BERT-SBIC-offensive / trainer_state.json
Cameron Clarke
add model
0e2363d
{
"best_metric": 0.8071770624962116,
"best_model_checkpoint": "/scratch/ccc779/sbic_offensiveness_model/checkpoint-55442",
"epoch": 2.0,
"global_step": 110884,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 1.2126612314737829e-05,
"loss": 0.917,
"step": 500
},
{
"epoch": 0.02,
"learning_rate": 1.2071683102556998e-05,
"loss": 0.9735,
"step": 1000
},
{
"epoch": 0.03,
"learning_rate": 1.2016753890376165e-05,
"loss": 1.045,
"step": 1500
},
{
"epoch": 0.04,
"learning_rate": 1.1961824678195334e-05,
"loss": 1.0762,
"step": 2000
},
{
"epoch": 0.05,
"learning_rate": 1.1906895466014503e-05,
"loss": 1.0967,
"step": 2500
},
{
"epoch": 0.05,
"learning_rate": 1.1851966253833671e-05,
"loss": 1.0201,
"step": 3000
},
{
"epoch": 0.06,
"learning_rate": 1.1797037041652839e-05,
"loss": 1.0907,
"step": 3500
},
{
"epoch": 0.07,
"learning_rate": 1.1742107829472006e-05,
"loss": 1.0534,
"step": 4000
},
{
"epoch": 0.08,
"learning_rate": 1.1687178617291175e-05,
"loss": 1.0246,
"step": 4500
},
{
"epoch": 0.09,
"learning_rate": 1.1632249405110344e-05,
"loss": 1.0328,
"step": 5000
},
{
"epoch": 0.1,
"learning_rate": 1.1577320192929512e-05,
"loss": 1.0292,
"step": 5500
},
{
"epoch": 0.11,
"learning_rate": 1.1522390980748681e-05,
"loss": 1.0483,
"step": 6000
},
{
"epoch": 0.12,
"learning_rate": 1.146746176856785e-05,
"loss": 1.0784,
"step": 6500
},
{
"epoch": 0.13,
"learning_rate": 1.1412532556387017e-05,
"loss": 0.9617,
"step": 7000
},
{
"epoch": 0.14,
"learning_rate": 1.1357603344206186e-05,
"loss": 0.9761,
"step": 7500
},
{
"epoch": 0.14,
"learning_rate": 1.1302674132025355e-05,
"loss": 0.9707,
"step": 8000
},
{
"epoch": 0.15,
"learning_rate": 1.1247744919844524e-05,
"loss": 1.0156,
"step": 8500
},
{
"epoch": 0.16,
"learning_rate": 1.1192815707663693e-05,
"loss": 1.0076,
"step": 9000
},
{
"epoch": 0.17,
"learning_rate": 1.113788649548286e-05,
"loss": 1.0648,
"step": 9500
},
{
"epoch": 0.18,
"learning_rate": 1.1082957283302027e-05,
"loss": 0.9915,
"step": 10000
},
{
"epoch": 0.19,
"learning_rate": 1.1028028071121196e-05,
"loss": 0.9352,
"step": 10500
},
{
"epoch": 0.2,
"learning_rate": 1.0973098858940365e-05,
"loss": 0.9666,
"step": 11000
},
{
"epoch": 0.21,
"learning_rate": 1.0918169646759534e-05,
"loss": 0.9732,
"step": 11500
},
{
"epoch": 0.22,
"learning_rate": 1.0863240434578701e-05,
"loss": 0.9314,
"step": 12000
},
{
"epoch": 0.23,
"learning_rate": 1.080831122239787e-05,
"loss": 0.9898,
"step": 12500
},
{
"epoch": 0.23,
"learning_rate": 1.0753382010217039e-05,
"loss": 0.9133,
"step": 13000
},
{
"epoch": 0.24,
"learning_rate": 1.0698452798036208e-05,
"loss": 0.9795,
"step": 13500
},
{
"epoch": 0.25,
"learning_rate": 1.0643523585855376e-05,
"loss": 0.9306,
"step": 14000
},
{
"epoch": 0.26,
"learning_rate": 1.0588594373674544e-05,
"loss": 1.0033,
"step": 14500
},
{
"epoch": 0.27,
"learning_rate": 1.0533665161493712e-05,
"loss": 0.958,
"step": 15000
},
{
"epoch": 0.28,
"learning_rate": 1.0478735949312881e-05,
"loss": 0.9671,
"step": 15500
},
{
"epoch": 0.29,
"learning_rate": 1.0423806737132048e-05,
"loss": 0.9541,
"step": 16000
},
{
"epoch": 0.3,
"learning_rate": 1.0368877524951217e-05,
"loss": 0.984,
"step": 16500
},
{
"epoch": 0.31,
"learning_rate": 1.0313948312770386e-05,
"loss": 0.9726,
"step": 17000
},
{
"epoch": 0.32,
"learning_rate": 1.0259019100589553e-05,
"loss": 0.9566,
"step": 17500
},
{
"epoch": 0.32,
"learning_rate": 1.0204089888408722e-05,
"loss": 0.8828,
"step": 18000
},
{
"epoch": 0.33,
"learning_rate": 1.0149160676227891e-05,
"loss": 0.9034,
"step": 18500
},
{
"epoch": 0.34,
"learning_rate": 1.009423146404706e-05,
"loss": 0.9196,
"step": 19000
},
{
"epoch": 0.35,
"learning_rate": 1.0039302251866229e-05,
"loss": 0.9549,
"step": 19500
},
{
"epoch": 0.36,
"learning_rate": 9.984373039685396e-06,
"loss": 0.9173,
"step": 20000
},
{
"epoch": 0.37,
"learning_rate": 9.929443827504565e-06,
"loss": 0.904,
"step": 20500
},
{
"epoch": 0.38,
"learning_rate": 9.874514615323734e-06,
"loss": 0.9472,
"step": 21000
},
{
"epoch": 0.39,
"learning_rate": 9.819585403142903e-06,
"loss": 0.9822,
"step": 21500
},
{
"epoch": 0.4,
"learning_rate": 9.76465619096207e-06,
"loss": 0.9483,
"step": 22000
},
{
"epoch": 0.41,
"learning_rate": 9.709726978781239e-06,
"loss": 0.9373,
"step": 22500
},
{
"epoch": 0.41,
"learning_rate": 9.654797766600406e-06,
"loss": 0.9935,
"step": 23000
},
{
"epoch": 0.42,
"learning_rate": 9.599868554419575e-06,
"loss": 0.9639,
"step": 23500
},
{
"epoch": 0.43,
"learning_rate": 9.544939342238744e-06,
"loss": 0.8847,
"step": 24000
},
{
"epoch": 0.44,
"learning_rate": 9.490010130057912e-06,
"loss": 0.8719,
"step": 24500
},
{
"epoch": 0.45,
"learning_rate": 9.43508091787708e-06,
"loss": 0.8415,
"step": 25000
},
{
"epoch": 0.46,
"learning_rate": 9.380151705696248e-06,
"loss": 1.0052,
"step": 25500
},
{
"epoch": 0.47,
"learning_rate": 9.325222493515417e-06,
"loss": 0.9453,
"step": 26000
},
{
"epoch": 0.48,
"learning_rate": 9.270293281334586e-06,
"loss": 0.8958,
"step": 26500
},
{
"epoch": 0.49,
"learning_rate": 9.215364069153755e-06,
"loss": 0.9676,
"step": 27000
},
{
"epoch": 0.5,
"learning_rate": 9.160434856972924e-06,
"loss": 0.961,
"step": 27500
},
{
"epoch": 0.51,
"learning_rate": 9.105505644792091e-06,
"loss": 0.9183,
"step": 28000
},
{
"epoch": 0.51,
"learning_rate": 9.05057643261126e-06,
"loss": 0.8921,
"step": 28500
},
{
"epoch": 0.52,
"learning_rate": 8.995647220430427e-06,
"loss": 0.8327,
"step": 29000
},
{
"epoch": 0.53,
"learning_rate": 8.940718008249596e-06,
"loss": 0.9493,
"step": 29500
},
{
"epoch": 0.54,
"learning_rate": 8.885788796068765e-06,
"loss": 0.8975,
"step": 30000
},
{
"epoch": 0.55,
"learning_rate": 8.830859583887932e-06,
"loss": 0.9427,
"step": 30500
},
{
"epoch": 0.56,
"learning_rate": 8.7759303717071e-06,
"loss": 0.91,
"step": 31000
},
{
"epoch": 0.57,
"learning_rate": 8.72100115952627e-06,
"loss": 0.9484,
"step": 31500
},
{
"epoch": 0.58,
"learning_rate": 8.666071947345439e-06,
"loss": 0.8844,
"step": 32000
},
{
"epoch": 0.59,
"learning_rate": 8.611142735164607e-06,
"loss": 0.9076,
"step": 32500
},
{
"epoch": 0.6,
"learning_rate": 8.556213522983775e-06,
"loss": 0.8977,
"step": 33000
},
{
"epoch": 0.6,
"learning_rate": 8.501284310802943e-06,
"loss": 0.8833,
"step": 33500
},
{
"epoch": 0.61,
"learning_rate": 8.446355098622112e-06,
"loss": 0.8685,
"step": 34000
},
{
"epoch": 0.62,
"learning_rate": 8.391425886441281e-06,
"loss": 0.8938,
"step": 34500
},
{
"epoch": 0.63,
"learning_rate": 8.336496674260448e-06,
"loss": 0.862,
"step": 35000
},
{
"epoch": 0.64,
"learning_rate": 8.281567462079617e-06,
"loss": 0.9537,
"step": 35500
},
{
"epoch": 0.65,
"learning_rate": 8.226638249898784e-06,
"loss": 0.8841,
"step": 36000
},
{
"epoch": 0.66,
"learning_rate": 8.171709037717953e-06,
"loss": 0.9267,
"step": 36500
},
{
"epoch": 0.67,
"learning_rate": 8.116779825537122e-06,
"loss": 0.929,
"step": 37000
},
{
"epoch": 0.68,
"learning_rate": 8.061850613356291e-06,
"loss": 0.9398,
"step": 37500
},
{
"epoch": 0.69,
"learning_rate": 8.00692140117546e-06,
"loss": 0.9289,
"step": 38000
},
{
"epoch": 0.69,
"learning_rate": 7.951992188994627e-06,
"loss": 0.9559,
"step": 38500
},
{
"epoch": 0.7,
"learning_rate": 7.897062976813796e-06,
"loss": 0.9307,
"step": 39000
},
{
"epoch": 0.71,
"learning_rate": 7.842133764632965e-06,
"loss": 0.8903,
"step": 39500
},
{
"epoch": 0.72,
"learning_rate": 7.787204552452134e-06,
"loss": 0.917,
"step": 40000
},
{
"epoch": 0.73,
"learning_rate": 7.732275340271303e-06,
"loss": 0.8122,
"step": 40500
},
{
"epoch": 0.74,
"learning_rate": 7.67734612809047e-06,
"loss": 0.8607,
"step": 41000
},
{
"epoch": 0.75,
"learning_rate": 7.622416915909638e-06,
"loss": 0.7791,
"step": 41500
},
{
"epoch": 0.76,
"learning_rate": 7.5674877037288066e-06,
"loss": 0.9309,
"step": 42000
},
{
"epoch": 0.77,
"learning_rate": 7.5125584915479754e-06,
"loss": 0.9187,
"step": 42500
},
{
"epoch": 0.78,
"learning_rate": 7.4576292793671434e-06,
"loss": 0.9128,
"step": 43000
},
{
"epoch": 0.78,
"learning_rate": 7.4027000671863115e-06,
"loss": 0.887,
"step": 43500
},
{
"epoch": 0.79,
"learning_rate": 7.3477708550054795e-06,
"loss": 0.8972,
"step": 44000
},
{
"epoch": 0.8,
"learning_rate": 7.292841642824648e-06,
"loss": 0.8416,
"step": 44500
},
{
"epoch": 0.81,
"learning_rate": 7.237912430643817e-06,
"loss": 0.8631,
"step": 45000
},
{
"epoch": 0.82,
"learning_rate": 7.182983218462986e-06,
"loss": 0.8923,
"step": 45500
},
{
"epoch": 0.83,
"learning_rate": 7.128054006282153e-06,
"loss": 0.8394,
"step": 46000
},
{
"epoch": 0.84,
"learning_rate": 7.073124794101322e-06,
"loss": 0.91,
"step": 46500
},
{
"epoch": 0.85,
"learning_rate": 7.01819558192049e-06,
"loss": 0.8782,
"step": 47000
},
{
"epoch": 0.86,
"learning_rate": 6.963266369739659e-06,
"loss": 0.8415,
"step": 47500
},
{
"epoch": 0.87,
"learning_rate": 6.908337157558828e-06,
"loss": 0.8202,
"step": 48000
},
{
"epoch": 0.87,
"learning_rate": 6.853407945377997e-06,
"loss": 0.8648,
"step": 48500
},
{
"epoch": 0.88,
"learning_rate": 6.798478733197164e-06,
"loss": 0.8353,
"step": 49000
},
{
"epoch": 0.89,
"learning_rate": 6.743549521016333e-06,
"loss": 0.8785,
"step": 49500
},
{
"epoch": 0.9,
"learning_rate": 6.688620308835501e-06,
"loss": 0.8593,
"step": 50000
},
{
"epoch": 0.91,
"learning_rate": 6.63369109665467e-06,
"loss": 0.866,
"step": 50500
},
{
"epoch": 0.92,
"learning_rate": 6.5787618844738385e-06,
"loss": 0.8312,
"step": 51000
},
{
"epoch": 0.93,
"learning_rate": 6.523832672293006e-06,
"loss": 0.8588,
"step": 51500
},
{
"epoch": 0.94,
"learning_rate": 6.4689034601121746e-06,
"loss": 0.9104,
"step": 52000
},
{
"epoch": 0.95,
"learning_rate": 6.413974247931343e-06,
"loss": 0.878,
"step": 52500
},
{
"epoch": 0.96,
"learning_rate": 6.3590450357505114e-06,
"loss": 0.9116,
"step": 53000
},
{
"epoch": 0.96,
"learning_rate": 6.30411582356968e-06,
"loss": 0.8848,
"step": 53500
},
{
"epoch": 0.97,
"learning_rate": 6.2491866113888475e-06,
"loss": 0.8784,
"step": 54000
},
{
"epoch": 0.98,
"learning_rate": 6.194257399208016e-06,
"loss": 0.8082,
"step": 54500
},
{
"epoch": 0.99,
"learning_rate": 6.139328187027185e-06,
"loss": 0.9775,
"step": 55000
},
{
"epoch": 1.0,
"eval_accuracy": 0.8071770624962115,
"eval_f1": 0.8071770624962116,
"eval_loss": 0.8160773515701294,
"eval_precision": 0.8071770624962115,
"eval_recall": 0.8071770624962115,
"eval_runtime": 106.0372,
"eval_samples_per_second": 155.577,
"step": 55442
},
{
"epoch": 1.0,
"learning_rate": 6.084398974846354e-06,
"loss": 0.8451,
"step": 55500
},
{
"epoch": 1.01,
"learning_rate": 6.029469762665522e-06,
"loss": 0.8463,
"step": 56000
},
{
"epoch": 1.02,
"learning_rate": 5.97454055048469e-06,
"loss": 0.8269,
"step": 56500
},
{
"epoch": 1.03,
"learning_rate": 5.919611338303859e-06,
"loss": 0.7929,
"step": 57000
},
{
"epoch": 1.04,
"learning_rate": 5.864682126123027e-06,
"loss": 0.7749,
"step": 57500
},
{
"epoch": 1.05,
"learning_rate": 5.809752913942196e-06,
"loss": 0.7263,
"step": 58000
},
{
"epoch": 1.06,
"learning_rate": 5.754823701761364e-06,
"loss": 0.783,
"step": 58500
},
{
"epoch": 1.06,
"learning_rate": 5.699894489580533e-06,
"loss": 0.7733,
"step": 59000
},
{
"epoch": 1.07,
"learning_rate": 5.644965277399701e-06,
"loss": 0.7848,
"step": 59500
},
{
"epoch": 1.08,
"learning_rate": 5.590036065218869e-06,
"loss": 0.7388,
"step": 60000
},
{
"epoch": 1.09,
"learning_rate": 5.535106853038038e-06,
"loss": 0.7464,
"step": 60500
},
{
"epoch": 1.1,
"learning_rate": 5.4801776408572065e-06,
"loss": 0.766,
"step": 61000
},
{
"epoch": 1.11,
"learning_rate": 5.4252484286763745e-06,
"loss": 0.7326,
"step": 61500
},
{
"epoch": 1.12,
"learning_rate": 5.370319216495543e-06,
"loss": 0.7069,
"step": 62000
},
{
"epoch": 1.13,
"learning_rate": 5.315390004314711e-06,
"loss": 0.7877,
"step": 62500
},
{
"epoch": 1.14,
"learning_rate": 5.2604607921338794e-06,
"loss": 0.7548,
"step": 63000
},
{
"epoch": 1.15,
"learning_rate": 5.205531579953048e-06,
"loss": 0.7907,
"step": 63500
},
{
"epoch": 1.15,
"learning_rate": 5.150602367772216e-06,
"loss": 0.7388,
"step": 64000
},
{
"epoch": 1.16,
"learning_rate": 5.095673155591385e-06,
"loss": 0.8162,
"step": 64500
},
{
"epoch": 1.17,
"learning_rate": 5.040743943410553e-06,
"loss": 0.7412,
"step": 65000
},
{
"epoch": 1.18,
"learning_rate": 4.985814731229722e-06,
"loss": 0.7125,
"step": 65500
},
{
"epoch": 1.19,
"learning_rate": 4.93088551904889e-06,
"loss": 0.7277,
"step": 66000
},
{
"epoch": 1.2,
"learning_rate": 4.875956306868058e-06,
"loss": 0.7701,
"step": 66500
},
{
"epoch": 1.21,
"learning_rate": 4.821027094687227e-06,
"loss": 0.8477,
"step": 67000
},
{
"epoch": 1.22,
"learning_rate": 4.766097882506396e-06,
"loss": 0.6889,
"step": 67500
},
{
"epoch": 1.23,
"learning_rate": 4.711168670325564e-06,
"loss": 0.7366,
"step": 68000
},
{
"epoch": 1.24,
"learning_rate": 4.656239458144733e-06,
"loss": 0.7619,
"step": 68500
},
{
"epoch": 1.24,
"learning_rate": 4.601310245963901e-06,
"loss": 0.7545,
"step": 69000
},
{
"epoch": 1.25,
"learning_rate": 4.546381033783069e-06,
"loss": 0.8154,
"step": 69500
},
{
"epoch": 1.26,
"learning_rate": 4.491451821602238e-06,
"loss": 0.7534,
"step": 70000
},
{
"epoch": 1.27,
"learning_rate": 4.436522609421406e-06,
"loss": 0.7664,
"step": 70500
},
{
"epoch": 1.28,
"learning_rate": 4.3815933972405745e-06,
"loss": 0.7648,
"step": 71000
},
{
"epoch": 1.29,
"learning_rate": 4.326664185059743e-06,
"loss": 0.7463,
"step": 71500
},
{
"epoch": 1.3,
"learning_rate": 4.271734972878911e-06,
"loss": 0.6641,
"step": 72000
},
{
"epoch": 1.31,
"learning_rate": 4.216805760698079e-06,
"loss": 0.8103,
"step": 72500
},
{
"epoch": 1.32,
"learning_rate": 4.161876548517247e-06,
"loss": 0.6905,
"step": 73000
},
{
"epoch": 1.33,
"learning_rate": 4.106947336336416e-06,
"loss": 0.7439,
"step": 73500
},
{
"epoch": 1.33,
"learning_rate": 4.052018124155585e-06,
"loss": 0.7842,
"step": 74000
},
{
"epoch": 1.34,
"learning_rate": 3.997088911974753e-06,
"loss": 0.7916,
"step": 74500
},
{
"epoch": 1.35,
"learning_rate": 3.942159699793922e-06,
"loss": 0.731,
"step": 75000
},
{
"epoch": 1.36,
"learning_rate": 3.88723048761309e-06,
"loss": 0.7484,
"step": 75500
},
{
"epoch": 1.37,
"learning_rate": 3.832301275432258e-06,
"loss": 0.8126,
"step": 76000
},
{
"epoch": 1.38,
"learning_rate": 3.777372063251427e-06,
"loss": 0.7575,
"step": 76500
},
{
"epoch": 1.39,
"learning_rate": 3.722442851070595e-06,
"loss": 0.758,
"step": 77000
},
{
"epoch": 1.4,
"learning_rate": 3.667513638889764e-06,
"loss": 0.7189,
"step": 77500
},
{
"epoch": 1.41,
"learning_rate": 3.6125844267089323e-06,
"loss": 0.7604,
"step": 78000
},
{
"epoch": 1.42,
"learning_rate": 3.5576552145281003e-06,
"loss": 0.7473,
"step": 78500
},
{
"epoch": 1.42,
"learning_rate": 3.502726002347269e-06,
"loss": 0.757,
"step": 79000
},
{
"epoch": 1.43,
"learning_rate": 3.4477967901664367e-06,
"loss": 0.7896,
"step": 79500
},
{
"epoch": 1.44,
"learning_rate": 3.3928675779856056e-06,
"loss": 0.8078,
"step": 80000
},
{
"epoch": 1.45,
"learning_rate": 3.3379383658047745e-06,
"loss": 0.8172,
"step": 80500
},
{
"epoch": 1.46,
"learning_rate": 3.283009153623942e-06,
"loss": 0.7771,
"step": 81000
},
{
"epoch": 1.47,
"learning_rate": 3.228079941443111e-06,
"loss": 0.7996,
"step": 81500
},
{
"epoch": 1.48,
"learning_rate": 3.1731507292622798e-06,
"loss": 0.7543,
"step": 82000
},
{
"epoch": 1.49,
"learning_rate": 3.1182215170814474e-06,
"loss": 0.7455,
"step": 82500
},
{
"epoch": 1.5,
"learning_rate": 3.0632923049006162e-06,
"loss": 0.7201,
"step": 83000
},
{
"epoch": 1.51,
"learning_rate": 3.0083630927197847e-06,
"loss": 0.7808,
"step": 83500
},
{
"epoch": 1.52,
"learning_rate": 2.9534338805389527e-06,
"loss": 0.7764,
"step": 84000
},
{
"epoch": 1.52,
"learning_rate": 2.898504668358121e-06,
"loss": 0.655,
"step": 84500
},
{
"epoch": 1.53,
"learning_rate": 2.84357545617729e-06,
"loss": 0.7855,
"step": 85000
},
{
"epoch": 1.54,
"learning_rate": 2.788646243996458e-06,
"loss": 0.7608,
"step": 85500
},
{
"epoch": 1.55,
"learning_rate": 2.7337170318156265e-06,
"loss": 0.8207,
"step": 86000
},
{
"epoch": 1.56,
"learning_rate": 2.678787819634795e-06,
"loss": 0.7595,
"step": 86500
},
{
"epoch": 1.57,
"learning_rate": 2.6238586074539634e-06,
"loss": 0.8041,
"step": 87000
},
{
"epoch": 1.58,
"learning_rate": 2.568929395273132e-06,
"loss": 0.6658,
"step": 87500
},
{
"epoch": 1.59,
"learning_rate": 2.5140001830923002e-06,
"loss": 0.7488,
"step": 88000
},
{
"epoch": 1.6,
"learning_rate": 2.4590709709114687e-06,
"loss": 0.7215,
"step": 88500
},
{
"epoch": 1.61,
"learning_rate": 2.4041417587306367e-06,
"loss": 0.7365,
"step": 89000
},
{
"epoch": 1.61,
"learning_rate": 2.3492125465498056e-06,
"loss": 0.8205,
"step": 89500
},
{
"epoch": 1.62,
"learning_rate": 2.294283334368974e-06,
"loss": 0.7802,
"step": 90000
},
{
"epoch": 1.63,
"learning_rate": 2.239354122188142e-06,
"loss": 0.7207,
"step": 90500
},
{
"epoch": 1.64,
"learning_rate": 2.1844249100073105e-06,
"loss": 0.8054,
"step": 91000
},
{
"epoch": 1.65,
"learning_rate": 2.1294956978264793e-06,
"loss": 0.7406,
"step": 91500
},
{
"epoch": 1.66,
"learning_rate": 2.0745664856456474e-06,
"loss": 0.741,
"step": 92000
},
{
"epoch": 1.67,
"learning_rate": 2.019637273464816e-06,
"loss": 0.7407,
"step": 92500
},
{
"epoch": 1.68,
"learning_rate": 1.9647080612839842e-06,
"loss": 0.7229,
"step": 93000
},
{
"epoch": 1.69,
"learning_rate": 1.9097788491031527e-06,
"loss": 0.7627,
"step": 93500
},
{
"epoch": 1.7,
"learning_rate": 1.8548496369223211e-06,
"loss": 0.6487,
"step": 94000
},
{
"epoch": 1.7,
"learning_rate": 1.7999204247414896e-06,
"loss": 0.773,
"step": 94500
},
{
"epoch": 1.71,
"learning_rate": 1.7449912125606578e-06,
"loss": 0.7363,
"step": 95000
},
{
"epoch": 1.72,
"learning_rate": 1.6900620003798264e-06,
"loss": 0.7422,
"step": 95500
},
{
"epoch": 1.73,
"learning_rate": 1.6351327881989949e-06,
"loss": 0.6581,
"step": 96000
},
{
"epoch": 1.74,
"learning_rate": 1.5802035760181631e-06,
"loss": 0.8168,
"step": 96500
},
{
"epoch": 1.75,
"learning_rate": 1.5252743638373316e-06,
"loss": 0.7826,
"step": 97000
},
{
"epoch": 1.76,
"learning_rate": 1.4703451516565e-06,
"loss": 0.7153,
"step": 97500
},
{
"epoch": 1.77,
"learning_rate": 1.4154159394756682e-06,
"loss": 0.7195,
"step": 98000
},
{
"epoch": 1.78,
"learning_rate": 1.3604867272948369e-06,
"loss": 0.7773,
"step": 98500
},
{
"epoch": 1.79,
"learning_rate": 1.3055575151140051e-06,
"loss": 0.6886,
"step": 99000
},
{
"epoch": 1.79,
"learning_rate": 1.2506283029331736e-06,
"loss": 0.7565,
"step": 99500
},
{
"epoch": 1.8,
"learning_rate": 1.195699090752342e-06,
"loss": 0.7796,
"step": 100000
},
{
"epoch": 1.81,
"learning_rate": 1.1407698785715104e-06,
"loss": 0.7513,
"step": 100500
},
{
"epoch": 1.82,
"learning_rate": 1.0858406663906789e-06,
"loss": 0.6886,
"step": 101000
},
{
"epoch": 1.83,
"learning_rate": 1.0309114542098473e-06,
"loss": 0.7378,
"step": 101500
},
{
"epoch": 1.84,
"learning_rate": 9.759822420290156e-07,
"loss": 0.6769,
"step": 102000
},
{
"epoch": 1.85,
"learning_rate": 9.210530298481841e-07,
"loss": 0.6617,
"step": 102500
},
{
"epoch": 1.86,
"learning_rate": 8.661238176673524e-07,
"loss": 0.6614,
"step": 103000
},
{
"epoch": 1.87,
"learning_rate": 8.11194605486521e-07,
"loss": 0.6952,
"step": 103500
},
{
"epoch": 1.88,
"learning_rate": 7.562653933056893e-07,
"loss": 0.7845,
"step": 104000
},
{
"epoch": 1.88,
"learning_rate": 7.013361811248578e-07,
"loss": 0.8129,
"step": 104500
},
{
"epoch": 1.89,
"learning_rate": 6.464069689440261e-07,
"loss": 0.6743,
"step": 105000
},
{
"epoch": 1.9,
"learning_rate": 5.914777567631945e-07,
"loss": 0.6937,
"step": 105500
},
{
"epoch": 1.91,
"learning_rate": 5.36548544582363e-07,
"loss": 0.7081,
"step": 106000
},
{
"epoch": 1.92,
"learning_rate": 4.816193324015313e-07,
"loss": 0.7547,
"step": 106500
},
{
"epoch": 1.93,
"learning_rate": 4.2669012022069976e-07,
"loss": 0.6936,
"step": 107000
},
{
"epoch": 1.94,
"learning_rate": 3.717609080398682e-07,
"loss": 0.7113,
"step": 107500
},
{
"epoch": 1.95,
"learning_rate": 3.168316958590366e-07,
"loss": 0.8074,
"step": 108000
},
{
"epoch": 1.96,
"learning_rate": 2.6190248367820503e-07,
"loss": 0.6368,
"step": 108500
},
{
"epoch": 1.97,
"learning_rate": 2.0697327149737345e-07,
"loss": 0.8334,
"step": 109000
},
{
"epoch": 1.98,
"learning_rate": 1.5204405931654183e-07,
"loss": 0.6496,
"step": 109500
},
{
"epoch": 1.98,
"learning_rate": 9.711484713571025e-08,
"loss": 0.7561,
"step": 110000
},
{
"epoch": 1.99,
"learning_rate": 4.218563495487866e-08,
"loss": 0.7072,
"step": 110500
},
{
"epoch": 2.0,
"eval_accuracy": 0.8014184397163121,
"eval_f1": 0.8014184397163121,
"eval_loss": 0.9014674425125122,
"eval_precision": 0.8014184397163121,
"eval_recall": 0.8014184397163121,
"eval_runtime": 105.7503,
"eval_samples_per_second": 155.999,
"step": 110884
}
],
"max_steps": 110884,
"num_train_epochs": 2,
"total_flos": 5755190618166264.0,
"trial_name": null,
"trial_params": null
}