|
{ |
|
"best_metric": 4.344208240509033, |
|
"best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/passive/lstm/3/checkpoints/checkpoint-228960", |
|
"epoch": 0.025000606015738065, |
|
"eval_steps": 10, |
|
"global_step": 228960, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.999998362119627e-05, |
|
"loss": 10.8209, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.999161405248948e-05, |
|
"loss": 7.5568, |
|
"step": 512 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.998322810497896e-05, |
|
"loss": 7.0563, |
|
"step": 1024 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.997484215746844e-05, |
|
"loss": 6.9949, |
|
"step": 1536 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.996645620995792e-05, |
|
"loss": 6.9457, |
|
"step": 2048 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.99580702624474e-05, |
|
"loss": 6.8667, |
|
"step": 2560 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.994968431493688e-05, |
|
"loss": 6.7087, |
|
"step": 3072 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.994129836742636e-05, |
|
"loss": 6.6099, |
|
"step": 3584 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.993291241991584e-05, |
|
"loss": 6.5194, |
|
"step": 4096 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.992452647240532e-05, |
|
"loss": 6.4426, |
|
"step": 4608 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.99161405248948e-05, |
|
"loss": 6.3859, |
|
"step": 5120 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.990775457738428e-05, |
|
"loss": 6.3178, |
|
"step": 5632 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.989936862987376e-05, |
|
"loss": 6.2549, |
|
"step": 6144 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.989098268236324e-05, |
|
"loss": 6.1917, |
|
"step": 6656 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.988259673485272e-05, |
|
"loss": 6.14, |
|
"step": 7168 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.98742107873422e-05, |
|
"loss": 6.0751, |
|
"step": 7680 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.986582483983168e-05, |
|
"loss": 6.0418, |
|
"step": 8192 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.985743889232116e-05, |
|
"loss": 5.9947, |
|
"step": 8704 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.984905294481064e-05, |
|
"loss": 5.9555, |
|
"step": 9216 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.984066699730012e-05, |
|
"loss": 5.917, |
|
"step": 9728 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.983229742859333e-05, |
|
"loss": 5.8862, |
|
"step": 10240 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.982391148108281e-05, |
|
"loss": 5.8472, |
|
"step": 10752 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.981552553357229e-05, |
|
"loss": 5.8198, |
|
"step": 11264 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.9807155964865506e-05, |
|
"loss": 5.778, |
|
"step": 11776 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.9798770017354986e-05, |
|
"loss": 5.7659, |
|
"step": 12288 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.9790384069844466e-05, |
|
"loss": 5.7291, |
|
"step": 12800 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.9781998122333946e-05, |
|
"loss": 5.7097, |
|
"step": 13312 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.9773612174823426e-05, |
|
"loss": 5.6721, |
|
"step": 13824 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.97652262273129e-05, |
|
"loss": 5.6552, |
|
"step": 14336 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.975684027980238e-05, |
|
"loss": 5.6355, |
|
"step": 14848 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.974845433229186e-05, |
|
"loss": 5.6104, |
|
"step": 15360 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.974008476358507e-05, |
|
"loss": 5.6008, |
|
"step": 15872 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.973169881607455e-05, |
|
"loss": 5.5742, |
|
"step": 16384 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.972331286856403e-05, |
|
"loss": 5.5579, |
|
"step": 16896 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9714926921053515e-05, |
|
"loss": 5.5499, |
|
"step": 17408 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9706540973542995e-05, |
|
"loss": 5.5264, |
|
"step": 17920 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9698171404836204e-05, |
|
"loss": 5.5173, |
|
"step": 18432 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9689785457325684e-05, |
|
"loss": 5.471, |
|
"step": 18944 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9681399509815164e-05, |
|
"loss": 5.477, |
|
"step": 19456 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9673013562304644e-05, |
|
"loss": 5.4379, |
|
"step": 19968 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9664627614794124e-05, |
|
"loss": 5.4437, |
|
"step": 20480 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.965625804608733e-05, |
|
"loss": 5.4203, |
|
"step": 20992 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.964787209857681e-05, |
|
"loss": 5.4179, |
|
"step": 21504 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.963948615106629e-05, |
|
"loss": 5.3885, |
|
"step": 22016 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.963110020355577e-05, |
|
"loss": 5.3944, |
|
"step": 22528 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.962271425604525e-05, |
|
"loss": 5.3831, |
|
"step": 23040 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.961432830853473e-05, |
|
"loss": 5.3749, |
|
"step": 23552 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.960595873982795e-05, |
|
"loss": 5.3678, |
|
"step": 24064 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.959757279231743e-05, |
|
"loss": 5.3383, |
|
"step": 24576 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.958918684480691e-05, |
|
"loss": 5.3293, |
|
"step": 25088 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.958080089729639e-05, |
|
"loss": 5.3294, |
|
"step": 25600 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.957241494978587e-05, |
|
"loss": 5.3153, |
|
"step": 26112 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.956402900227535e-05, |
|
"loss": 5.3102, |
|
"step": 26624 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.955564305476483e-05, |
|
"loss": 5.2874, |
|
"step": 27136 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.954727348605804e-05, |
|
"loss": 5.2838, |
|
"step": 27648 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.953888753854752e-05, |
|
"loss": 5.2638, |
|
"step": 28160 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9530501591037e-05, |
|
"loss": 5.2849, |
|
"step": 28672 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.952211564352648e-05, |
|
"loss": 5.2387, |
|
"step": 29184 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.951372969601596e-05, |
|
"loss": 5.2546, |
|
"step": 29696 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.950534374850543e-05, |
|
"loss": 5.2441, |
|
"step": 30208 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.949695780099492e-05, |
|
"loss": 5.2235, |
|
"step": 30720 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.948858823228813e-05, |
|
"loss": 5.2221, |
|
"step": 31232 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.948020228477761e-05, |
|
"loss": 5.2082, |
|
"step": 31744 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9471816337267086e-05, |
|
"loss": 5.1982, |
|
"step": 32256 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9463430389756566e-05, |
|
"loss": 5.191, |
|
"step": 32768 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.945507719985351e-05, |
|
"loss": 5.2049, |
|
"step": 33280 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.944669125234299e-05, |
|
"loss": 5.1792, |
|
"step": 33792 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.943830530483247e-05, |
|
"loss": 5.1701, |
|
"step": 34304 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.942991935732195e-05, |
|
"loss": 5.1601, |
|
"step": 34816 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.942153340981143e-05, |
|
"loss": 5.157, |
|
"step": 35328 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9413147462300904e-05, |
|
"loss": 5.1632, |
|
"step": 35840 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9404761514790384e-05, |
|
"loss": 5.1577, |
|
"step": 36352 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.939637556727987e-05, |
|
"loss": 5.1337, |
|
"step": 36864 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.938798961976935e-05, |
|
"loss": 5.1489, |
|
"step": 37376 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9379636429866296e-05, |
|
"loss": 5.1414, |
|
"step": 37888 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9371250482355776e-05, |
|
"loss": 5.1144, |
|
"step": 38400 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9362864534845256e-05, |
|
"loss": 5.1127, |
|
"step": 38912 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.935447858733473e-05, |
|
"loss": 5.103, |
|
"step": 39424 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.934609263982421e-05, |
|
"loss": 5.0947, |
|
"step": 39936 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.933770669231369e-05, |
|
"loss": 5.0891, |
|
"step": 40448 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.932932074480317e-05, |
|
"loss": 5.0944, |
|
"step": 40960 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.932093479729265e-05, |
|
"loss": 5.0902, |
|
"step": 41472 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.931254884978213e-05, |
|
"loss": 5.0915, |
|
"step": 41984 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.930417928107534e-05, |
|
"loss": 5.0665, |
|
"step": 42496 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9295793333564825e-05, |
|
"loss": 5.0488, |
|
"step": 43008 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9287423764858034e-05, |
|
"loss": 5.0552, |
|
"step": 43520 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9279037817347514e-05, |
|
"loss": 5.0559, |
|
"step": 44032 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9270651869836994e-05, |
|
"loss": 5.0544, |
|
"step": 44544 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9262265922326474e-05, |
|
"loss": 5.0371, |
|
"step": 45056 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9253879974815954e-05, |
|
"loss": 5.0355, |
|
"step": 45568 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9245494027305433e-05, |
|
"loss": 5.0355, |
|
"step": 46080 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9237108079794913e-05, |
|
"loss": 5.0281, |
|
"step": 46592 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.922872213228439e-05, |
|
"loss": 5.0165, |
|
"step": 47104 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.922033618477387e-05, |
|
"loss": 5.0128, |
|
"step": 47616 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.921195023726335e-05, |
|
"loss": 5.0001, |
|
"step": 48128 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.920356428975283e-05, |
|
"loss": 5.0129, |
|
"step": 48640 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.919519472104604e-05, |
|
"loss": 5.0006, |
|
"step": 49152 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.918680877353552e-05, |
|
"loss": 4.979, |
|
"step": 49664 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.917842282602501e-05, |
|
"loss": 4.9847, |
|
"step": 50176 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.917003687851449e-05, |
|
"loss": 4.982, |
|
"step": 50688 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.916165093100397e-05, |
|
"loss": 4.9889, |
|
"step": 51200 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.915328136229718e-05, |
|
"loss": 4.9709, |
|
"step": 51712 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.914489541478666e-05, |
|
"loss": 4.9677, |
|
"step": 52224 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.913650946727614e-05, |
|
"loss": 4.9518, |
|
"step": 52736 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.912812351976562e-05, |
|
"loss": 4.9576, |
|
"step": 53248 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.911973757225509e-05, |
|
"loss": 4.9371, |
|
"step": 53760 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.911136800354831e-05, |
|
"loss": 4.9483, |
|
"step": 54272 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.910298205603779e-05, |
|
"loss": 4.9349, |
|
"step": 54784 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.909459610852727e-05, |
|
"loss": 4.9384, |
|
"step": 55296 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.908621016101675e-05, |
|
"loss": 4.9274, |
|
"step": 55808 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.907782421350623e-05, |
|
"loss": 4.9304, |
|
"step": 56320 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.906943826599571e-05, |
|
"loss": 4.9166, |
|
"step": 56832 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.906105231848519e-05, |
|
"loss": 4.9207, |
|
"step": 57344 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.905266637097467e-05, |
|
"loss": 4.9128, |
|
"step": 57856 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9044296802267876e-05, |
|
"loss": 4.9061, |
|
"step": 58368 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9035910854757356e-05, |
|
"loss": 4.9177, |
|
"step": 58880 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9027524907246836e-05, |
|
"loss": 4.9003, |
|
"step": 59392 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9019155338540045e-05, |
|
"loss": 4.9002, |
|
"step": 59904 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9010769391029525e-05, |
|
"loss": 4.8912, |
|
"step": 60416 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.9002383443519005e-05, |
|
"loss": 4.8944, |
|
"step": 60928 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.8994013874812214e-05, |
|
"loss": 4.8892, |
|
"step": 61440 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.89856279273017e-05, |
|
"loss": 4.8825, |
|
"step": 61952 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.897724197979118e-05, |
|
"loss": 4.8876, |
|
"step": 62464 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.896885603228066e-05, |
|
"loss": 4.874, |
|
"step": 62976 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.896047008477014e-05, |
|
"loss": 4.8753, |
|
"step": 63488 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.895208413725962e-05, |
|
"loss": 4.8537, |
|
"step": 64000 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.89436981897491e-05, |
|
"loss": 4.8603, |
|
"step": 64512 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.893531224223858e-05, |
|
"loss": 4.8618, |
|
"step": 65024 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.892694267353179e-05, |
|
"loss": 4.8521, |
|
"step": 65536 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.891855672602127e-05, |
|
"loss": 4.8429, |
|
"step": 66048 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.891017077851075e-05, |
|
"loss": 4.8658, |
|
"step": 66560 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.890178483100023e-05, |
|
"loss": 4.8481, |
|
"step": 67072 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.889339888348971e-05, |
|
"loss": 4.8407, |
|
"step": 67584 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.888501293597919e-05, |
|
"loss": 4.8387, |
|
"step": 68096 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.88766433672724e-05, |
|
"loss": 4.8334, |
|
"step": 68608 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.886825741976188e-05, |
|
"loss": 4.8353, |
|
"step": 69120 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.8859871472251365e-05, |
|
"loss": 4.8324, |
|
"step": 69632 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.8851485524740845e-05, |
|
"loss": 4.8275, |
|
"step": 70144 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.8843099577230325e-05, |
|
"loss": 4.829, |
|
"step": 70656 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.8834713629719805e-05, |
|
"loss": 4.828, |
|
"step": 71168 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.882632768220928e-05, |
|
"loss": 4.8157, |
|
"step": 71680 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.881794173469876e-05, |
|
"loss": 4.8062, |
|
"step": 72192 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.8809572165991974e-05, |
|
"loss": 4.8178, |
|
"step": 72704 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.880120259728518e-05, |
|
"loss": 4.8037, |
|
"step": 73216 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.879281664977466e-05, |
|
"loss": 4.8074, |
|
"step": 73728 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.878443070226414e-05, |
|
"loss": 4.7987, |
|
"step": 74240 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.877604475475362e-05, |
|
"loss": 4.7962, |
|
"step": 74752 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.87676588072431e-05, |
|
"loss": 4.8014, |
|
"step": 75264 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.875927285973258e-05, |
|
"loss": 4.7887, |
|
"step": 75776 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.87509032910258e-05, |
|
"loss": 4.7862, |
|
"step": 76288 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"eval_loss": 4.75404691696167, |
|
"eval_runtime": 294.4935, |
|
"eval_samples_per_second": 1295.754, |
|
"eval_steps_per_second": 40.493, |
|
"step": 76320 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.874251734351528e-05, |
|
"loss": 4.7728, |
|
"step": 76800 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.873414777480849e-05, |
|
"loss": 4.7719, |
|
"step": 77312 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.872576182729797e-05, |
|
"loss": 4.7959, |
|
"step": 77824 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.871737587978745e-05, |
|
"loss": 4.7803, |
|
"step": 78336 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.870898993227693e-05, |
|
"loss": 4.7712, |
|
"step": 78848 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.87006039847664e-05, |
|
"loss": 4.7681, |
|
"step": 79360 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.869221803725588e-05, |
|
"loss": 4.7691, |
|
"step": 79872 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.868383208974536e-05, |
|
"loss": 4.7519, |
|
"step": 80384 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8675462521038576e-05, |
|
"loss": 4.7617, |
|
"step": 80896 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8667076573528056e-05, |
|
"loss": 4.7605, |
|
"step": 81408 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8658690626017536e-05, |
|
"loss": 4.7619, |
|
"step": 81920 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8650304678507016e-05, |
|
"loss": 4.7615, |
|
"step": 82432 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8641935109800225e-05, |
|
"loss": 4.7428, |
|
"step": 82944 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8633549162289705e-05, |
|
"loss": 4.7437, |
|
"step": 83456 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8625163214779185e-05, |
|
"loss": 4.7324, |
|
"step": 83968 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8616777267268665e-05, |
|
"loss": 4.7343, |
|
"step": 84480 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8608391319758145e-05, |
|
"loss": 4.7403, |
|
"step": 84992 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8600005372247625e-05, |
|
"loss": 4.7298, |
|
"step": 85504 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8591619424737105e-05, |
|
"loss": 4.7359, |
|
"step": 86016 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8583233477226585e-05, |
|
"loss": 4.7484, |
|
"step": 86528 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8574847529716065e-05, |
|
"loss": 4.7308, |
|
"step": 87040 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8566477961009274e-05, |
|
"loss": 4.7333, |
|
"step": 87552 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.8558092013498754e-05, |
|
"loss": 4.7154, |
|
"step": 88064 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.854970606598824e-05, |
|
"loss": 4.7366, |
|
"step": 88576 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.854132011847772e-05, |
|
"loss": 4.7145, |
|
"step": 89088 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.85329341709672e-05, |
|
"loss": 4.7163, |
|
"step": 89600 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.852454822345668e-05, |
|
"loss": 4.7035, |
|
"step": 90112 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.851617865474989e-05, |
|
"loss": 4.7129, |
|
"step": 90624 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 4.850779270723937e-05, |
|
"loss": 4.7032, |
|
"step": 91136 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.849940675972885e-05, |
|
"loss": 4.6993, |
|
"step": 91648 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.849102081221833e-05, |
|
"loss": 4.7045, |
|
"step": 92160 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.848263486470781e-05, |
|
"loss": 4.7088, |
|
"step": 92672 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.847426529600102e-05, |
|
"loss": 4.702, |
|
"step": 93184 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.84658793484905e-05, |
|
"loss": 4.7027, |
|
"step": 93696 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.845750977978371e-05, |
|
"loss": 4.6948, |
|
"step": 94208 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8449123832273194e-05, |
|
"loss": 4.702, |
|
"step": 94720 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8440737884762674e-05, |
|
"loss": 4.6742, |
|
"step": 95232 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8432351937252154e-05, |
|
"loss": 4.6881, |
|
"step": 95744 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8423965989741634e-05, |
|
"loss": 4.6658, |
|
"step": 96256 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8415580042231114e-05, |
|
"loss": 4.6852, |
|
"step": 96768 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.840719409472059e-05, |
|
"loss": 4.669, |
|
"step": 97280 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.839880814721007e-05, |
|
"loss": 4.6818, |
|
"step": 97792 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.839043857850328e-05, |
|
"loss": 4.6603, |
|
"step": 98304 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.838205263099276e-05, |
|
"loss": 4.6788, |
|
"step": 98816 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8373666683482236e-05, |
|
"loss": 4.6764, |
|
"step": 99328 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8365280735971716e-05, |
|
"loss": 4.6732, |
|
"step": 99840 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8356894788461196e-05, |
|
"loss": 4.6797, |
|
"step": 100352 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.834852521975441e-05, |
|
"loss": 4.6501, |
|
"step": 100864 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.834013927224389e-05, |
|
"loss": 4.6538, |
|
"step": 101376 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.833175332473337e-05, |
|
"loss": 4.6663, |
|
"step": 101888 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.832336737722285e-05, |
|
"loss": 4.6582, |
|
"step": 102400 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.831498142971233e-05, |
|
"loss": 4.6559, |
|
"step": 102912 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.830659548220181e-05, |
|
"loss": 4.6488, |
|
"step": 103424 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.829820953469129e-05, |
|
"loss": 4.6436, |
|
"step": 103936 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.828982358718077e-05, |
|
"loss": 4.6389, |
|
"step": 104448 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.828145401847398e-05, |
|
"loss": 4.6534, |
|
"step": 104960 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.827306807096346e-05, |
|
"loss": 4.6291, |
|
"step": 105472 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.826468212345294e-05, |
|
"loss": 4.641, |
|
"step": 105984 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.825629617594242e-05, |
|
"loss": 4.6469, |
|
"step": 106496 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.82479102284319e-05, |
|
"loss": 4.6295, |
|
"step": 107008 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.823952428092138e-05, |
|
"loss": 4.6298, |
|
"step": 107520 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.823113833341086e-05, |
|
"loss": 4.6277, |
|
"step": 108032 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.822275238590034e-05, |
|
"loss": 4.6148, |
|
"step": 108544 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8214382817193557e-05, |
|
"loss": 4.6224, |
|
"step": 109056 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8205996869683037e-05, |
|
"loss": 4.6378, |
|
"step": 109568 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8197610922172516e-05, |
|
"loss": 4.6195, |
|
"step": 110080 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8189224974661996e-05, |
|
"loss": 4.6139, |
|
"step": 110592 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8180855405955206e-05, |
|
"loss": 4.6128, |
|
"step": 111104 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8172469458444685e-05, |
|
"loss": 4.6137, |
|
"step": 111616 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8164083510934165e-05, |
|
"loss": 4.6279, |
|
"step": 112128 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8155697563423645e-05, |
|
"loss": 4.6195, |
|
"step": 112640 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8147327994716854e-05, |
|
"loss": 4.6104, |
|
"step": 113152 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8138942047206334e-05, |
|
"loss": 4.624, |
|
"step": 113664 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.8130556099695814e-05, |
|
"loss": 4.6227, |
|
"step": 114176 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.812218653098903e-05, |
|
"loss": 4.6056, |
|
"step": 114688 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.811380058347851e-05, |
|
"loss": 4.6045, |
|
"step": 115200 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.810541463596799e-05, |
|
"loss": 4.6032, |
|
"step": 115712 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.809702868845747e-05, |
|
"loss": 4.6, |
|
"step": 116224 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.808864274094695e-05, |
|
"loss": 4.5967, |
|
"step": 116736 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.808027317224016e-05, |
|
"loss": 4.6001, |
|
"step": 117248 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.807188722472964e-05, |
|
"loss": 4.6069, |
|
"step": 117760 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.806350127721912e-05, |
|
"loss": 4.6072, |
|
"step": 118272 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.80551153297086e-05, |
|
"loss": 4.595, |
|
"step": 118784 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.804672938219807e-05, |
|
"loss": 4.5728, |
|
"step": 119296 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.803834343468755e-05, |
|
"loss": 4.5889, |
|
"step": 119808 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.802995748717703e-05, |
|
"loss": 4.5938, |
|
"step": 120320 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.802157153966652e-05, |
|
"loss": 4.595, |
|
"step": 120832 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.801320197095973e-05, |
|
"loss": 4.587, |
|
"step": 121344 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 4.800481602344921e-05, |
|
"loss": 4.5837, |
|
"step": 121856 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7996446454742424e-05, |
|
"loss": 4.5903, |
|
"step": 122368 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.79880605072319e-05, |
|
"loss": 4.5806, |
|
"step": 122880 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.797967455972138e-05, |
|
"loss": 4.5765, |
|
"step": 123392 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.797128861221086e-05, |
|
"loss": 4.5836, |
|
"step": 123904 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.796290266470034e-05, |
|
"loss": 4.5677, |
|
"step": 124416 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7954533095993546e-05, |
|
"loss": 4.5869, |
|
"step": 124928 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7946147148483026e-05, |
|
"loss": 4.5784, |
|
"step": 125440 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7937761200972506e-05, |
|
"loss": 4.5583, |
|
"step": 125952 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7929375253461986e-05, |
|
"loss": 4.5695, |
|
"step": 126464 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.79210056847552e-05, |
|
"loss": 4.5689, |
|
"step": 126976 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.791261973724468e-05, |
|
"loss": 4.5807, |
|
"step": 127488 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.790423378973416e-05, |
|
"loss": 4.5667, |
|
"step": 128000 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.789584784222364e-05, |
|
"loss": 4.562, |
|
"step": 128512 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.788746189471312e-05, |
|
"loss": 4.5555, |
|
"step": 129024 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.787909232600633e-05, |
|
"loss": 4.5643, |
|
"step": 129536 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.787070637849581e-05, |
|
"loss": 4.5466, |
|
"step": 130048 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.786232043098529e-05, |
|
"loss": 4.555, |
|
"step": 130560 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.785393448347477e-05, |
|
"loss": 4.5523, |
|
"step": 131072 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.784554853596425e-05, |
|
"loss": 4.5537, |
|
"step": 131584 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.783716258845373e-05, |
|
"loss": 4.5444, |
|
"step": 132096 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.782877664094321e-05, |
|
"loss": 4.5542, |
|
"step": 132608 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.782039069343269e-05, |
|
"loss": 4.5456, |
|
"step": 133120 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.781200474592217e-05, |
|
"loss": 4.5466, |
|
"step": 133632 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7803635177215386e-05, |
|
"loss": 4.5481, |
|
"step": 134144 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7795249229704866e-05, |
|
"loss": 4.5391, |
|
"step": 134656 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7786863282194346e-05, |
|
"loss": 4.5621, |
|
"step": 135168 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7778493713487555e-05, |
|
"loss": 4.5475, |
|
"step": 135680 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7770107765977035e-05, |
|
"loss": 4.5449, |
|
"step": 136192 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7761721818466515e-05, |
|
"loss": 4.5359, |
|
"step": 136704 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7753335870955995e-05, |
|
"loss": 4.5425, |
|
"step": 137216 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7744966302249204e-05, |
|
"loss": 4.5402, |
|
"step": 137728 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7736580354738684e-05, |
|
"loss": 4.532, |
|
"step": 138240 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7728194407228164e-05, |
|
"loss": 4.5474, |
|
"step": 138752 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7719808459717644e-05, |
|
"loss": 4.537, |
|
"step": 139264 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.771143889101086e-05, |
|
"loss": 4.5293, |
|
"step": 139776 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.770305294350034e-05, |
|
"loss": 4.5245, |
|
"step": 140288 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.769466699598982e-05, |
|
"loss": 4.5262, |
|
"step": 140800 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.76862810484793e-05, |
|
"loss": 4.5262, |
|
"step": 141312 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.767789510096878e-05, |
|
"loss": 4.5286, |
|
"step": 141824 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.766950915345826e-05, |
|
"loss": 4.5142, |
|
"step": 142336 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.766112320594773e-05, |
|
"loss": 4.5427, |
|
"step": 142848 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.765273725843721e-05, |
|
"loss": 4.5329, |
|
"step": 143360 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.764435131092669e-05, |
|
"loss": 4.5214, |
|
"step": 143872 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.763599812102364e-05, |
|
"loss": 4.5208, |
|
"step": 144384 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.762761217351312e-05, |
|
"loss": 4.5145, |
|
"step": 144896 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.76192262260026e-05, |
|
"loss": 4.5281, |
|
"step": 145408 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.761084027849208e-05, |
|
"loss": 4.5241, |
|
"step": 145920 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.760245433098156e-05, |
|
"loss": 4.5192, |
|
"step": 146432 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.759406838347104e-05, |
|
"loss": 4.5236, |
|
"step": 146944 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.758568243596052e-05, |
|
"loss": 4.5215, |
|
"step": 147456 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.757731286725373e-05, |
|
"loss": 4.5189, |
|
"step": 147968 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7568926919743206e-05, |
|
"loss": 4.5086, |
|
"step": 148480 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7560540972232686e-05, |
|
"loss": 4.5205, |
|
"step": 148992 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7552155024722166e-05, |
|
"loss": 4.5103, |
|
"step": 149504 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7543769077211646e-05, |
|
"loss": 4.5139, |
|
"step": 150016 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7535383129701126e-05, |
|
"loss": 4.5072, |
|
"step": 150528 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7526997182190606e-05, |
|
"loss": 4.5097, |
|
"step": 151040 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7518611234680086e-05, |
|
"loss": 4.5106, |
|
"step": 151552 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.7510241665973295e-05, |
|
"loss": 4.4977, |
|
"step": 152064 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 4.750185571846278e-05, |
|
"loss": 4.5018, |
|
"step": 152576 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"eval_loss": 4.476864814758301, |
|
"eval_runtime": 295.321, |
|
"eval_samples_per_second": 1292.123, |
|
"eval_steps_per_second": 40.38, |
|
"step": 152640 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.749346977095226e-05, |
|
"loss": 4.4904, |
|
"step": 153088 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.748508382344174e-05, |
|
"loss": 4.4898, |
|
"step": 153600 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.747669787593122e-05, |
|
"loss": 4.5159, |
|
"step": 154112 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.74683119284207e-05, |
|
"loss": 4.5074, |
|
"step": 154624 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.745992598091018e-05, |
|
"loss": 4.4951, |
|
"step": 155136 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.745154003339966e-05, |
|
"loss": 4.4946, |
|
"step": 155648 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.744315408588914e-05, |
|
"loss": 4.4962, |
|
"step": 156160 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.743476813837862e-05, |
|
"loss": 4.4806, |
|
"step": 156672 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.7426382190868095e-05, |
|
"loss": 4.4997, |
|
"step": 157184 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.7417996243357575e-05, |
|
"loss": 4.4921, |
|
"step": 157696 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.7409610295847055e-05, |
|
"loss": 4.4957, |
|
"step": 158208 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.7401224348336535e-05, |
|
"loss": 4.4977, |
|
"step": 158720 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.739285477962975e-05, |
|
"loss": 4.4807, |
|
"step": 159232 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.738446883211923e-05, |
|
"loss": 4.4857, |
|
"step": 159744 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.737608288460871e-05, |
|
"loss": 4.479, |
|
"step": 160256 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.736769693709819e-05, |
|
"loss": 4.4784, |
|
"step": 160768 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.735931098958767e-05, |
|
"loss": 4.4827, |
|
"step": 161280 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.735092504207715e-05, |
|
"loss": 4.4749, |
|
"step": 161792 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.734253909456663e-05, |
|
"loss": 4.481, |
|
"step": 162304 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.733415314705611e-05, |
|
"loss": 4.5013, |
|
"step": 162816 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.732576719954559e-05, |
|
"loss": 4.4794, |
|
"step": 163328 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.731738125203507e-05, |
|
"loss": 4.4827, |
|
"step": 163840 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.730899530452455e-05, |
|
"loss": 4.4755, |
|
"step": 164352 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.730062573581776e-05, |
|
"loss": 4.4884, |
|
"step": 164864 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.729223978830724e-05, |
|
"loss": 4.4684, |
|
"step": 165376 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.728385384079672e-05, |
|
"loss": 4.4794, |
|
"step": 165888 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.72754678932862e-05, |
|
"loss": 4.4631, |
|
"step": 166400 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.726708194577568e-05, |
|
"loss": 4.4686, |
|
"step": 166912 |
|
}, |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 4.725869599826516e-05, |
|
"loss": 4.465, |
|
"step": 167424 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.725031005075464e-05, |
|
"loss": 4.4623, |
|
"step": 167936 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.724192410324412e-05, |
|
"loss": 4.4693, |
|
"step": 168448 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.7233570913341064e-05, |
|
"loss": 4.4752, |
|
"step": 168960 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.7225184965830544e-05, |
|
"loss": 4.4674, |
|
"step": 169472 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.7216799018320024e-05, |
|
"loss": 4.4732, |
|
"step": 169984 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.720842944961323e-05, |
|
"loss": 4.4586, |
|
"step": 170496 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.720004350210271e-05, |
|
"loss": 4.4756, |
|
"step": 171008 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.719165755459219e-05, |
|
"loss": 4.443, |
|
"step": 171520 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.718327160708167e-05, |
|
"loss": 4.4657, |
|
"step": 172032 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.717488565957115e-05, |
|
"loss": 4.4399, |
|
"step": 172544 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.716649971206063e-05, |
|
"loss": 4.4612, |
|
"step": 173056 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.715811376455011e-05, |
|
"loss": 4.4485, |
|
"step": 173568 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.714972781703959e-05, |
|
"loss": 4.4624, |
|
"step": 174080 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.714134186952907e-05, |
|
"loss": 4.4441, |
|
"step": 174592 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.713295592201855e-05, |
|
"loss": 4.456, |
|
"step": 175104 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.712456997450803e-05, |
|
"loss": 4.4587, |
|
"step": 175616 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.711618402699751e-05, |
|
"loss": 4.4576, |
|
"step": 176128 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.710781445829072e-05, |
|
"loss": 4.4598, |
|
"step": 176640 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.70994285107802e-05, |
|
"loss": 4.4386, |
|
"step": 177152 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.709104256326968e-05, |
|
"loss": 4.4354, |
|
"step": 177664 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.708265661575916e-05, |
|
"loss": 4.4563, |
|
"step": 178176 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.707428704705237e-05, |
|
"loss": 4.4488, |
|
"step": 178688 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.706590109954185e-05, |
|
"loss": 4.4428, |
|
"step": 179200 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.705751515203134e-05, |
|
"loss": 4.441, |
|
"step": 179712 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.7049145583324546e-05, |
|
"loss": 4.4372, |
|
"step": 180224 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.7040759635814026e-05, |
|
"loss": 4.4278, |
|
"step": 180736 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.7032373688303506e-05, |
|
"loss": 4.443, |
|
"step": 181248 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.7023987740792986e-05, |
|
"loss": 4.4233, |
|
"step": 181760 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.7015601793282466e-05, |
|
"loss": 4.4356, |
|
"step": 182272 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.7007215845771946e-05, |
|
"loss": 4.4463, |
|
"step": 182784 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.6998829898261426e-05, |
|
"loss": 4.4244, |
|
"step": 183296 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.6990443950750906e-05, |
|
"loss": 4.4282, |
|
"step": 183808 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.6982074382044115e-05, |
|
"loss": 4.4301, |
|
"step": 184320 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.6973688434533595e-05, |
|
"loss": 4.4157, |
|
"step": 184832 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.6965302487023075e-05, |
|
"loss": 4.4185, |
|
"step": 185344 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.6956916539512555e-05, |
|
"loss": 4.4405, |
|
"step": 185856 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.6948530592002035e-05, |
|
"loss": 4.4234, |
|
"step": 186368 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.694016102329525e-05, |
|
"loss": 4.417, |
|
"step": 186880 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.693179145458846e-05, |
|
"loss": 4.4173, |
|
"step": 187392 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.692340550707794e-05, |
|
"loss": 4.4221, |
|
"step": 187904 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.691501955956742e-05, |
|
"loss": 4.4354, |
|
"step": 188416 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.69066336120569e-05, |
|
"loss": 4.4297, |
|
"step": 188928 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.689824766454638e-05, |
|
"loss": 4.4174, |
|
"step": 189440 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.688986171703586e-05, |
|
"loss": 4.4331, |
|
"step": 189952 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.688147576952534e-05, |
|
"loss": 4.4278, |
|
"step": 190464 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.687308982201481e-05, |
|
"loss": 4.4187, |
|
"step": 190976 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.686472025330803e-05, |
|
"loss": 4.415, |
|
"step": 191488 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.685633430579751e-05, |
|
"loss": 4.4204, |
|
"step": 192000 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.6847964737090725e-05, |
|
"loss": 4.4136, |
|
"step": 192512 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.6839578789580204e-05, |
|
"loss": 4.4126, |
|
"step": 193024 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.6831192842069684e-05, |
|
"loss": 4.4129, |
|
"step": 193536 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.6822806894559164e-05, |
|
"loss": 4.4213, |
|
"step": 194048 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.6814420947048644e-05, |
|
"loss": 4.4263, |
|
"step": 194560 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.680603499953812e-05, |
|
"loss": 4.4119, |
|
"step": 195072 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.6797665430831333e-05, |
|
"loss": 4.3973, |
|
"step": 195584 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.678929586212454e-05, |
|
"loss": 4.4047, |
|
"step": 196096 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.678090991461402e-05, |
|
"loss": 4.4111, |
|
"step": 196608 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.67725239671035e-05, |
|
"loss": 4.4173, |
|
"step": 197120 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.676413801959298e-05, |
|
"loss": 4.4071, |
|
"step": 197632 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.675575207208246e-05, |
|
"loss": 4.4053, |
|
"step": 198144 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.674736612457194e-05, |
|
"loss": 4.4154, |
|
"step": 198656 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.673898017706142e-05, |
|
"loss": 4.4014, |
|
"step": 199168 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.673061060835464e-05, |
|
"loss": 4.4028, |
|
"step": 199680 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.672222466084412e-05, |
|
"loss": 4.4053, |
|
"step": 200192 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.671383871333359e-05, |
|
"loss": 4.3974, |
|
"step": 200704 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.670545276582307e-05, |
|
"loss": 4.4155, |
|
"step": 201216 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.669706681831255e-05, |
|
"loss": 4.4004, |
|
"step": 201728 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.668868087080203e-05, |
|
"loss": 4.3895, |
|
"step": 202240 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.668029492329151e-05, |
|
"loss": 4.3991, |
|
"step": 202752 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.667190897578099e-05, |
|
"loss": 4.3966, |
|
"step": 203264 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.666352302827047e-05, |
|
"loss": 4.412, |
|
"step": 203776 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.665513708075995e-05, |
|
"loss": 4.4012, |
|
"step": 204288 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.664675113324943e-05, |
|
"loss": 4.3922, |
|
"step": 204800 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.663838156454265e-05, |
|
"loss": 4.3852, |
|
"step": 205312 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.662999561703213e-05, |
|
"loss": 4.3967, |
|
"step": 205824 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.662160966952161e-05, |
|
"loss": 4.3808, |
|
"step": 206336 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.661322372201109e-05, |
|
"loss": 4.3932, |
|
"step": 206848 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6604837774500567e-05, |
|
"loss": 4.3884, |
|
"step": 207360 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6596468205793776e-05, |
|
"loss": 4.3886, |
|
"step": 207872 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6588082258283256e-05, |
|
"loss": 4.3789, |
|
"step": 208384 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6579696310772736e-05, |
|
"loss": 4.398, |
|
"step": 208896 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6571326742065945e-05, |
|
"loss": 4.3804, |
|
"step": 209408 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6562940794555425e-05, |
|
"loss": 4.3823, |
|
"step": 209920 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6554554847044905e-05, |
|
"loss": 4.3901, |
|
"step": 210432 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6546168899534385e-05, |
|
"loss": 4.3801, |
|
"step": 210944 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6537782952023865e-05, |
|
"loss": 4.3996, |
|
"step": 211456 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6529397004513344e-05, |
|
"loss": 4.3879, |
|
"step": 211968 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.652101105700283e-05, |
|
"loss": 4.3885, |
|
"step": 212480 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6512625109492304e-05, |
|
"loss": 4.3786, |
|
"step": 212992 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6504239161981784e-05, |
|
"loss": 4.3841, |
|
"step": 213504 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6495869593275e-05, |
|
"loss": 4.3851, |
|
"step": 214016 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.648748364576447e-05, |
|
"loss": 4.3739, |
|
"step": 214528 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.647909769825395e-05, |
|
"loss": 4.3891, |
|
"step": 215040 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.647071175074343e-05, |
|
"loss": 4.3822, |
|
"step": 215552 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.646232580323291e-05, |
|
"loss": 4.3759, |
|
"step": 216064 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.645393985572239e-05, |
|
"loss": 4.378, |
|
"step": 216576 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.644555390821187e-05, |
|
"loss": 4.3759, |
|
"step": 217088 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.643716796070135e-05, |
|
"loss": 4.3674, |
|
"step": 217600 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.642879839199457e-05, |
|
"loss": 4.3816, |
|
"step": 218112 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.642041244448405e-05, |
|
"loss": 4.3651, |
|
"step": 218624 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.641202649697353e-05, |
|
"loss": 4.3877, |
|
"step": 219136 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.640364054946301e-05, |
|
"loss": 4.3838, |
|
"step": 219648 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.639527098075622e-05, |
|
"loss": 4.3703, |
|
"step": 220160 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.63868850332457e-05, |
|
"loss": 4.3692, |
|
"step": 220672 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.637851546453891e-05, |
|
"loss": 4.3682, |
|
"step": 221184 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.637012951702839e-05, |
|
"loss": 4.3788, |
|
"step": 221696 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.636174356951787e-05, |
|
"loss": 4.3756, |
|
"step": 222208 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6353374000811076e-05, |
|
"loss": 4.3718, |
|
"step": 222720 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6344988053300556e-05, |
|
"loss": 4.3748, |
|
"step": 223232 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.6336602105790036e-05, |
|
"loss": 4.3754, |
|
"step": 223744 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.632823253708325e-05, |
|
"loss": 4.3754, |
|
"step": 224256 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.631984658957273e-05, |
|
"loss": 4.3619, |
|
"step": 224768 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.631146064206221e-05, |
|
"loss": 4.3801, |
|
"step": 225280 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.630307469455169e-05, |
|
"loss": 4.3694, |
|
"step": 225792 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.629468874704117e-05, |
|
"loss": 4.3655, |
|
"step": 226304 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.628630279953065e-05, |
|
"loss": 4.367, |
|
"step": 226816 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.627791685202013e-05, |
|
"loss": 4.3685, |
|
"step": 227328 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.626953090450961e-05, |
|
"loss": 4.3661, |
|
"step": 227840 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.626114495699909e-05, |
|
"loss": 4.3543, |
|
"step": 228352 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 4.625275900948857e-05, |
|
"loss": 4.3581, |
|
"step": 228864 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"eval_loss": 4.344208240509033, |
|
"eval_runtime": 292.4539, |
|
"eval_samples_per_second": 1304.79, |
|
"eval_steps_per_second": 40.776, |
|
"step": 228960 |
|
} |
|
], |
|
"logging_steps": 512, |
|
"max_steps": 3052726, |
|
"num_train_epochs": 9223372036854775807, |
|
"save_steps": 10, |
|
"total_flos": 9.460672397103816e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|