195, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.053192947821556855, | |
| "grad_norm": 2.0168747405099747, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7000571727752686, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.05346713827424529, | |
| "grad_norm": 1.6344897385611805, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7235137939453125, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.053741328726933726, | |
| "grad_norm": 3.5766753006414125, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7371991634368896, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.05401551917962216, | |
| "grad_norm": 1.4953479982574733, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7229987621307373, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.054289709632310605, | |
| "grad_norm": 1.6598347475638233, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6758342742919922, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.05456390008499904, | |
| "grad_norm": 1.3876361065240026, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6621166229248047, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.05483809053768748, | |
| "grad_norm": 1.2953540155757288, | |
| "learning_rate": 2e-05, | |
| "loss": 0.701081657409668, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.05511228099037591, | |
| "grad_norm": 1.0757376098134102, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6949031829833985, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.055386471443064356, | |
| "grad_norm": 1.3515745918409625, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6849285125732422, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.05566066189575279, | |
| "grad_norm": 3.4678908742290884, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6919690608978272, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.05593485234844123, | |
| "grad_norm": 1.9456705464214417, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6663742065429688, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.05620904280112966, | |
| "grad_norm": 1.2852941271459164, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6898897171020508, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.0564832332538181, | |
| "grad_norm": 7.958437531519753, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7008088588714599, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.05675742370650654, | |
| "grad_norm": 3.3113911508682685, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6721383571624756, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.05703161415919498, | |
| "grad_norm": 1.1175929950295507, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7099536895751953, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.057305804611883414, | |
| "grad_norm": 0.9164381223894205, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6982614040374756, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.05757999506457185, | |
| "grad_norm": 1.3064363870590416, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6582142829895019, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.057854185517260286, | |
| "grad_norm": 1.9295155170705978, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7130876064300538, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.05812837596994873, | |
| "grad_norm": 1.598328187450677, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7056316375732422, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.058402566422637164, | |
| "grad_norm": 1.3733360145402107, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6908935546875, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.0586767568753256, | |
| "grad_norm": 1.5835542241632976, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6900417327880859, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.058950947328014036, | |
| "grad_norm": 15.343250543565158, | |
| "learning_rate": 2e-05, | |
| "loss": 0.661806869506836, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.05922513778070248, | |
| "grad_norm": 1.7535753341635816, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6718528270721436, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.059499328233390915, | |
| "grad_norm": 1.9376370630167024, | |
| "learning_rate": 2e-05, | |
| "loss": 0.690641212463379, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.05977351868607935, | |
| "grad_norm": 2.5374679136605596, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6993424415588378, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.06004770913876779, | |
| "grad_norm": 8.884410141851701, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6886792182922363, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.06032189959145622, | |
| "grad_norm": 1.3716294166066423, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7170482635498047, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.060596090044144665, | |
| "grad_norm": 15.068560848475203, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6805070877075196, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.0608702804968331, | |
| "grad_norm": 1.3577298589073343, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7325131893157959, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.06114447094952154, | |
| "grad_norm": 1.5605906922192767, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7512709617614746, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.06141866140220997, | |
| "grad_norm": 2.7582484830984075, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6785373687744141, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.061692851854898416, | |
| "grad_norm": 1.5143827466792619, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6751925945281982, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.06196704230758685, | |
| "grad_norm": 2.0161013083730874, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7139543533325196, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.06224123276027529, | |
| "grad_norm": 1.5476890937684233, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6949531555175781, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.06251542321296373, | |
| "grad_norm": 2.854149288642599, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7019545555114746, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.06278961366565217, | |
| "grad_norm": 2.8910444598110145, | |
| "learning_rate": 2e-05, | |
| "loss": 0.679999828338623, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.0630638041183406, | |
| "grad_norm": 4.5131786865492165, | |
| "learning_rate": 2e-05, | |
| "loss": 0.743858814239502, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.06333799457102904, | |
| "grad_norm": 1.6666983651920946, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6903061866760254, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.06361218502371747, | |
| "grad_norm": 3.9236409182044927, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6770228385925293, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.06388637547640591, | |
| "grad_norm": 3.5304751421823375, | |
| "learning_rate": 2e-05, | |
| "loss": 0.683766508102417, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.06416056592909435, | |
| "grad_norm": 2.3440415121115508, | |
| "learning_rate": 2e-05, | |
| "loss": 0.685692024230957, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.06443475638178278, | |
| "grad_norm": 2.023495973731089, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6623273372650147, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.06470894683447122, | |
| "grad_norm": 1.3644101729258715, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6689095497131348, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.06498313728715967, | |
| "grad_norm": 1.9596672876784174, | |
| "learning_rate": 2e-05, | |
| "loss": 0.691974401473999, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.0652573277398481, | |
| "grad_norm": 1.3195698986606514, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6821478843688965, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.06553151819253654, | |
| "grad_norm": 1.7573022373302587, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6735329627990723, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.06580570864522498, | |
| "grad_norm": 1.53838950058506, | |
| "learning_rate": 2e-05, | |
| "loss": 0.748023509979248, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.06607989909791341, | |
| "grad_norm": 1.2532586297274881, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7004266738891601, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.06635408955060185, | |
| "grad_norm": 1.0700352542953793, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6959473609924316, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.06662828000329028, | |
| "grad_norm": 4.556276241049152, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8061809539794922, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.06690247045597872, | |
| "grad_norm": 2.152509533567621, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6806859016418457, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.06717666090866715, | |
| "grad_norm": 2.844321748025424, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6640607833862304, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.0674508513613556, | |
| "grad_norm": 1.1844711198885922, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6736173629760742, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.06772504181404404, | |
| "grad_norm": 1.7441815113714467, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6737575531005859, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.06799923226673248, | |
| "grad_norm": 15.110436266375306, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6922641277313233, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.06827342271942091, | |
| "grad_norm": 2.2896995422971513, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6922240257263184, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.06854761317210935, | |
| "grad_norm": 3.3946567144349524, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7507569313049316, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.06882180362479778, | |
| "grad_norm": 2.4027551835720606, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6868194580078125, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.06909599407748622, | |
| "grad_norm": 2.642622943517862, | |
| "learning_rate": 2e-05, | |
| "loss": 0.671518611907959, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.06937018453017466, | |
| "grad_norm": 1.5309404635845658, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7356051921844482, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.06964437498286309, | |
| "grad_norm": 3.0548677174511956, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6758328437805176, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.06991856543555153, | |
| "grad_norm": 1.193438740119827, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6692746639251709, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.07019275588823998, | |
| "grad_norm": 1.2975925737053815, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6996981620788574, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.07046694634092841, | |
| "grad_norm": 2.604623408290158, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6900700569152832, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.07074113679361685, | |
| "grad_norm": 1.3002551593133926, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6943139553070068, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.07101532724630528, | |
| "grad_norm": 1.2531714833158747, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6756250381469726, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.07128951769899372, | |
| "grad_norm": 1.2937652650463045, | |
| "learning_rate": 2e-05, | |
| "loss": 0.663780403137207, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.07156370815168216, | |
| "grad_norm": 2.6360485085938437, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7122066020965576, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.07183789860437059, | |
| "grad_norm": 1.4934934194790765, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7055806159973145, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.07211208905705903, | |
| "grad_norm": 1.265554029972572, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7235059261322021, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.07238627950974746, | |
| "grad_norm": 1.9958554767672445, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7027205467224121, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.07266046996243591, | |
| "grad_norm": 1.2413434845129339, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6815381526947022, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.07293466041512435, | |
| "grad_norm": 1.1371354821804263, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6793872356414795, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.07320885086781279, | |
| "grad_norm": 1.509310628738819, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6791254997253418, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.07348304132050122, | |
| "grad_norm": 1.1060317096479009, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6734872341156006, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.07375723177318966, | |
| "grad_norm": 2.300558523814945, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7129344940185547, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.0740314222258781, | |
| "grad_norm": 1.4427965933652498, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6588455200195312, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.07430561267856653, | |
| "grad_norm": 1.294782697054497, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6733306884765625, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.07457980313125497, | |
| "grad_norm": 1.5340099682623456, | |
| "learning_rate": 2e-05, | |
| "loss": 0.67325758934021, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.0748539935839434, | |
| "grad_norm": 1.206231978423083, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6762287139892578, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.07512818403663185, | |
| "grad_norm": 1.9615078302959699, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6494486808776856, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.07540237448932029, | |
| "grad_norm": 2.7069935077316156, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6673858642578125, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.07567656494200872, | |
| "grad_norm": 1.244130150553988, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6522541046142578, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.07595075539469716, | |
| "grad_norm": 1.2098821554953647, | |
| "learning_rate": 2e-05, | |
| "loss": 0.683398962020874, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 0.0762249458473856, | |
| "grad_norm": 1.288135083537603, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6805778503417969, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.07649913630007403, | |
| "grad_norm": 1.6304988366424138, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7026628017425537, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.07677332675276247, | |
| "grad_norm": 2.7936841529325167, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6661033630371094, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.0770475172054509, | |
| "grad_norm": 1.0298921513328017, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6871449470520019, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 0.07732170765813934, | |
| "grad_norm": 1.1609549565762298, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7018230438232422, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.07759589811082779, | |
| "grad_norm": 1.4937627603643764, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6921014785766602, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 0.07787008856351622, | |
| "grad_norm": 1.06750496448494, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6660943508148194, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.07814427901620466, | |
| "grad_norm": 1.3483148967347722, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6800715923309326, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.0784184694688931, | |
| "grad_norm": 0.9156475317981986, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6963984489440918, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.07869265992158153, | |
| "grad_norm": 2.541415851603761, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6630027294158936, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 0.07896685037426997, | |
| "grad_norm": 16.943114843325624, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6456388473510742, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.0792410408269584, | |
| "grad_norm": 1.5105418227638865, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6730733394622803, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 0.07951523127964684, | |
| "grad_norm": 1.0845416608133904, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7171577453613281, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.07978942173233528, | |
| "grad_norm": 1.1519480568569667, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6709290504455566, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 0.08006361218502371, | |
| "grad_norm": 1.2117290732120638, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6620571136474609, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.08033780263771216, | |
| "grad_norm": 1.2315110130143458, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6733248710632325, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 0.0806119930904006, | |
| "grad_norm": 2.440287172046704, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6643840789794921, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.08088618354308903, | |
| "grad_norm": 1.679547493244505, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6582764625549317, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.08116037399577747, | |
| "grad_norm": 1.4899450244629049, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6351748466491699, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.0814345644484659, | |
| "grad_norm": 8.727623946787714, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6684012413024902, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 0.08170875490115434, | |
| "grad_norm": 1.579015056993268, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7062510490417481, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.08198294535384278, | |
| "grad_norm": 1.1464807229444751, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7049516201019287, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 0.08225713580653121, | |
| "grad_norm": 5.869960507362621, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6782597541809082, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.08253132625921965, | |
| "grad_norm": 1.4182529764185894, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6684169769287109, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 0.0828055167119081, | |
| "grad_norm": 1.6893769027297443, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6985646247863769, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.08307970716459653, | |
| "grad_norm": 2.5342955627283548, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6566522121429443, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 0.08335389761728497, | |
| "grad_norm": 2.2039912383675504, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6650538444519043, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.0836280880699734, | |
| "grad_norm": 1.1365614280802196, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7018685340881348, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.08390227852266184, | |
| "grad_norm": 9.019421975743805, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6758551597595215, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.08417646897535028, | |
| "grad_norm": 3.6659007510462036, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6577141761779786, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 0.08445065942803871, | |
| "grad_norm": 1.5678832350052356, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6269432067871094, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.08472484988072715, | |
| "grad_norm": 1.727670242813246, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6586098670959473, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 0.08499904033341558, | |
| "grad_norm": 3.755002398572796, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7226817607879639, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.08527323078610403, | |
| "grad_norm": 2.1330764933548174, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7848432064056396, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 0.08554742123879247, | |
| "grad_norm": 7.344569339937305, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6657351970672607, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.0858216116914809, | |
| "grad_norm": 10.468008663230526, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7568781852722168, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 0.08609580214416934, | |
| "grad_norm": 2.1138000067535527, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6460443496704101, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.08636999259685778, | |
| "grad_norm": 4.547865280742984, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6882412910461426, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.08664418304954621, | |
| "grad_norm": 3.5154327639954515, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6506322860717774, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.08691837350223465, | |
| "grad_norm": 2.7275478121939303, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6696713447570801, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 0.08719256395492309, | |
| "grad_norm": 1.3787630646286215, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6836231231689454, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.08746675440761152, | |
| "grad_norm": 2.047844653500189, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6477073669433594, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 0.08774094486029996, | |
| "grad_norm": 1.3852125913708124, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6763352870941162, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.08801513531298841, | |
| "grad_norm": 2.3040797052670854, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6652051448822022, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 0.08828932576567684, | |
| "grad_norm": 14.697316208995012, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7217847347259522, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 0.08856351621836528, | |
| "grad_norm": 2.51269929770817, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6437290191650391, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 0.08883770667105371, | |
| "grad_norm": 1.0839928596474344, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6739638328552247, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.08911189712374215, | |
| "grad_norm": 1.4232969395596793, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6545791625976562, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.08938608757643059, | |
| "grad_norm": 2.7705768107632407, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6883786678314209, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 0.08966027802911902, | |
| "grad_norm": 1.6294188609939928, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6384977340698242, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 0.08993446848180746, | |
| "grad_norm": 1.594582926676938, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6840876579284668, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 0.0902086589344959, | |
| "grad_norm": 3.419979140068578, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6966186046600342, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 0.09048284938718434, | |
| "grad_norm": 1.1720984784354762, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7145553588867187, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.09075703983987278, | |
| "grad_norm": 2.5865997974005905, | |
| "learning_rate": 2e-05, | |
| "loss": 0.670125675201416, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 0.09103123029256122, | |
| "grad_norm": 1.7653476493006024, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6601348876953125, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 0.09130542074524965, | |
| "grad_norm": 1.078475603753133, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6758632659912109, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 0.09157961119793809, | |
| "grad_norm": 1.9813753893710975, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6324165344238282, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 0.09185380165062652, | |
| "grad_norm": 5.357267664807992, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7044012069702148, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.09212799210331496, | |
| "grad_norm": 1.9661122621370217, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6732365608215332, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.0924021825560034, | |
| "grad_norm": 1.463741721870027, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6740346431732178, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 0.09267637300869183, | |
| "grad_norm": 2.1100577721208684, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6793565273284912, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 0.09295056346138028, | |
| "grad_norm": 1.647800752209185, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6734954357147217, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 0.09322475391406872, | |
| "grad_norm": 5.2930839705776425, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6649512767791748, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.09349894436675715, | |
| "grad_norm": 1.2957879552995335, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6656761169433594, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 0.09377313481944559, | |
| "grad_norm": 2.2505114790067173, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6644682884216309, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 0.09404732527213402, | |
| "grad_norm": 3.242876354488533, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7099063873291016, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 0.09432151572482246, | |
| "grad_norm": 1.2326965640901657, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6550569057464599, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 0.0945957061775109, | |
| "grad_norm": 1.1460797675833456, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7175025463104248, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.09486989663019933, | |
| "grad_norm": 1.2339953246226074, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6755492687225342, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 0.09514408708288777, | |
| "grad_norm": 1.2239178623958433, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7038548469543457, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 0.09541827753557622, | |
| "grad_norm": 2.054453414781126, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7407256126403808, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 0.09569246798826465, | |
| "grad_norm": 0.962580766048047, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6635346412658691, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 0.09596665844095309, | |
| "grad_norm": 1.2912100361591508, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7218295574188233, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.09624084889364153, | |
| "grad_norm": 1.1758629619276566, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6495042324066163, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 0.09651503934632996, | |
| "grad_norm": 3.5366805734644338, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6897785186767578, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.0967892297990184, | |
| "grad_norm": 2.272234739463371, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6400076866149902, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 0.09706342025170683, | |
| "grad_norm": 1.0410457309022894, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6666034698486328, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 0.09733761070439527, | |
| "grad_norm": 1.47293937630036, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6687037467956543, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.0976118011570837, | |
| "grad_norm": 1.9748563321883545, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6717099666595459, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 0.09788599160977214, | |
| "grad_norm": 1.5933750546326055, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6562514305114746, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 0.09816018206246059, | |
| "grad_norm": 3.128593619633481, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7533950805664062, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 0.09843437251514903, | |
| "grad_norm": 2.891793023072984, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6750978469848633, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 0.09870856296783746, | |
| "grad_norm": 2.1897121019394965, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6912845134735107, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.0989827534205259, | |
| "grad_norm": 2.004346028083513, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6787797927856445, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 0.09925694387321433, | |
| "grad_norm": 2.915237046123075, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6627199172973632, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 0.09953113432590277, | |
| "grad_norm": 1.135046615758623, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6445761680603027, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 0.0998053247785912, | |
| "grad_norm": 1.3959275315420312, | |
| "learning_rate": 2e-05, | |
| "loss": 0.667326545715332, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 0.10007951523127964, | |
| "grad_norm": 0.9909291231965143, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6794458389282226, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.10035370568396808, | |
| "grad_norm": 1.8845470713145256, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6390461921691895, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 0.10062789613665653, | |
| "grad_norm": 1.5618410538502554, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6542781829833985, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 0.10090208658934496, | |
| "grad_norm": 1.1053530735550423, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6584223747253418, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 0.1011762770420334, | |
| "grad_norm": 1.8050013341368214, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6776100158691406, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 0.10145046749472184, | |
| "grad_norm": 1.3509209662153294, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7505715847015381, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.10172465794741027, | |
| "grad_norm": 1.3171737413682911, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6496476650238037, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 0.10199884840009871, | |
| "grad_norm": 2.037728719646489, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6601945400238037, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 0.10227303885278714, | |
| "grad_norm": 0.9419204630678888, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6456241607666016, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 0.10254722930547558, | |
| "grad_norm": 0.9978934523669455, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6376785278320313, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 0.10282141975816401, | |
| "grad_norm": 1.2463996045138312, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6607570648193359, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.10309561021085246, | |
| "grad_norm": 2.5633755688709, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6273582458496094, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 0.1033698006635409, | |
| "grad_norm": 1.3319182781528256, | |
| "learning_rate": 2e-05, | |
| "loss": 0.653649091720581, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 0.10364399111622934, | |
| "grad_norm": 1.5739184819076273, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6797492027282714, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 0.10391818156891777, | |
| "grad_norm": 1.0307904527745226, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6382014274597168, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 0.10419237202160621, | |
| "grad_norm": 1.367080564181282, | |
| "learning_rate": 2e-05, | |
| "loss": 0.656506872177124, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.10446656247429464, | |
| "grad_norm": 1.1306043565346662, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6709275722503663, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 0.10474075292698308, | |
| "grad_norm": 1.1484780247773891, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6394906520843506, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 0.10501494337967152, | |
| "grad_norm": 1.1364725089670198, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6385705947875977, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 0.10528913383235995, | |
| "grad_norm": 0.8720469361835742, | |
| "learning_rate": 2e-05, | |
| "loss": 0.648165512084961, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.10556332428504839, | |
| "grad_norm": 1.0753251518024716, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7535962581634521, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.10583751473773684, | |
| "grad_norm": 2.0842578780523646, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6594803810119629, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 0.10611170519042527, | |
| "grad_norm": 1.2009333278099745, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6642222881317139, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 0.10638589564311371, | |
| "grad_norm": 1.336651409297708, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6247560024261475, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 0.10666008609580215, | |
| "grad_norm": 1.2693616052452827, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6595481395721435, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 0.10693427654849058, | |
| "grad_norm": 1.0968413832844812, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6259689331054688, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.10720846700117902, | |
| "grad_norm": 1.4406885841286199, | |
| "learning_rate": 2e-05, | |
| "loss": 0.635016393661499, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 0.10748265745386745, | |
| "grad_norm": 1.1149430537042344, | |
| "learning_rate": 2e-05, | |
| "loss": 0.652572774887085, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 0.10775684790655589, | |
| "grad_norm": 1.4284277973722026, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6603116512298584, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 0.10803103835924432, | |
| "grad_norm": 2.191046426715608, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6569153785705566, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 0.10830522881193277, | |
| "grad_norm": 0.9755637485958417, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6817933559417725, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.10857941926462121, | |
| "grad_norm": 0.9207426742604556, | |
| "learning_rate": 2e-05, | |
| "loss": 0.632993507385254, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 0.10885360971730965, | |
| "grad_norm": 1.1075412445990842, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6479109764099121, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 0.10912780016999808, | |
| "grad_norm": 1.02734487686768, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6399119853973388, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 0.10940199062268652, | |
| "grad_norm": 1.0062247863415956, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6592807292938232, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 0.10967618107537495, | |
| "grad_norm": 1.7207229565430802, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6665008544921875, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.10995037152806339, | |
| "grad_norm": 8.645042736815913, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6616263866424561, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 0.11022456198075183, | |
| "grad_norm": 1.2238886957987656, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6552308559417724, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 0.11049875243344026, | |
| "grad_norm": 1.1553638761775333, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6816983699798584, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 0.11077294288612871, | |
| "grad_norm": 1.054398000948271, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6408511161804199, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 0.11104713333881715, | |
| "grad_norm": 0.9428030212455841, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6279231071472168, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.11132132379150558, | |
| "grad_norm": 1.388635722554096, | |
| "learning_rate": 2e-05, | |
| "loss": 0.6294130325317383, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 0.11159551424419402, | |
| "grad_norm": 1.7370583970603126, | |
| "learning_rate": 2e-05, | |
| "loss": 0.7167839050292969, | |
| "step": 4070 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 5935, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 2035, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1927226987642880.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |