5035 lines
136 KiB
JSON
5035 lines
136 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.05586155272771962,
|
|
"eval_steps": 500,
|
|
"global_step": 2500,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 10.69185152053833,
|
|
"epoch": 0.00011172310545543924,
|
|
"grad_norm": 18.5,
|
|
"learning_rate": 2e-06,
|
|
"loss": 10.5751,
|
|
"mean_token_accuracy": 0.00030030030757188797,
|
|
"num_tokens": 4264.0,
|
|
"step": 5
|
|
},
|
|
{
|
|
"entropy": 10.691861820220947,
|
|
"epoch": 0.00022344621091087847,
|
|
"grad_norm": 16.75,
|
|
"learning_rate": 4.5e-06,
|
|
"loss": 10.4625,
|
|
"mean_token_accuracy": 0.0,
|
|
"num_tokens": 8413.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 10.690832233428955,
|
|
"epoch": 0.0003351693163663177,
|
|
"grad_norm": 13.4375,
|
|
"learning_rate": 7e-06,
|
|
"loss": 10.1956,
|
|
"mean_token_accuracy": 0.03499803415033966,
|
|
"num_tokens": 12394.0,
|
|
"step": 15
|
|
},
|
|
{
|
|
"entropy": 10.681817626953125,
|
|
"epoch": 0.00044689242182175694,
|
|
"grad_norm": 8.4375,
|
|
"learning_rate": 9.5e-06,
|
|
"loss": 9.8891,
|
|
"mean_token_accuracy": 0.052673622593283655,
|
|
"num_tokens": 16610.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 10.648191356658936,
|
|
"epoch": 0.0005586155272771962,
|
|
"grad_norm": 5.84375,
|
|
"learning_rate": 1.2e-05,
|
|
"loss": 9.6415,
|
|
"mean_token_accuracy": 0.059712447226047516,
|
|
"num_tokens": 21243.0,
|
|
"step": 25
|
|
},
|
|
{
|
|
"entropy": 10.600524044036865,
|
|
"epoch": 0.0006703386327326354,
|
|
"grad_norm": 4.21875,
|
|
"learning_rate": 1.4500000000000002e-05,
|
|
"loss": 9.3955,
|
|
"mean_token_accuracy": 0.06313439272344112,
|
|
"num_tokens": 25571.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 10.565627956390381,
|
|
"epoch": 0.0007820617381880747,
|
|
"grad_norm": 4.28125,
|
|
"learning_rate": 1.7000000000000003e-05,
|
|
"loss": 9.2501,
|
|
"mean_token_accuracy": 0.058146678283810614,
|
|
"num_tokens": 29412.0,
|
|
"step": 35
|
|
},
|
|
{
|
|
"entropy": 10.537028217315674,
|
|
"epoch": 0.0008937848436435139,
|
|
"grad_norm": 3.796875,
|
|
"learning_rate": 1.95e-05,
|
|
"loss": 9.2173,
|
|
"mean_token_accuracy": 0.061766361258924006,
|
|
"num_tokens": 33893.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 10.526171112060547,
|
|
"epoch": 0.0010055079490989532,
|
|
"grad_norm": 3.765625,
|
|
"learning_rate": 2.2e-05,
|
|
"loss": 9.1499,
|
|
"mean_token_accuracy": 0.06647200398147106,
|
|
"num_tokens": 38043.0,
|
|
"step": 45
|
|
},
|
|
{
|
|
"entropy": 10.459559440612793,
|
|
"epoch": 0.0011172310545543925,
|
|
"grad_norm": 3.296875,
|
|
"learning_rate": 2.4500000000000003e-05,
|
|
"loss": 9.0927,
|
|
"mean_token_accuracy": 0.06208832561969757,
|
|
"num_tokens": 42222.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"entropy": 10.321250820159912,
|
|
"epoch": 0.0012289541600098315,
|
|
"grad_norm": 3.640625,
|
|
"learning_rate": 2.7e-05,
|
|
"loss": 8.9929,
|
|
"mean_token_accuracy": 0.06888641528785229,
|
|
"num_tokens": 46321.0,
|
|
"step": 55
|
|
},
|
|
{
|
|
"entropy": 10.221459579467773,
|
|
"epoch": 0.0013406772654652708,
|
|
"grad_norm": 3.140625,
|
|
"learning_rate": 2.95e-05,
|
|
"loss": 9.0121,
|
|
"mean_token_accuracy": 0.07621248066425323,
|
|
"num_tokens": 50816.0,
|
|
"step": 60
|
|
},
|
|
{
|
|
"entropy": 10.167527294158935,
|
|
"epoch": 0.0014524003709207101,
|
|
"grad_norm": 3.046875,
|
|
"learning_rate": 3.2e-05,
|
|
"loss": 8.9766,
|
|
"mean_token_accuracy": 0.06631188206374646,
|
|
"num_tokens": 55331.0,
|
|
"step": 65
|
|
},
|
|
{
|
|
"entropy": 10.158582305908203,
|
|
"epoch": 0.0015641234763761494,
|
|
"grad_norm": 2.828125,
|
|
"learning_rate": 3.4500000000000005e-05,
|
|
"loss": 8.8916,
|
|
"mean_token_accuracy": 0.06474936045706273,
|
|
"num_tokens": 59379.0,
|
|
"step": 70
|
|
},
|
|
{
|
|
"entropy": 10.132518291473389,
|
|
"epoch": 0.0016758465818315885,
|
|
"grad_norm": 2.671875,
|
|
"learning_rate": 3.7e-05,
|
|
"loss": 8.7691,
|
|
"mean_token_accuracy": 0.07134371958673,
|
|
"num_tokens": 63463.0,
|
|
"step": 75
|
|
},
|
|
{
|
|
"entropy": 10.115495491027833,
|
|
"epoch": 0.0017875696872870278,
|
|
"grad_norm": 2.625,
|
|
"learning_rate": 3.95e-05,
|
|
"loss": 8.8345,
|
|
"mean_token_accuracy": 0.07139550410211086,
|
|
"num_tokens": 67897.0,
|
|
"step": 80
|
|
},
|
|
{
|
|
"entropy": 10.076316165924073,
|
|
"epoch": 0.001899292792742467,
|
|
"grad_norm": 2.796875,
|
|
"learning_rate": 4.2000000000000004e-05,
|
|
"loss": 8.6081,
|
|
"mean_token_accuracy": 0.07515238299965858,
|
|
"num_tokens": 71995.0,
|
|
"step": 85
|
|
},
|
|
{
|
|
"entropy": 10.003674030303955,
|
|
"epoch": 0.0020110158981979064,
|
|
"grad_norm": 2.78125,
|
|
"learning_rate": 4.45e-05,
|
|
"loss": 8.4613,
|
|
"mean_token_accuracy": 0.07963024936616421,
|
|
"num_tokens": 75745.0,
|
|
"step": 90
|
|
},
|
|
{
|
|
"entropy": 9.955099964141846,
|
|
"epoch": 0.0021227390036533456,
|
|
"grad_norm": 2.34375,
|
|
"learning_rate": 4.7000000000000004e-05,
|
|
"loss": 8.4039,
|
|
"mean_token_accuracy": 0.0783398538827896,
|
|
"num_tokens": 79641.0,
|
|
"step": 95
|
|
},
|
|
{
|
|
"entropy": 9.907726192474366,
|
|
"epoch": 0.002234462109108785,
|
|
"grad_norm": 2.15625,
|
|
"learning_rate": 4.9500000000000004e-05,
|
|
"loss": 8.3639,
|
|
"mean_token_accuracy": 0.07564271241426468,
|
|
"num_tokens": 84368.0,
|
|
"step": 100
|
|
},
|
|
{
|
|
"entropy": 9.777185916900635,
|
|
"epoch": 0.0023461852145642242,
|
|
"grad_norm": 3.046875,
|
|
"learning_rate": 5.2e-05,
|
|
"loss": 8.3562,
|
|
"mean_token_accuracy": 0.07499068863689899,
|
|
"num_tokens": 89177.0,
|
|
"step": 105
|
|
},
|
|
{
|
|
"entropy": 9.874152565002442,
|
|
"epoch": 0.002457908320019663,
|
|
"grad_norm": 2.09375,
|
|
"learning_rate": 5.45e-05,
|
|
"loss": 8.1616,
|
|
"mean_token_accuracy": 0.08200128600001336,
|
|
"num_tokens": 93266.0,
|
|
"step": 110
|
|
},
|
|
{
|
|
"entropy": 9.573559665679932,
|
|
"epoch": 0.0025696314254751024,
|
|
"grad_norm": 2.515625,
|
|
"learning_rate": 5.7e-05,
|
|
"loss": 8.0628,
|
|
"mean_token_accuracy": 0.08064724728465081,
|
|
"num_tokens": 97385.0,
|
|
"step": 115
|
|
},
|
|
{
|
|
"entropy": 9.477795314788818,
|
|
"epoch": 0.0026813545309305417,
|
|
"grad_norm": 2.03125,
|
|
"learning_rate": 5.9499999999999996e-05,
|
|
"loss": 7.8909,
|
|
"mean_token_accuracy": 0.08364077880978585,
|
|
"num_tokens": 100911.0,
|
|
"step": 120
|
|
},
|
|
{
|
|
"entropy": 9.406207180023193,
|
|
"epoch": 0.002793077636385981,
|
|
"grad_norm": 1.765625,
|
|
"learning_rate": 6.2e-05,
|
|
"loss": 7.8101,
|
|
"mean_token_accuracy": 0.08597532883286477,
|
|
"num_tokens": 104785.0,
|
|
"step": 125
|
|
},
|
|
{
|
|
"entropy": 9.198468017578126,
|
|
"epoch": 0.0029048007418414202,
|
|
"grad_norm": 2.265625,
|
|
"learning_rate": 6.450000000000001e-05,
|
|
"loss": 7.7507,
|
|
"mean_token_accuracy": 0.08100780844688416,
|
|
"num_tokens": 109434.0,
|
|
"step": 130
|
|
},
|
|
{
|
|
"entropy": 9.137882709503174,
|
|
"epoch": 0.0030165238472968595,
|
|
"grad_norm": 1.734375,
|
|
"learning_rate": 6.7e-05,
|
|
"loss": 7.7526,
|
|
"mean_token_accuracy": 0.08647556751966476,
|
|
"num_tokens": 113869.0,
|
|
"step": 135
|
|
},
|
|
{
|
|
"entropy": 8.982115554809571,
|
|
"epoch": 0.003128246952752299,
|
|
"grad_norm": 1.828125,
|
|
"learning_rate": 6.950000000000001e-05,
|
|
"loss": 7.6855,
|
|
"mean_token_accuracy": 0.0863442875444889,
|
|
"num_tokens": 118332.0,
|
|
"step": 140
|
|
},
|
|
{
|
|
"entropy": 8.803977680206298,
|
|
"epoch": 0.003239970058207738,
|
|
"grad_norm": 1.8828125,
|
|
"learning_rate": 7.2e-05,
|
|
"loss": 7.5951,
|
|
"mean_token_accuracy": 0.07965997867286205,
|
|
"num_tokens": 122140.0,
|
|
"step": 145
|
|
},
|
|
{
|
|
"entropy": 8.717557144165038,
|
|
"epoch": 0.003351693163663177,
|
|
"grad_norm": 1.796875,
|
|
"learning_rate": 7.45e-05,
|
|
"loss": 7.4489,
|
|
"mean_token_accuracy": 0.08853708282113075,
|
|
"num_tokens": 126515.0,
|
|
"step": 150
|
|
},
|
|
{
|
|
"entropy": 8.480121898651124,
|
|
"epoch": 0.0034634162691186163,
|
|
"grad_norm": 1.9375,
|
|
"learning_rate": 7.7e-05,
|
|
"loss": 7.5078,
|
|
"mean_token_accuracy": 0.0842710092663765,
|
|
"num_tokens": 130397.0,
|
|
"step": 155
|
|
},
|
|
{
|
|
"entropy": 8.55941562652588,
|
|
"epoch": 0.0035751393745740555,
|
|
"grad_norm": 1.796875,
|
|
"learning_rate": 7.950000000000001e-05,
|
|
"loss": 7.448,
|
|
"mean_token_accuracy": 0.086439348757267,
|
|
"num_tokens": 134428.0,
|
|
"step": 160
|
|
},
|
|
{
|
|
"entropy": 8.38999900817871,
|
|
"epoch": 0.003686862480029495,
|
|
"grad_norm": 2.0625,
|
|
"learning_rate": 8.2e-05,
|
|
"loss": 7.3523,
|
|
"mean_token_accuracy": 0.09324855357408524,
|
|
"num_tokens": 138294.0,
|
|
"step": 165
|
|
},
|
|
{
|
|
"entropy": 8.155976009368896,
|
|
"epoch": 0.003798585585484934,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 8.450000000000001e-05,
|
|
"loss": 7.3474,
|
|
"mean_token_accuracy": 0.08743810728192329,
|
|
"num_tokens": 142750.0,
|
|
"step": 170
|
|
},
|
|
{
|
|
"entropy": 8.256286525726319,
|
|
"epoch": 0.003910308690940373,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 8.7e-05,
|
|
"loss": 7.3177,
|
|
"mean_token_accuracy": 0.08836911171674729,
|
|
"num_tokens": 147220.0,
|
|
"step": 175
|
|
},
|
|
{
|
|
"entropy": 7.98900089263916,
|
|
"epoch": 0.004022031796395813,
|
|
"grad_norm": 1.765625,
|
|
"learning_rate": 8.95e-05,
|
|
"loss": 7.2059,
|
|
"mean_token_accuracy": 0.09879823476076126,
|
|
"num_tokens": 151396.0,
|
|
"step": 180
|
|
},
|
|
{
|
|
"entropy": 8.085029649734498,
|
|
"epoch": 0.0041337549018512516,
|
|
"grad_norm": 1.859375,
|
|
"learning_rate": 9.2e-05,
|
|
"loss": 7.2565,
|
|
"mean_token_accuracy": 0.0990697756409645,
|
|
"num_tokens": 155336.0,
|
|
"step": 185
|
|
},
|
|
{
|
|
"entropy": 8.046411085128785,
|
|
"epoch": 0.004245478007306691,
|
|
"grad_norm": 3.03125,
|
|
"learning_rate": 9.45e-05,
|
|
"loss": 7.2963,
|
|
"mean_token_accuracy": 0.08894443586468696,
|
|
"num_tokens": 159752.0,
|
|
"step": 190
|
|
},
|
|
{
|
|
"entropy": 8.043665552139283,
|
|
"epoch": 0.00435720111276213,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 9.7e-05,
|
|
"loss": 7.3874,
|
|
"mean_token_accuracy": 0.08465041853487491,
|
|
"num_tokens": 163895.0,
|
|
"step": 195
|
|
},
|
|
{
|
|
"entropy": 7.914860630035401,
|
|
"epoch": 0.00446892421821757,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 9.95e-05,
|
|
"loss": 7.1927,
|
|
"mean_token_accuracy": 0.093280029296875,
|
|
"num_tokens": 168312.0,
|
|
"step": 200
|
|
},
|
|
{
|
|
"entropy": 7.915866088867188,
|
|
"epoch": 0.004580647323673009,
|
|
"grad_norm": 1.9375,
|
|
"learning_rate": 0.000102,
|
|
"loss": 7.2334,
|
|
"mean_token_accuracy": 0.09746119827032089,
|
|
"num_tokens": 172358.0,
|
|
"step": 205
|
|
},
|
|
{
|
|
"entropy": 7.660448884963989,
|
|
"epoch": 0.0046923704291284484,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 0.00010449999999999999,
|
|
"loss": 7.2206,
|
|
"mean_token_accuracy": 0.09569770470261574,
|
|
"num_tokens": 177085.0,
|
|
"step": 210
|
|
},
|
|
{
|
|
"entropy": 7.943923139572144,
|
|
"epoch": 0.004804093534583887,
|
|
"grad_norm": 2.015625,
|
|
"learning_rate": 0.000107,
|
|
"loss": 7.3482,
|
|
"mean_token_accuracy": 0.08808798864483833,
|
|
"num_tokens": 181410.0,
|
|
"step": 215
|
|
},
|
|
{
|
|
"entropy": 7.640953445434571,
|
|
"epoch": 0.004915816640039326,
|
|
"grad_norm": 1.6875,
|
|
"learning_rate": 0.0001095,
|
|
"loss": 7.0931,
|
|
"mean_token_accuracy": 0.0959208883345127,
|
|
"num_tokens": 185649.0,
|
|
"step": 220
|
|
},
|
|
{
|
|
"entropy": 7.714695310592651,
|
|
"epoch": 0.005027539745494766,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 0.000112,
|
|
"loss": 7.1461,
|
|
"mean_token_accuracy": 0.09635853916406631,
|
|
"num_tokens": 190055.0,
|
|
"step": 225
|
|
},
|
|
{
|
|
"entropy": 7.722128820419312,
|
|
"epoch": 0.005139262850950205,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 0.0001145,
|
|
"loss": 7.1545,
|
|
"mean_token_accuracy": 0.10052503794431686,
|
|
"num_tokens": 194267.0,
|
|
"step": 230
|
|
},
|
|
{
|
|
"entropy": 7.67925763130188,
|
|
"epoch": 0.0052509859564056445,
|
|
"grad_norm": 1.78125,
|
|
"learning_rate": 0.00011700000000000001,
|
|
"loss": 7.0976,
|
|
"mean_token_accuracy": 0.09923605695366859,
|
|
"num_tokens": 198483.0,
|
|
"step": 235
|
|
},
|
|
{
|
|
"entropy": 7.5812575340271,
|
|
"epoch": 0.005362709061861083,
|
|
"grad_norm": 1.5703125,
|
|
"learning_rate": 0.00011949999999999999,
|
|
"loss": 7.0173,
|
|
"mean_token_accuracy": 0.10283883661031723,
|
|
"num_tokens": 202514.0,
|
|
"step": 240
|
|
},
|
|
{
|
|
"entropy": 7.598950576782227,
|
|
"epoch": 0.005474432167316523,
|
|
"grad_norm": 1.796875,
|
|
"learning_rate": 0.000122,
|
|
"loss": 7.1592,
|
|
"mean_token_accuracy": 0.09925142973661423,
|
|
"num_tokens": 206748.0,
|
|
"step": 245
|
|
},
|
|
{
|
|
"entropy": 7.573145723342895,
|
|
"epoch": 0.005586155272771962,
|
|
"grad_norm": 1.875,
|
|
"learning_rate": 0.0001245,
|
|
"loss": 7.0229,
|
|
"mean_token_accuracy": 0.10259764194488526,
|
|
"num_tokens": 210981.0,
|
|
"step": 250
|
|
},
|
|
{
|
|
"entropy": 7.5371081829071045,
|
|
"epoch": 0.005697878378227402,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 0.000127,
|
|
"loss": 7.1679,
|
|
"mean_token_accuracy": 0.09469496235251426,
|
|
"num_tokens": 215551.0,
|
|
"step": 255
|
|
},
|
|
{
|
|
"entropy": 7.520426607131958,
|
|
"epoch": 0.0058096014836828405,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.0001295,
|
|
"loss": 6.9739,
|
|
"mean_token_accuracy": 0.11107034981250763,
|
|
"num_tokens": 219677.0,
|
|
"step": 260
|
|
},
|
|
{
|
|
"entropy": 7.612557506561279,
|
|
"epoch": 0.005921324589138279,
|
|
"grad_norm": 1.75,
|
|
"learning_rate": 0.000132,
|
|
"loss": 7.072,
|
|
"mean_token_accuracy": 0.10267340168356895,
|
|
"num_tokens": 223457.0,
|
|
"step": 265
|
|
},
|
|
{
|
|
"entropy": 7.546310329437256,
|
|
"epoch": 0.006033047694593719,
|
|
"grad_norm": 1.734375,
|
|
"learning_rate": 0.00013450000000000002,
|
|
"loss": 6.9226,
|
|
"mean_token_accuracy": 0.1154308445751667,
|
|
"num_tokens": 227135.0,
|
|
"step": 270
|
|
},
|
|
{
|
|
"entropy": 7.572003746032715,
|
|
"epoch": 0.006144770800049158,
|
|
"grad_norm": 1.6328125,
|
|
"learning_rate": 0.00013700000000000002,
|
|
"loss": 7.1372,
|
|
"mean_token_accuracy": 0.09673138484358787,
|
|
"num_tokens": 231393.0,
|
|
"step": 275
|
|
},
|
|
{
|
|
"entropy": 7.611022663116455,
|
|
"epoch": 0.006256493905504598,
|
|
"grad_norm": 1.6484375,
|
|
"learning_rate": 0.0001395,
|
|
"loss": 7.1052,
|
|
"mean_token_accuracy": 0.10218687430024147,
|
|
"num_tokens": 235959.0,
|
|
"step": 280
|
|
},
|
|
{
|
|
"entropy": 7.403985118865966,
|
|
"epoch": 0.0063682170109600365,
|
|
"grad_norm": 1.734375,
|
|
"learning_rate": 0.00014199999999999998,
|
|
"loss": 6.9434,
|
|
"mean_token_accuracy": 0.10860099717974663,
|
|
"num_tokens": 239875.0,
|
|
"step": 285
|
|
},
|
|
{
|
|
"entropy": 7.420692682266235,
|
|
"epoch": 0.006479940116415476,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.0001445,
|
|
"loss": 7.0656,
|
|
"mean_token_accuracy": 0.11386927664279937,
|
|
"num_tokens": 244397.0,
|
|
"step": 290
|
|
},
|
|
{
|
|
"entropy": 7.5308808326721195,
|
|
"epoch": 0.006591663221870915,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 0.000147,
|
|
"loss": 7.0,
|
|
"mean_token_accuracy": 0.10292632281780242,
|
|
"num_tokens": 248441.0,
|
|
"step": 295
|
|
},
|
|
{
|
|
"entropy": 7.551111650466919,
|
|
"epoch": 0.006703386327326354,
|
|
"grad_norm": 2.0625,
|
|
"learning_rate": 0.0001495,
|
|
"loss": 7.0078,
|
|
"mean_token_accuracy": 0.10877085924148559,
|
|
"num_tokens": 252693.0,
|
|
"step": 300
|
|
},
|
|
{
|
|
"entropy": 7.420186233520508,
|
|
"epoch": 0.006815109432781794,
|
|
"grad_norm": 1.96875,
|
|
"learning_rate": 0.000152,
|
|
"loss": 6.9778,
|
|
"mean_token_accuracy": 0.1077166736125946,
|
|
"num_tokens": 257135.0,
|
|
"step": 305
|
|
},
|
|
{
|
|
"entropy": 7.455549049377441,
|
|
"epoch": 0.0069268325382372325,
|
|
"grad_norm": 1.7421875,
|
|
"learning_rate": 0.00015450000000000001,
|
|
"loss": 7.101,
|
|
"mean_token_accuracy": 0.09917000532150269,
|
|
"num_tokens": 261919.0,
|
|
"step": 310
|
|
},
|
|
{
|
|
"entropy": 7.428725242614746,
|
|
"epoch": 0.007038555643692672,
|
|
"grad_norm": 1.71875,
|
|
"learning_rate": 0.000157,
|
|
"loss": 7.0018,
|
|
"mean_token_accuracy": 0.1032940112054348,
|
|
"num_tokens": 266656.0,
|
|
"step": 315
|
|
},
|
|
{
|
|
"entropy": 7.4056861877441404,
|
|
"epoch": 0.007150278749148111,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 0.0001595,
|
|
"loss": 6.9649,
|
|
"mean_token_accuracy": 0.10402853935956954,
|
|
"num_tokens": 270879.0,
|
|
"step": 320
|
|
},
|
|
{
|
|
"entropy": 7.356624603271484,
|
|
"epoch": 0.007262001854603551,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 0.000162,
|
|
"loss": 6.9156,
|
|
"mean_token_accuracy": 0.11899577528238296,
|
|
"num_tokens": 274884.0,
|
|
"step": 325
|
|
},
|
|
{
|
|
"entropy": 7.359720230102539,
|
|
"epoch": 0.00737372496005899,
|
|
"grad_norm": 1.6953125,
|
|
"learning_rate": 0.00016450000000000001,
|
|
"loss": 6.8492,
|
|
"mean_token_accuracy": 0.10880193561315536,
|
|
"num_tokens": 279096.0,
|
|
"step": 330
|
|
},
|
|
{
|
|
"entropy": 7.420133590698242,
|
|
"epoch": 0.007485448065514429,
|
|
"grad_norm": 1.875,
|
|
"learning_rate": 0.00016700000000000002,
|
|
"loss": 6.9547,
|
|
"mean_token_accuracy": 0.10481383726000786,
|
|
"num_tokens": 282923.0,
|
|
"step": 335
|
|
},
|
|
{
|
|
"entropy": 7.375938367843628,
|
|
"epoch": 0.007597171170969868,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 0.00016950000000000003,
|
|
"loss": 6.8854,
|
|
"mean_token_accuracy": 0.11555279865860939,
|
|
"num_tokens": 287283.0,
|
|
"step": 340
|
|
},
|
|
{
|
|
"entropy": 7.266809701919556,
|
|
"epoch": 0.007708894276425307,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 0.00017199999999999998,
|
|
"loss": 6.8566,
|
|
"mean_token_accuracy": 0.11106687858700752,
|
|
"num_tokens": 291409.0,
|
|
"step": 345
|
|
},
|
|
{
|
|
"entropy": 7.351815128326416,
|
|
"epoch": 0.007820617381880746,
|
|
"grad_norm": 1.640625,
|
|
"learning_rate": 0.00017449999999999999,
|
|
"loss": 6.9102,
|
|
"mean_token_accuracy": 0.10942614153027534,
|
|
"num_tokens": 295157.0,
|
|
"step": 350
|
|
},
|
|
{
|
|
"entropy": 7.481018209457398,
|
|
"epoch": 0.007932340487336187,
|
|
"grad_norm": 1.9765625,
|
|
"learning_rate": 0.000177,
|
|
"loss": 6.9203,
|
|
"mean_token_accuracy": 0.10715582817792893,
|
|
"num_tokens": 299370.0,
|
|
"step": 355
|
|
},
|
|
{
|
|
"entropy": 7.2193928241729735,
|
|
"epoch": 0.008044063592791625,
|
|
"grad_norm": 1.765625,
|
|
"learning_rate": 0.0001795,
|
|
"loss": 6.9256,
|
|
"mean_token_accuracy": 0.11601178944110871,
|
|
"num_tokens": 303955.0,
|
|
"step": 360
|
|
},
|
|
{
|
|
"entropy": 7.46614179611206,
|
|
"epoch": 0.008155786698247064,
|
|
"grad_norm": 2.28125,
|
|
"learning_rate": 0.000182,
|
|
"loss": 6.8992,
|
|
"mean_token_accuracy": 0.11738249063491821,
|
|
"num_tokens": 308047.0,
|
|
"step": 365
|
|
},
|
|
{
|
|
"entropy": 7.339001560211182,
|
|
"epoch": 0.008267509803702503,
|
|
"grad_norm": 1.8359375,
|
|
"learning_rate": 0.0001845,
|
|
"loss": 6.8353,
|
|
"mean_token_accuracy": 0.12007508799433708,
|
|
"num_tokens": 312050.0,
|
|
"step": 370
|
|
},
|
|
{
|
|
"entropy": 7.393199253082275,
|
|
"epoch": 0.008379232909157944,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 0.000187,
|
|
"loss": 6.789,
|
|
"mean_token_accuracy": 0.11748188808560371,
|
|
"num_tokens": 315901.0,
|
|
"step": 375
|
|
},
|
|
{
|
|
"entropy": 7.206725311279297,
|
|
"epoch": 0.008490956014613383,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 0.0001895,
|
|
"loss": 6.8387,
|
|
"mean_token_accuracy": 0.11932774782180786,
|
|
"num_tokens": 320077.0,
|
|
"step": 380
|
|
},
|
|
{
|
|
"entropy": 7.309018516540528,
|
|
"epoch": 0.008602679120068821,
|
|
"grad_norm": 1.859375,
|
|
"learning_rate": 0.000192,
|
|
"loss": 6.7772,
|
|
"mean_token_accuracy": 0.12389342486858368,
|
|
"num_tokens": 324365.0,
|
|
"step": 385
|
|
},
|
|
{
|
|
"entropy": 7.1502117156982425,
|
|
"epoch": 0.00871440222552426,
|
|
"grad_norm": 1.7421875,
|
|
"learning_rate": 0.0001945,
|
|
"loss": 6.7342,
|
|
"mean_token_accuracy": 0.12538171485066413,
|
|
"num_tokens": 327971.0,
|
|
"step": 390
|
|
},
|
|
{
|
|
"entropy": 7.207239198684692,
|
|
"epoch": 0.0088261253309797,
|
|
"grad_norm": 1.8046875,
|
|
"learning_rate": 0.00019700000000000002,
|
|
"loss": 6.7814,
|
|
"mean_token_accuracy": 0.12070702090859413,
|
|
"num_tokens": 331791.0,
|
|
"step": 395
|
|
},
|
|
{
|
|
"entropy": 7.188077354431153,
|
|
"epoch": 0.00893784843643514,
|
|
"grad_norm": 1.7265625,
|
|
"learning_rate": 0.00019950000000000002,
|
|
"loss": 6.8347,
|
|
"mean_token_accuracy": 0.11772776916623115,
|
|
"num_tokens": 335616.0,
|
|
"step": 400
|
|
},
|
|
{
|
|
"entropy": 7.238067245483398,
|
|
"epoch": 0.009049571541890579,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.000202,
|
|
"loss": 6.7448,
|
|
"mean_token_accuracy": 0.12914832159876824,
|
|
"num_tokens": 339815.0,
|
|
"step": 405
|
|
},
|
|
{
|
|
"entropy": 7.271863889694214,
|
|
"epoch": 0.009161294647346017,
|
|
"grad_norm": 1.6875,
|
|
"learning_rate": 0.00020449999999999998,
|
|
"loss": 6.7314,
|
|
"mean_token_accuracy": 0.12098882421851158,
|
|
"num_tokens": 344045.0,
|
|
"step": 410
|
|
},
|
|
{
|
|
"entropy": 7.054260921478272,
|
|
"epoch": 0.009273017752801456,
|
|
"grad_norm": 1.9296875,
|
|
"learning_rate": 0.000207,
|
|
"loss": 6.8258,
|
|
"mean_token_accuracy": 0.12370770499110222,
|
|
"num_tokens": 348726.0,
|
|
"step": 415
|
|
},
|
|
{
|
|
"entropy": 7.25892219543457,
|
|
"epoch": 0.009384740858256897,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 0.0002095,
|
|
"loss": 6.811,
|
|
"mean_token_accuracy": 0.12439991980791092,
|
|
"num_tokens": 353228.0,
|
|
"step": 420
|
|
},
|
|
{
|
|
"entropy": 7.089362478256225,
|
|
"epoch": 0.009496463963712336,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 0.000212,
|
|
"loss": 6.7731,
|
|
"mean_token_accuracy": 0.12979045361280442,
|
|
"num_tokens": 357379.0,
|
|
"step": 425
|
|
},
|
|
{
|
|
"entropy": 7.264574861526489,
|
|
"epoch": 0.009608187069167775,
|
|
"grad_norm": 2.015625,
|
|
"learning_rate": 0.0002145,
|
|
"loss": 6.7426,
|
|
"mean_token_accuracy": 0.11879147663712501,
|
|
"num_tokens": 361347.0,
|
|
"step": 430
|
|
},
|
|
{
|
|
"entropy": 7.1300811767578125,
|
|
"epoch": 0.009719910174623213,
|
|
"grad_norm": 1.7265625,
|
|
"learning_rate": 0.00021700000000000002,
|
|
"loss": 6.7019,
|
|
"mean_token_accuracy": 0.12289151027798653,
|
|
"num_tokens": 365364.0,
|
|
"step": 435
|
|
},
|
|
{
|
|
"entropy": 7.1097527027130125,
|
|
"epoch": 0.009831633280078652,
|
|
"grad_norm": 1.6171875,
|
|
"learning_rate": 0.0002195,
|
|
"loss": 6.7534,
|
|
"mean_token_accuracy": 0.12068467438220978,
|
|
"num_tokens": 369648.0,
|
|
"step": 440
|
|
},
|
|
{
|
|
"entropy": 7.229344892501831,
|
|
"epoch": 0.009943356385534093,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 0.000222,
|
|
"loss": 6.7309,
|
|
"mean_token_accuracy": 0.12263526245951653,
|
|
"num_tokens": 373734.0,
|
|
"step": 445
|
|
},
|
|
{
|
|
"entropy": 7.1849524021148685,
|
|
"epoch": 0.010055079490989532,
|
|
"grad_norm": 1.765625,
|
|
"learning_rate": 0.0002245,
|
|
"loss": 6.7276,
|
|
"mean_token_accuracy": 0.12742977142333983,
|
|
"num_tokens": 378184.0,
|
|
"step": 450
|
|
},
|
|
{
|
|
"entropy": 7.064081621170044,
|
|
"epoch": 0.01016680259644497,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 0.00022700000000000002,
|
|
"loss": 6.8377,
|
|
"mean_token_accuracy": 0.12056938409805298,
|
|
"num_tokens": 382686.0,
|
|
"step": 455
|
|
},
|
|
{
|
|
"entropy": 7.035848474502563,
|
|
"epoch": 0.01027852570190041,
|
|
"grad_norm": 1.84375,
|
|
"learning_rate": 0.00022950000000000002,
|
|
"loss": 6.7467,
|
|
"mean_token_accuracy": 0.13145566657185553,
|
|
"num_tokens": 387210.0,
|
|
"step": 460
|
|
},
|
|
{
|
|
"entropy": 7.06201696395874,
|
|
"epoch": 0.01039024880735585,
|
|
"grad_norm": 1.734375,
|
|
"learning_rate": 0.00023200000000000003,
|
|
"loss": 6.7263,
|
|
"mean_token_accuracy": 0.12420300841331482,
|
|
"num_tokens": 391247.0,
|
|
"step": 465
|
|
},
|
|
{
|
|
"entropy": 7.2231800079345705,
|
|
"epoch": 0.010501971912811289,
|
|
"grad_norm": 1.8125,
|
|
"learning_rate": 0.00023449999999999998,
|
|
"loss": 6.6101,
|
|
"mean_token_accuracy": 0.12484818920493126,
|
|
"num_tokens": 395188.0,
|
|
"step": 470
|
|
},
|
|
{
|
|
"entropy": 6.963715267181397,
|
|
"epoch": 0.010613695018266728,
|
|
"grad_norm": 1.7734375,
|
|
"learning_rate": 0.000237,
|
|
"loss": 6.6759,
|
|
"mean_token_accuracy": 0.1323227606713772,
|
|
"num_tokens": 399513.0,
|
|
"step": 475
|
|
},
|
|
{
|
|
"entropy": 7.180826234817505,
|
|
"epoch": 0.010725418123722167,
|
|
"grad_norm": 1.6875,
|
|
"learning_rate": 0.0002395,
|
|
"loss": 6.7397,
|
|
"mean_token_accuracy": 0.1241152174770832,
|
|
"num_tokens": 404052.0,
|
|
"step": 480
|
|
},
|
|
{
|
|
"entropy": 7.000790357589722,
|
|
"epoch": 0.010837141229177606,
|
|
"grad_norm": 1.6875,
|
|
"learning_rate": 0.000242,
|
|
"loss": 6.6031,
|
|
"mean_token_accuracy": 0.13010914325714112,
|
|
"num_tokens": 408299.0,
|
|
"step": 485
|
|
},
|
|
{
|
|
"entropy": 6.951791143417358,
|
|
"epoch": 0.010948864334633046,
|
|
"grad_norm": 1.890625,
|
|
"learning_rate": 0.0002445,
|
|
"loss": 6.597,
|
|
"mean_token_accuracy": 0.13655143678188325,
|
|
"num_tokens": 412376.0,
|
|
"step": 490
|
|
},
|
|
{
|
|
"entropy": 6.940329837799072,
|
|
"epoch": 0.011060587440088485,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.000247,
|
|
"loss": 6.6731,
|
|
"mean_token_accuracy": 0.13289386332035064,
|
|
"num_tokens": 417010.0,
|
|
"step": 495
|
|
},
|
|
{
|
|
"entropy": 7.088097476959229,
|
|
"epoch": 0.011172310545543924,
|
|
"grad_norm": 1.796875,
|
|
"learning_rate": 0.0002495,
|
|
"loss": 6.5194,
|
|
"mean_token_accuracy": 0.14706601724028587,
|
|
"num_tokens": 421167.0,
|
|
"step": 500
|
|
},
|
|
{
|
|
"entropy": 6.922516012191773,
|
|
"epoch": 0.011284033650999363,
|
|
"grad_norm": 1.703125,
|
|
"learning_rate": 0.000252,
|
|
"loss": 6.6178,
|
|
"mean_token_accuracy": 0.13619204014539718,
|
|
"num_tokens": 425942.0,
|
|
"step": 505
|
|
},
|
|
{
|
|
"entropy": 7.03532772064209,
|
|
"epoch": 0.011395756756454803,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 0.0002545,
|
|
"loss": 6.6648,
|
|
"mean_token_accuracy": 0.13011146634817122,
|
|
"num_tokens": 430529.0,
|
|
"step": 510
|
|
},
|
|
{
|
|
"entropy": 6.945074939727784,
|
|
"epoch": 0.011507479861910242,
|
|
"grad_norm": 1.828125,
|
|
"learning_rate": 0.000257,
|
|
"loss": 6.5181,
|
|
"mean_token_accuracy": 0.12947220131754875,
|
|
"num_tokens": 434464.0,
|
|
"step": 515
|
|
},
|
|
{
|
|
"entropy": 6.959482002258301,
|
|
"epoch": 0.011619202967365681,
|
|
"grad_norm": 1.7421875,
|
|
"learning_rate": 0.0002595,
|
|
"loss": 6.6253,
|
|
"mean_token_accuracy": 0.13869500905275345,
|
|
"num_tokens": 438763.0,
|
|
"step": 520
|
|
},
|
|
{
|
|
"entropy": 6.914068031311035,
|
|
"epoch": 0.01173092607282112,
|
|
"grad_norm": 1.6875,
|
|
"learning_rate": 0.000262,
|
|
"loss": 6.5756,
|
|
"mean_token_accuracy": 0.1341098003089428,
|
|
"num_tokens": 442945.0,
|
|
"step": 525
|
|
},
|
|
{
|
|
"entropy": 7.105465316772461,
|
|
"epoch": 0.011842649178276559,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 0.00026450000000000003,
|
|
"loss": 6.6796,
|
|
"mean_token_accuracy": 0.12466075941920281,
|
|
"num_tokens": 447301.0,
|
|
"step": 530
|
|
},
|
|
{
|
|
"entropy": 6.853219223022461,
|
|
"epoch": 0.011954372283732,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 0.00026700000000000004,
|
|
"loss": 6.5349,
|
|
"mean_token_accuracy": 0.13604073524475097,
|
|
"num_tokens": 451365.0,
|
|
"step": 535
|
|
},
|
|
{
|
|
"entropy": 6.9650640964508055,
|
|
"epoch": 0.012066095389187438,
|
|
"grad_norm": 1.7734375,
|
|
"learning_rate": 0.00026950000000000005,
|
|
"loss": 6.5806,
|
|
"mean_token_accuracy": 0.13407194539904593,
|
|
"num_tokens": 455469.0,
|
|
"step": 540
|
|
},
|
|
{
|
|
"entropy": 6.94967565536499,
|
|
"epoch": 0.012177818494642877,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 0.00027200000000000005,
|
|
"loss": 6.6521,
|
|
"mean_token_accuracy": 0.13110672906041146,
|
|
"num_tokens": 459641.0,
|
|
"step": 545
|
|
},
|
|
{
|
|
"entropy": 6.912797451019287,
|
|
"epoch": 0.012289541600098316,
|
|
"grad_norm": 1.6796875,
|
|
"learning_rate": 0.0002745,
|
|
"loss": 6.5912,
|
|
"mean_token_accuracy": 0.1325491264462471,
|
|
"num_tokens": 463866.0,
|
|
"step": 550
|
|
},
|
|
{
|
|
"entropy": 6.911716938018799,
|
|
"epoch": 0.012401264705553756,
|
|
"grad_norm": 1.921875,
|
|
"learning_rate": 0.000277,
|
|
"loss": 6.4776,
|
|
"mean_token_accuracy": 0.1369607262313366,
|
|
"num_tokens": 467692.0,
|
|
"step": 555
|
|
},
|
|
{
|
|
"entropy": 6.93188853263855,
|
|
"epoch": 0.012512987811009195,
|
|
"grad_norm": 1.84375,
|
|
"learning_rate": 0.0002795,
|
|
"loss": 6.611,
|
|
"mean_token_accuracy": 0.1433392733335495,
|
|
"num_tokens": 471719.0,
|
|
"step": 560
|
|
},
|
|
{
|
|
"entropy": 6.887971878051758,
|
|
"epoch": 0.012624710916464634,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 0.00028199999999999997,
|
|
"loss": 6.5399,
|
|
"mean_token_accuracy": 0.13064411506056786,
|
|
"num_tokens": 475978.0,
|
|
"step": 565
|
|
},
|
|
{
|
|
"entropy": 6.961443710327148,
|
|
"epoch": 0.012736434021920073,
|
|
"grad_norm": 1.6171875,
|
|
"learning_rate": 0.0002845,
|
|
"loss": 6.6299,
|
|
"mean_token_accuracy": 0.1282844863831997,
|
|
"num_tokens": 480538.0,
|
|
"step": 570
|
|
},
|
|
{
|
|
"entropy": 6.9848442554473875,
|
|
"epoch": 0.012848157127375512,
|
|
"grad_norm": 1.75,
|
|
"learning_rate": 0.000287,
|
|
"loss": 6.5282,
|
|
"mean_token_accuracy": 0.13599977716803552,
|
|
"num_tokens": 484604.0,
|
|
"step": 575
|
|
},
|
|
{
|
|
"entropy": 6.9366988182067875,
|
|
"epoch": 0.012959880232830952,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.0002895,
|
|
"loss": 6.5757,
|
|
"mean_token_accuracy": 0.12946860045194625,
|
|
"num_tokens": 489117.0,
|
|
"step": 580
|
|
},
|
|
{
|
|
"entropy": 6.9284679889678955,
|
|
"epoch": 0.013071603338286391,
|
|
"grad_norm": 1.90625,
|
|
"learning_rate": 0.000292,
|
|
"loss": 6.454,
|
|
"mean_token_accuracy": 0.13936082720756532,
|
|
"num_tokens": 493020.0,
|
|
"step": 585
|
|
},
|
|
{
|
|
"entropy": 6.81143798828125,
|
|
"epoch": 0.01318332644374183,
|
|
"grad_norm": 1.765625,
|
|
"learning_rate": 0.0002945,
|
|
"loss": 6.6045,
|
|
"mean_token_accuracy": 0.1356659173965454,
|
|
"num_tokens": 497224.0,
|
|
"step": 590
|
|
},
|
|
{
|
|
"entropy": 6.871704006195069,
|
|
"epoch": 0.013295049549197269,
|
|
"grad_norm": 1.640625,
|
|
"learning_rate": 0.000297,
|
|
"loss": 6.5519,
|
|
"mean_token_accuracy": 0.1327143579721451,
|
|
"num_tokens": 501323.0,
|
|
"step": 595
|
|
},
|
|
{
|
|
"entropy": 6.897035217285156,
|
|
"epoch": 0.013406772654652708,
|
|
"grad_norm": 1.765625,
|
|
"learning_rate": 0.0002995,
|
|
"loss": 6.6141,
|
|
"mean_token_accuracy": 0.13350649401545525,
|
|
"num_tokens": 505544.0,
|
|
"step": 600
|
|
},
|
|
{
|
|
"entropy": 6.890090608596802,
|
|
"epoch": 0.013518495760108148,
|
|
"grad_norm": 1.6796875,
|
|
"learning_rate": 0.000302,
|
|
"loss": 6.5533,
|
|
"mean_token_accuracy": 0.13723233789205552,
|
|
"num_tokens": 509980.0,
|
|
"step": 605
|
|
},
|
|
{
|
|
"entropy": 6.711225366592407,
|
|
"epoch": 0.013630218865563587,
|
|
"grad_norm": 1.7265625,
|
|
"learning_rate": 0.0003045,
|
|
"loss": 6.4153,
|
|
"mean_token_accuracy": 0.14487348720431328,
|
|
"num_tokens": 514264.0,
|
|
"step": 610
|
|
},
|
|
{
|
|
"entropy": 6.843616533279419,
|
|
"epoch": 0.013741941971019026,
|
|
"grad_norm": 1.8359375,
|
|
"learning_rate": 0.000307,
|
|
"loss": 6.498,
|
|
"mean_token_accuracy": 0.13974266350269318,
|
|
"num_tokens": 518482.0,
|
|
"step": 615
|
|
},
|
|
{
|
|
"entropy": 6.874301528930664,
|
|
"epoch": 0.013853665076474465,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 0.0003095,
|
|
"loss": 6.5465,
|
|
"mean_token_accuracy": 0.13032930791378022,
|
|
"num_tokens": 522765.0,
|
|
"step": 620
|
|
},
|
|
{
|
|
"entropy": 6.749671411514282,
|
|
"epoch": 0.013965388181929906,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.000312,
|
|
"loss": 6.369,
|
|
"mean_token_accuracy": 0.14341583847999573,
|
|
"num_tokens": 526794.0,
|
|
"step": 625
|
|
},
|
|
{
|
|
"entropy": 6.787525320053101,
|
|
"epoch": 0.014077111287385344,
|
|
"grad_norm": 1.640625,
|
|
"learning_rate": 0.0003145,
|
|
"loss": 6.528,
|
|
"mean_token_accuracy": 0.14095058366656305,
|
|
"num_tokens": 531070.0,
|
|
"step": 630
|
|
},
|
|
{
|
|
"entropy": 6.8752782344818115,
|
|
"epoch": 0.014188834392840783,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.000317,
|
|
"loss": 6.5556,
|
|
"mean_token_accuracy": 0.1387290321290493,
|
|
"num_tokens": 535655.0,
|
|
"step": 635
|
|
},
|
|
{
|
|
"entropy": 6.793731594085694,
|
|
"epoch": 0.014300557498296222,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.0003195,
|
|
"loss": 6.5852,
|
|
"mean_token_accuracy": 0.13649069741368294,
|
|
"num_tokens": 540231.0,
|
|
"step": 640
|
|
},
|
|
{
|
|
"entropy": 6.764686965942383,
|
|
"epoch": 0.014412280603751661,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 0.000322,
|
|
"loss": 6.4429,
|
|
"mean_token_accuracy": 0.14507341384887695,
|
|
"num_tokens": 544180.0,
|
|
"step": 645
|
|
},
|
|
{
|
|
"entropy": 6.750355339050293,
|
|
"epoch": 0.014524003709207102,
|
|
"grad_norm": 1.703125,
|
|
"learning_rate": 0.00032450000000000003,
|
|
"loss": 6.413,
|
|
"mean_token_accuracy": 0.14032018259167672,
|
|
"num_tokens": 547905.0,
|
|
"step": 650
|
|
},
|
|
{
|
|
"entropy": 6.633134460449218,
|
|
"epoch": 0.01463572681466254,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 0.00032700000000000003,
|
|
"loss": 6.4971,
|
|
"mean_token_accuracy": 0.14342024102807044,
|
|
"num_tokens": 552248.0,
|
|
"step": 655
|
|
},
|
|
{
|
|
"entropy": 6.69278130531311,
|
|
"epoch": 0.01474744992011798,
|
|
"grad_norm": 1.8671875,
|
|
"learning_rate": 0.00032950000000000004,
|
|
"loss": 6.3103,
|
|
"mean_token_accuracy": 0.15002134442329407,
|
|
"num_tokens": 556430.0,
|
|
"step": 660
|
|
},
|
|
{
|
|
"entropy": 6.606019735336304,
|
|
"epoch": 0.014859173025573418,
|
|
"grad_norm": 1.796875,
|
|
"learning_rate": 0.00033200000000000005,
|
|
"loss": 6.4126,
|
|
"mean_token_accuracy": 0.14947109222412108,
|
|
"num_tokens": 560114.0,
|
|
"step": 665
|
|
},
|
|
{
|
|
"entropy": 6.456874704360962,
|
|
"epoch": 0.014970896131028859,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 0.00033450000000000005,
|
|
"loss": 6.432,
|
|
"mean_token_accuracy": 0.1395042322576046,
|
|
"num_tokens": 564624.0,
|
|
"step": 670
|
|
},
|
|
{
|
|
"entropy": 6.712696886062622,
|
|
"epoch": 0.015082619236484298,
|
|
"grad_norm": 1.7265625,
|
|
"learning_rate": 0.000337,
|
|
"loss": 6.3378,
|
|
"mean_token_accuracy": 0.1465103656053543,
|
|
"num_tokens": 568918.0,
|
|
"step": 675
|
|
},
|
|
{
|
|
"entropy": 6.637308168411255,
|
|
"epoch": 0.015194342341939737,
|
|
"grad_norm": 1.6171875,
|
|
"learning_rate": 0.0003395,
|
|
"loss": 6.4062,
|
|
"mean_token_accuracy": 0.14732098281383516,
|
|
"num_tokens": 572972.0,
|
|
"step": 680
|
|
},
|
|
{
|
|
"entropy": 6.601571273803711,
|
|
"epoch": 0.015306065447395175,
|
|
"grad_norm": 1.65625,
|
|
"learning_rate": 0.000342,
|
|
"loss": 6.4433,
|
|
"mean_token_accuracy": 0.150725756585598,
|
|
"num_tokens": 577250.0,
|
|
"step": 685
|
|
},
|
|
{
|
|
"entropy": 6.742843151092529,
|
|
"epoch": 0.015417788552850614,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 0.00034449999999999997,
|
|
"loss": 6.343,
|
|
"mean_token_accuracy": 0.14816580414772035,
|
|
"num_tokens": 581418.0,
|
|
"step": 690
|
|
},
|
|
{
|
|
"entropy": 6.67585711479187,
|
|
"epoch": 0.015529511658306055,
|
|
"grad_norm": 1.796875,
|
|
"learning_rate": 0.000347,
|
|
"loss": 6.4489,
|
|
"mean_token_accuracy": 0.14566415771842003,
|
|
"num_tokens": 585675.0,
|
|
"step": 695
|
|
},
|
|
{
|
|
"entropy": 6.479352045059204,
|
|
"epoch": 0.015641234763761492,
|
|
"grad_norm": 1.6640625,
|
|
"learning_rate": 0.0003495,
|
|
"loss": 6.35,
|
|
"mean_token_accuracy": 0.1425018362700939,
|
|
"num_tokens": 590032.0,
|
|
"step": 700
|
|
},
|
|
{
|
|
"entropy": 6.537110137939453,
|
|
"epoch": 0.015752957869216933,
|
|
"grad_norm": 1.7578125,
|
|
"learning_rate": 0.000352,
|
|
"loss": 6.3371,
|
|
"mean_token_accuracy": 0.15346041396260263,
|
|
"num_tokens": 594185.0,
|
|
"step": 705
|
|
},
|
|
{
|
|
"entropy": 6.561723518371582,
|
|
"epoch": 0.015864680974672373,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 0.0003545,
|
|
"loss": 6.2759,
|
|
"mean_token_accuracy": 0.15528617724776267,
|
|
"num_tokens": 598246.0,
|
|
"step": 710
|
|
},
|
|
{
|
|
"entropy": 6.575086498260498,
|
|
"epoch": 0.01597640408012781,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 0.000357,
|
|
"loss": 6.4107,
|
|
"mean_token_accuracy": 0.14987436532974244,
|
|
"num_tokens": 602676.0,
|
|
"step": 715
|
|
},
|
|
{
|
|
"entropy": 6.5285728931427,
|
|
"epoch": 0.01608812718558325,
|
|
"grad_norm": 1.6171875,
|
|
"learning_rate": 0.0003595,
|
|
"loss": 6.2682,
|
|
"mean_token_accuracy": 0.15596837252378465,
|
|
"num_tokens": 606693.0,
|
|
"step": 720
|
|
},
|
|
{
|
|
"entropy": 6.538558626174927,
|
|
"epoch": 0.01619985029103869,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 0.000362,
|
|
"loss": 6.3284,
|
|
"mean_token_accuracy": 0.13880535289645196,
|
|
"num_tokens": 610780.0,
|
|
"step": 725
|
|
},
|
|
{
|
|
"entropy": 6.484487867355346,
|
|
"epoch": 0.01631157339649413,
|
|
"grad_norm": 1.640625,
|
|
"learning_rate": 0.0003645,
|
|
"loss": 6.365,
|
|
"mean_token_accuracy": 0.148425579816103,
|
|
"num_tokens": 614746.0,
|
|
"step": 730
|
|
},
|
|
{
|
|
"entropy": 6.57484130859375,
|
|
"epoch": 0.01642329650194957,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.000367,
|
|
"loss": 6.3217,
|
|
"mean_token_accuracy": 0.1501600444316864,
|
|
"num_tokens": 619222.0,
|
|
"step": 735
|
|
},
|
|
{
|
|
"entropy": 6.699773979187012,
|
|
"epoch": 0.016535019607405006,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.0003695,
|
|
"loss": 6.4575,
|
|
"mean_token_accuracy": 0.1470234327018261,
|
|
"num_tokens": 623746.0,
|
|
"step": 740
|
|
},
|
|
{
|
|
"entropy": 6.542045211791992,
|
|
"epoch": 0.016646742712860447,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.000372,
|
|
"loss": 6.3108,
|
|
"mean_token_accuracy": 0.15406130626797676,
|
|
"num_tokens": 628392.0,
|
|
"step": 745
|
|
},
|
|
{
|
|
"entropy": 6.443634414672852,
|
|
"epoch": 0.016758465818315887,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 0.0003745,
|
|
"loss": 6.3252,
|
|
"mean_token_accuracy": 0.1483919471502304,
|
|
"num_tokens": 632439.0,
|
|
"step": 750
|
|
},
|
|
{
|
|
"entropy": 6.462843322753907,
|
|
"epoch": 0.016870188923771325,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.000377,
|
|
"loss": 6.3639,
|
|
"mean_token_accuracy": 0.14475394636392594,
|
|
"num_tokens": 636801.0,
|
|
"step": 755
|
|
},
|
|
{
|
|
"entropy": 6.5299022674560545,
|
|
"epoch": 0.016981912029226765,
|
|
"grad_norm": 1.71875,
|
|
"learning_rate": 0.0003795,
|
|
"loss": 6.1376,
|
|
"mean_token_accuracy": 0.16324445307254792,
|
|
"num_tokens": 640818.0,
|
|
"step": 760
|
|
},
|
|
{
|
|
"entropy": 6.448936939239502,
|
|
"epoch": 0.017093635134682202,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 0.000382,
|
|
"loss": 6.3332,
|
|
"mean_token_accuracy": 0.1428905963897705,
|
|
"num_tokens": 644914.0,
|
|
"step": 765
|
|
},
|
|
{
|
|
"entropy": 6.681206798553466,
|
|
"epoch": 0.017205358240137643,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 0.0003845,
|
|
"loss": 6.4011,
|
|
"mean_token_accuracy": 0.1510247729718685,
|
|
"num_tokens": 649368.0,
|
|
"step": 770
|
|
},
|
|
{
|
|
"entropy": 6.323616456985474,
|
|
"epoch": 0.017317081345593083,
|
|
"grad_norm": 1.6796875,
|
|
"learning_rate": 0.00038700000000000003,
|
|
"loss": 6.2719,
|
|
"mean_token_accuracy": 0.1547230713069439,
|
|
"num_tokens": 653537.0,
|
|
"step": 775
|
|
},
|
|
{
|
|
"entropy": 6.604541254043579,
|
|
"epoch": 0.01742880445104852,
|
|
"grad_norm": 1.625,
|
|
"learning_rate": 0.00038950000000000003,
|
|
"loss": 6.3935,
|
|
"mean_token_accuracy": 0.1366790473461151,
|
|
"num_tokens": 658232.0,
|
|
"step": 780
|
|
},
|
|
{
|
|
"entropy": 6.442852210998535,
|
|
"epoch": 0.01754052755650396,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.00039200000000000004,
|
|
"loss": 6.3431,
|
|
"mean_token_accuracy": 0.1412769600749016,
|
|
"num_tokens": 662512.0,
|
|
"step": 785
|
|
},
|
|
{
|
|
"entropy": 6.617573547363281,
|
|
"epoch": 0.0176522506619594,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.00039450000000000005,
|
|
"loss": 6.317,
|
|
"mean_token_accuracy": 0.15297196432948112,
|
|
"num_tokens": 666759.0,
|
|
"step": 790
|
|
},
|
|
{
|
|
"entropy": 6.2801289558410645,
|
|
"epoch": 0.01776397376741484,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 0.00039700000000000005,
|
|
"loss": 6.2359,
|
|
"mean_token_accuracy": 0.15913384407758713,
|
|
"num_tokens": 670870.0,
|
|
"step": 795
|
|
},
|
|
{
|
|
"entropy": 6.4234747886657715,
|
|
"epoch": 0.01787569687287028,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 0.0003995,
|
|
"loss": 6.2002,
|
|
"mean_token_accuracy": 0.15476782992482185,
|
|
"num_tokens": 675150.0,
|
|
"step": 800
|
|
},
|
|
{
|
|
"entropy": 6.242582082748413,
|
|
"epoch": 0.017987419978325717,
|
|
"grad_norm": 1.6953125,
|
|
"learning_rate": 0.000402,
|
|
"loss": 6.1478,
|
|
"mean_token_accuracy": 0.16312235742807388,
|
|
"num_tokens": 679083.0,
|
|
"step": 805
|
|
},
|
|
{
|
|
"entropy": 6.390512132644654,
|
|
"epoch": 0.018099143083781157,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 0.0004045,
|
|
"loss": 6.3339,
|
|
"mean_token_accuracy": 0.14840762168169022,
|
|
"num_tokens": 683477.0,
|
|
"step": 810
|
|
},
|
|
{
|
|
"entropy": 6.346619319915772,
|
|
"epoch": 0.018210866189236594,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 0.00040699999999999997,
|
|
"loss": 6.1433,
|
|
"mean_token_accuracy": 0.15934223383665086,
|
|
"num_tokens": 687819.0,
|
|
"step": 815
|
|
},
|
|
{
|
|
"entropy": 6.47499303817749,
|
|
"epoch": 0.018322589294692035,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.0004095,
|
|
"loss": 6.3563,
|
|
"mean_token_accuracy": 0.15130842924118043,
|
|
"num_tokens": 692017.0,
|
|
"step": 820
|
|
},
|
|
{
|
|
"entropy": 6.57763991355896,
|
|
"epoch": 0.018434312400147475,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.000412,
|
|
"loss": 6.3471,
|
|
"mean_token_accuracy": 0.1486021101474762,
|
|
"num_tokens": 696853.0,
|
|
"step": 825
|
|
},
|
|
{
|
|
"entropy": 6.5021766185760494,
|
|
"epoch": 0.018546035505602913,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.0004145,
|
|
"loss": 6.4022,
|
|
"mean_token_accuracy": 0.13754147440195083,
|
|
"num_tokens": 701131.0,
|
|
"step": 830
|
|
},
|
|
{
|
|
"entropy": 6.329108572006225,
|
|
"epoch": 0.018657758611058353,
|
|
"grad_norm": 1.7109375,
|
|
"learning_rate": 0.000417,
|
|
"loss": 6.3438,
|
|
"mean_token_accuracy": 0.14031846597790718,
|
|
"num_tokens": 705428.0,
|
|
"step": 835
|
|
},
|
|
{
|
|
"entropy": 6.359441089630127,
|
|
"epoch": 0.018769481716513794,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.0004195,
|
|
"loss": 6.1931,
|
|
"mean_token_accuracy": 0.16036254167556763,
|
|
"num_tokens": 709760.0,
|
|
"step": 840
|
|
},
|
|
{
|
|
"entropy": 6.268203020095825,
|
|
"epoch": 0.01888120482196923,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.000422,
|
|
"loss": 6.1527,
|
|
"mean_token_accuracy": 0.16414240151643752,
|
|
"num_tokens": 713964.0,
|
|
"step": 845
|
|
},
|
|
{
|
|
"entropy": 6.395886707305908,
|
|
"epoch": 0.01899292792742467,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 0.0004245,
|
|
"loss": 6.3232,
|
|
"mean_token_accuracy": 0.15316279977560043,
|
|
"num_tokens": 718305.0,
|
|
"step": 850
|
|
},
|
|
{
|
|
"entropy": 6.378367567062378,
|
|
"epoch": 0.01910465103288011,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 0.000427,
|
|
"loss": 6.2283,
|
|
"mean_token_accuracy": 0.15503655076026918,
|
|
"num_tokens": 722705.0,
|
|
"step": 855
|
|
},
|
|
{
|
|
"entropy": 6.2019463062286375,
|
|
"epoch": 0.01921637413833555,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.0004295,
|
|
"loss": 6.1186,
|
|
"mean_token_accuracy": 0.16495653316378595,
|
|
"num_tokens": 726529.0,
|
|
"step": 860
|
|
},
|
|
{
|
|
"entropy": 6.353125715255738,
|
|
"epoch": 0.01932809724379099,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 0.000432,
|
|
"loss": 6.2228,
|
|
"mean_token_accuracy": 0.15027105957269668,
|
|
"num_tokens": 730535.0,
|
|
"step": 865
|
|
},
|
|
{
|
|
"entropy": 6.341917848587036,
|
|
"epoch": 0.019439820349246427,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.0004345,
|
|
"loss": 6.1982,
|
|
"mean_token_accuracy": 0.15795509219169618,
|
|
"num_tokens": 734836.0,
|
|
"step": 870
|
|
},
|
|
{
|
|
"entropy": 6.281454563140869,
|
|
"epoch": 0.019551543454701868,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.000437,
|
|
"loss": 6.2734,
|
|
"mean_token_accuracy": 0.15110851228237152,
|
|
"num_tokens": 739176.0,
|
|
"step": 875
|
|
},
|
|
{
|
|
"entropy": 6.2023228168487545,
|
|
"epoch": 0.019663266560157305,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 0.0004395,
|
|
"loss": 6.2473,
|
|
"mean_token_accuracy": 0.15088646113872528,
|
|
"num_tokens": 743497.0,
|
|
"step": 880
|
|
},
|
|
{
|
|
"entropy": 6.359633541107177,
|
|
"epoch": 0.019774989665612745,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 0.000442,
|
|
"loss": 6.2555,
|
|
"mean_token_accuracy": 0.15800251960754394,
|
|
"num_tokens": 747185.0,
|
|
"step": 885
|
|
},
|
|
{
|
|
"entropy": 6.299527931213379,
|
|
"epoch": 0.019886712771068186,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 0.0004445,
|
|
"loss": 6.3705,
|
|
"mean_token_accuracy": 0.1479213811457157,
|
|
"num_tokens": 751345.0,
|
|
"step": 890
|
|
},
|
|
{
|
|
"entropy": 6.348771095275879,
|
|
"epoch": 0.019998435876523623,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 0.000447,
|
|
"loss": 6.2641,
|
|
"mean_token_accuracy": 0.1576582223176956,
|
|
"num_tokens": 755618.0,
|
|
"step": 895
|
|
},
|
|
{
|
|
"entropy": 6.318030929565429,
|
|
"epoch": 0.020110158981979064,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 0.00044950000000000003,
|
|
"loss": 6.2539,
|
|
"mean_token_accuracy": 0.15391020849347115,
|
|
"num_tokens": 759583.0,
|
|
"step": 900
|
|
},
|
|
{
|
|
"entropy": 6.339395999908447,
|
|
"epoch": 0.0202218820874345,
|
|
"grad_norm": 1.5625,
|
|
"learning_rate": 0.00045200000000000004,
|
|
"loss": 6.1856,
|
|
"mean_token_accuracy": 0.16274880915880202,
|
|
"num_tokens": 763861.0,
|
|
"step": 905
|
|
},
|
|
{
|
|
"entropy": 6.222519493103027,
|
|
"epoch": 0.02033360519288994,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.00045450000000000004,
|
|
"loss": 6.1597,
|
|
"mean_token_accuracy": 0.16332034766674042,
|
|
"num_tokens": 768046.0,
|
|
"step": 910
|
|
},
|
|
{
|
|
"entropy": 6.363137149810791,
|
|
"epoch": 0.020445328298345382,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 0.00045700000000000005,
|
|
"loss": 6.1448,
|
|
"mean_token_accuracy": 0.15950068086385727,
|
|
"num_tokens": 772570.0,
|
|
"step": 915
|
|
},
|
|
{
|
|
"entropy": 6.369995403289795,
|
|
"epoch": 0.02055705140380082,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.00045950000000000006,
|
|
"loss": 6.2981,
|
|
"mean_token_accuracy": 0.1562412589788437,
|
|
"num_tokens": 777324.0,
|
|
"step": 920
|
|
},
|
|
{
|
|
"entropy": 6.437484455108643,
|
|
"epoch": 0.02066877450925626,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.000462,
|
|
"loss": 6.273,
|
|
"mean_token_accuracy": 0.14749887958168983,
|
|
"num_tokens": 781829.0,
|
|
"step": 925
|
|
},
|
|
{
|
|
"entropy": 6.290871667861938,
|
|
"epoch": 0.0207804976147117,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.0004645,
|
|
"loss": 6.139,
|
|
"mean_token_accuracy": 0.15945375710725784,
|
|
"num_tokens": 786089.0,
|
|
"step": 930
|
|
},
|
|
{
|
|
"entropy": 6.188903331756592,
|
|
"epoch": 0.020892220720167137,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.000467,
|
|
"loss": 6.135,
|
|
"mean_token_accuracy": 0.16696064472198485,
|
|
"num_tokens": 790273.0,
|
|
"step": 935
|
|
},
|
|
{
|
|
"entropy": 6.331109428405762,
|
|
"epoch": 0.021003943825622578,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.0004695,
|
|
"loss": 6.1418,
|
|
"mean_token_accuracy": 0.16708167791366577,
|
|
"num_tokens": 794600.0,
|
|
"step": 940
|
|
},
|
|
{
|
|
"entropy": 6.217516183853149,
|
|
"epoch": 0.021115666931078015,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 0.000472,
|
|
"loss": 6.1176,
|
|
"mean_token_accuracy": 0.1658843532204628,
|
|
"num_tokens": 798756.0,
|
|
"step": 945
|
|
},
|
|
{
|
|
"entropy": 6.395580863952636,
|
|
"epoch": 0.021227390036533456,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.0004745,
|
|
"loss": 6.1259,
|
|
"mean_token_accuracy": 0.16052196100354194,
|
|
"num_tokens": 803122.0,
|
|
"step": 950
|
|
},
|
|
{
|
|
"entropy": 6.081810760498047,
|
|
"epoch": 0.021339113141988896,
|
|
"grad_norm": 1.625,
|
|
"learning_rate": 0.000477,
|
|
"loss": 6.164,
|
|
"mean_token_accuracy": 0.16162788346409798,
|
|
"num_tokens": 807393.0,
|
|
"step": 955
|
|
},
|
|
{
|
|
"entropy": 6.248359203338623,
|
|
"epoch": 0.021450836247444333,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 0.0004795,
|
|
"loss": 6.1023,
|
|
"mean_token_accuracy": 0.16735041439533233,
|
|
"num_tokens": 811735.0,
|
|
"step": 960
|
|
},
|
|
{
|
|
"entropy": 6.347312164306641,
|
|
"epoch": 0.021562559352899774,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.000482,
|
|
"loss": 6.309,
|
|
"mean_token_accuracy": 0.15177485793828965,
|
|
"num_tokens": 816327.0,
|
|
"step": 965
|
|
},
|
|
{
|
|
"entropy": 6.2607207775115965,
|
|
"epoch": 0.02167428245835521,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 0.0004845,
|
|
"loss": 6.0873,
|
|
"mean_token_accuracy": 0.16931966543197632,
|
|
"num_tokens": 820476.0,
|
|
"step": 970
|
|
},
|
|
{
|
|
"entropy": 6.144009685516357,
|
|
"epoch": 0.02178600556381065,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.000487,
|
|
"loss": 6.1081,
|
|
"mean_token_accuracy": 0.16204737722873688,
|
|
"num_tokens": 824563.0,
|
|
"step": 975
|
|
},
|
|
{
|
|
"entropy": 6.143129920959472,
|
|
"epoch": 0.021897728669266092,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.0004895,
|
|
"loss": 6.2294,
|
|
"mean_token_accuracy": 0.15659647509455682,
|
|
"num_tokens": 829307.0,
|
|
"step": 980
|
|
},
|
|
{
|
|
"entropy": 6.235242652893066,
|
|
"epoch": 0.02200945177472153,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.000492,
|
|
"loss": 6.2047,
|
|
"mean_token_accuracy": 0.15783022493124008,
|
|
"num_tokens": 833876.0,
|
|
"step": 985
|
|
},
|
|
{
|
|
"entropy": 6.273316383361816,
|
|
"epoch": 0.02212117488017697,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.0004945,
|
|
"loss": 6.2855,
|
|
"mean_token_accuracy": 0.15405729338526725,
|
|
"num_tokens": 838484.0,
|
|
"step": 990
|
|
},
|
|
{
|
|
"entropy": 6.237334489822388,
|
|
"epoch": 0.022232897985632407,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 0.000497,
|
|
"loss": 6.1138,
|
|
"mean_token_accuracy": 0.15904016494750978,
|
|
"num_tokens": 842813.0,
|
|
"step": 995
|
|
},
|
|
{
|
|
"entropy": 6.182960510253906,
|
|
"epoch": 0.022344621091087848,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 0.0004995,
|
|
"loss": 6.0759,
|
|
"mean_token_accuracy": 0.16501354575157165,
|
|
"num_tokens": 846638.0,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"entropy": 6.176758861541748,
|
|
"epoch": 0.022456344196543288,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 0.000499998026082006,
|
|
"loss": 6.0475,
|
|
"mean_token_accuracy": 0.1734422266483307,
|
|
"num_tokens": 850577.0,
|
|
"step": 1005
|
|
},
|
|
{
|
|
"entropy": 6.231747055053711,
|
|
"epoch": 0.022568067301998725,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.0004999900070995136,
|
|
"loss": 6.0858,
|
|
"mean_token_accuracy": 0.16294893622398376,
|
|
"num_tokens": 854904.0,
|
|
"step": 1010
|
|
},
|
|
{
|
|
"entropy": 6.121539163589477,
|
|
"epoch": 0.022679790407454166,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.0004999758199023239,
|
|
"loss": 6.081,
|
|
"mean_token_accuracy": 0.16951507180929185,
|
|
"num_tokens": 858859.0,
|
|
"step": 1015
|
|
},
|
|
{
|
|
"entropy": 6.083382177352905,
|
|
"epoch": 0.022791513512909607,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 0.0004999554648793858,
|
|
"loss": 6.0267,
|
|
"mean_token_accuracy": 0.17120128124952316,
|
|
"num_tokens": 863121.0,
|
|
"step": 1020
|
|
},
|
|
{
|
|
"entropy": 6.1603466987609865,
|
|
"epoch": 0.022903236618365044,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 0.0004999289425887425,
|
|
"loss": 6.1551,
|
|
"mean_token_accuracy": 0.15693418234586715,
|
|
"num_tokens": 867805.0,
|
|
"step": 1025
|
|
},
|
|
{
|
|
"entropy": 6.104901504516602,
|
|
"epoch": 0.023014959723820484,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.0004998962537575161,
|
|
"loss": 6.1175,
|
|
"mean_token_accuracy": 0.15979633256793022,
|
|
"num_tokens": 871988.0,
|
|
"step": 1030
|
|
},
|
|
{
|
|
"entropy": 6.240704679489136,
|
|
"epoch": 0.02312668282927592,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.0004998573992818874,
|
|
"loss": 6.0985,
|
|
"mean_token_accuracy": 0.15933127254247664,
|
|
"num_tokens": 876181.0,
|
|
"step": 1035
|
|
},
|
|
{
|
|
"entropy": 5.971805095672607,
|
|
"epoch": 0.023238405934731362,
|
|
"grad_norm": 1.9765625,
|
|
"learning_rate": 0.0004998123802270715,
|
|
"loss": 5.9561,
|
|
"mean_token_accuracy": 0.178880412876606,
|
|
"num_tokens": 880521.0,
|
|
"step": 1040
|
|
},
|
|
{
|
|
"entropy": 5.992498254776001,
|
|
"epoch": 0.023350129040186803,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.0004997611978272886,
|
|
"loss": 6.0967,
|
|
"mean_token_accuracy": 0.16714801341295243,
|
|
"num_tokens": 885006.0,
|
|
"step": 1045
|
|
},
|
|
{
|
|
"entropy": 6.120235300064087,
|
|
"epoch": 0.02346185214564224,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.0004997038534857298,
|
|
"loss": 6.0853,
|
|
"mean_token_accuracy": 0.16448681205511093,
|
|
"num_tokens": 889184.0,
|
|
"step": 1050
|
|
},
|
|
{
|
|
"entropy": 6.119455003738404,
|
|
"epoch": 0.02357357525109768,
|
|
"grad_norm": 1.546875,
|
|
"learning_rate": 0.0004996403487745194,
|
|
"loss": 6.0012,
|
|
"mean_token_accuracy": 0.17143121510744094,
|
|
"num_tokens": 893806.0,
|
|
"step": 1055
|
|
},
|
|
{
|
|
"entropy": 6.055149364471435,
|
|
"epoch": 0.023685298356553117,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.000499570685434671,
|
|
"loss": 6.0071,
|
|
"mean_token_accuracy": 0.16834600120782853,
|
|
"num_tokens": 897578.0,
|
|
"step": 1060
|
|
},
|
|
{
|
|
"entropy": 6.227984523773193,
|
|
"epoch": 0.023797021462008558,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.0004994948653760405,
|
|
"loss": 6.2897,
|
|
"mean_token_accuracy": 0.15649080276489258,
|
|
"num_tokens": 902225.0,
|
|
"step": 1065
|
|
},
|
|
{
|
|
"entropy": 6.1664598941802975,
|
|
"epoch": 0.023908744567464,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.0004994128906772729,
|
|
"loss": 6.1097,
|
|
"mean_token_accuracy": 0.16405512094497682,
|
|
"num_tokens": 906477.0,
|
|
"step": 1070
|
|
},
|
|
{
|
|
"entropy": 6.110595417022705,
|
|
"epoch": 0.024020467672919436,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.000499324763585746,
|
|
"loss": 6.1071,
|
|
"mean_token_accuracy": 0.15544368103146552,
|
|
"num_tokens": 910948.0,
|
|
"step": 1075
|
|
},
|
|
{
|
|
"entropy": 6.197442293167114,
|
|
"epoch": 0.024132190778374876,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 0.0004992304865175085,
|
|
"loss": 6.2299,
|
|
"mean_token_accuracy": 0.1479525662958622,
|
|
"num_tokens": 915157.0,
|
|
"step": 1080
|
|
},
|
|
{
|
|
"entropy": 6.140390396118164,
|
|
"epoch": 0.024243913883830313,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.0004991300620572138,
|
|
"loss": 6.0491,
|
|
"mean_token_accuracy": 0.16512299925088883,
|
|
"num_tokens": 920251.0,
|
|
"step": 1085
|
|
},
|
|
{
|
|
"entropy": 6.003958606719971,
|
|
"epoch": 0.024355636989285754,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.0004990234929580494,
|
|
"loss": 6.1163,
|
|
"mean_token_accuracy": 0.15651995092630386,
|
|
"num_tokens": 924510.0,
|
|
"step": 1090
|
|
},
|
|
{
|
|
"entropy": 6.04054217338562,
|
|
"epoch": 0.024467360094741195,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.0004989107821416609,
|
|
"loss": 6.0239,
|
|
"mean_token_accuracy": 0.16540934443473815,
|
|
"num_tokens": 928925.0,
|
|
"step": 1095
|
|
},
|
|
{
|
|
"entropy": 6.079621315002441,
|
|
"epoch": 0.02457908320019663,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.0004987919326980723,
|
|
"loss": 6.0004,
|
|
"mean_token_accuracy": 0.17250575870275497,
|
|
"num_tokens": 933211.0,
|
|
"step": 1100
|
|
},
|
|
{
|
|
"entropy": 6.115182685852051,
|
|
"epoch": 0.024690806305652072,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.0004986669478856011,
|
|
"loss": 6.1689,
|
|
"mean_token_accuracy": 0.16165492981672286,
|
|
"num_tokens": 937621.0,
|
|
"step": 1105
|
|
},
|
|
{
|
|
"entropy": 6.006317281723023,
|
|
"epoch": 0.024802529411107513,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.0004985358311307688,
|
|
"loss": 6.087,
|
|
"mean_token_accuracy": 0.15697798803448676,
|
|
"num_tokens": 941629.0,
|
|
"step": 1110
|
|
},
|
|
{
|
|
"entropy": 6.04360728263855,
|
|
"epoch": 0.02491425251656295,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.0004983985860282081,
|
|
"loss": 6.0674,
|
|
"mean_token_accuracy": 0.15867182463407517,
|
|
"num_tokens": 945853.0,
|
|
"step": 1115
|
|
},
|
|
{
|
|
"entropy": 6.042965602874756,
|
|
"epoch": 0.02502597562201839,
|
|
"grad_norm": 1.609375,
|
|
"learning_rate": 0.0004982552163405623,
|
|
"loss": 5.8914,
|
|
"mean_token_accuracy": 0.17837537676095963,
|
|
"num_tokens": 949758.0,
|
|
"step": 1120
|
|
},
|
|
{
|
|
"entropy": 5.911374568939209,
|
|
"epoch": 0.025137698727473828,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.0004981057259983839,
|
|
"loss": 6.046,
|
|
"mean_token_accuracy": 0.17550223916769028,
|
|
"num_tokens": 954028.0,
|
|
"step": 1125
|
|
},
|
|
{
|
|
"entropy": 6.2100794315338135,
|
|
"epoch": 0.02524942183292927,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 0.0004979501191000262,
|
|
"loss": 6.0439,
|
|
"mean_token_accuracy": 0.1597377523779869,
|
|
"num_tokens": 958006.0,
|
|
"step": 1130
|
|
},
|
|
{
|
|
"entropy": 5.997774934768676,
|
|
"epoch": 0.02536114493838471,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 0.0004977883999115311,
|
|
"loss": 5.9673,
|
|
"mean_token_accuracy": 0.1754068598151207,
|
|
"num_tokens": 962637.0,
|
|
"step": 1135
|
|
},
|
|
{
|
|
"entropy": 6.133125972747803,
|
|
"epoch": 0.025472868043840146,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.0004976205728665113,
|
|
"loss": 5.9539,
|
|
"mean_token_accuracy": 0.17636077851057053,
|
|
"num_tokens": 966770.0,
|
|
"step": 1140
|
|
},
|
|
{
|
|
"entropy": 5.795867204666138,
|
|
"epoch": 0.025584591149295587,
|
|
"grad_norm": 1.5859375,
|
|
"learning_rate": 0.0004974466425660307,
|
|
"loss": 5.9624,
|
|
"mean_token_accuracy": 0.17465363889932634,
|
|
"num_tokens": 970976.0,
|
|
"step": 1145
|
|
},
|
|
{
|
|
"entropy": 6.01137866973877,
|
|
"epoch": 0.025696314254751024,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 0.0004972666137784759,
|
|
"loss": 6.0146,
|
|
"mean_token_accuracy": 0.16153744012117385,
|
|
"num_tokens": 975253.0,
|
|
"step": 1150
|
|
},
|
|
{
|
|
"entropy": 6.057316732406616,
|
|
"epoch": 0.025808037360206464,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.0004970804914394271,
|
|
"loss": 5.9926,
|
|
"mean_token_accuracy": 0.16764254570007325,
|
|
"num_tokens": 979388.0,
|
|
"step": 1155
|
|
},
|
|
{
|
|
"entropy": 5.929347705841065,
|
|
"epoch": 0.025919760465661905,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 0.0004968882806515225,
|
|
"loss": 5.9724,
|
|
"mean_token_accuracy": 0.1787356838583946,
|
|
"num_tokens": 983216.0,
|
|
"step": 1160
|
|
},
|
|
{
|
|
"entropy": 6.045448160171508,
|
|
"epoch": 0.026031483571117342,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.0004966899866843177,
|
|
"loss": 6.0631,
|
|
"mean_token_accuracy": 0.16194660663604737,
|
|
"num_tokens": 987850.0,
|
|
"step": 1165
|
|
},
|
|
{
|
|
"entropy": 6.053802442550659,
|
|
"epoch": 0.026143206676572783,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.000496485614974142,
|
|
"loss": 6.0138,
|
|
"mean_token_accuracy": 0.16602955460548402,
|
|
"num_tokens": 992232.0,
|
|
"step": 1170
|
|
},
|
|
{
|
|
"entropy": 6.0259240627288815,
|
|
"epoch": 0.02625492978202822,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.0004962751711239492,
|
|
"loss": 6.0693,
|
|
"mean_token_accuracy": 0.16301065385341645,
|
|
"num_tokens": 996428.0,
|
|
"step": 1175
|
|
},
|
|
{
|
|
"entropy": 6.157499313354492,
|
|
"epoch": 0.02636665288748366,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.0004960586609031636,
|
|
"loss": 5.9524,
|
|
"mean_token_accuracy": 0.17110942751169206,
|
|
"num_tokens": 1000336.0,
|
|
"step": 1180
|
|
},
|
|
{
|
|
"entropy": 6.04528169631958,
|
|
"epoch": 0.0264783759929391,
|
|
"grad_norm": 1.6328125,
|
|
"learning_rate": 0.0004958360902475224,
|
|
"loss": 5.9189,
|
|
"mean_token_accuracy": 0.16904981136322023,
|
|
"num_tokens": 1004312.0,
|
|
"step": 1185
|
|
},
|
|
{
|
|
"entropy": 6.089739894866943,
|
|
"epoch": 0.026590099098394538,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.0004956074652589125,
|
|
"loss": 6.174,
|
|
"mean_token_accuracy": 0.15594123005867006,
|
|
"num_tokens": 1008728.0,
|
|
"step": 1190
|
|
},
|
|
{
|
|
"entropy": 6.1193163871765135,
|
|
"epoch": 0.02670182220384998,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.0004953727922052035,
|
|
"loss": 6.0671,
|
|
"mean_token_accuracy": 0.16603883504867553,
|
|
"num_tokens": 1013385.0,
|
|
"step": 1195
|
|
},
|
|
{
|
|
"entropy": 6.139393329620361,
|
|
"epoch": 0.026813545309305416,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.0004951320775200756,
|
|
"loss": 6.0219,
|
|
"mean_token_accuracy": 0.16773877292871475,
|
|
"num_tokens": 1017537.0,
|
|
"step": 1200
|
|
},
|
|
{
|
|
"entropy": 5.894366598129272,
|
|
"epoch": 0.026925268414760856,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.0004948853278028436,
|
|
"loss": 5.9409,
|
|
"mean_token_accuracy": 0.17285942137241364,
|
|
"num_tokens": 1021753.0,
|
|
"step": 1205
|
|
},
|
|
{
|
|
"entropy": 5.971589708328247,
|
|
"epoch": 0.027036991520216297,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.0004946325498182755,
|
|
"loss": 6.0589,
|
|
"mean_token_accuracy": 0.1600156843662262,
|
|
"num_tokens": 1026248.0,
|
|
"step": 1210
|
|
},
|
|
{
|
|
"entropy": 5.97881760597229,
|
|
"epoch": 0.027148714625671734,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.0004943737504964076,
|
|
"loss": 5.8084,
|
|
"mean_token_accuracy": 0.18238039016723634,
|
|
"num_tokens": 1030284.0,
|
|
"step": 1215
|
|
},
|
|
{
|
|
"entropy": 6.004219436645508,
|
|
"epoch": 0.027260437731127175,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.000494108936932354,
|
|
"loss": 6.1125,
|
|
"mean_token_accuracy": 0.1722887262701988,
|
|
"num_tokens": 1034444.0,
|
|
"step": 1220
|
|
},
|
|
{
|
|
"entropy": 6.072319364547729,
|
|
"epoch": 0.027372160836582615,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.0004938381163861124,
|
|
"loss": 5.9253,
|
|
"mean_token_accuracy": 0.18728673458099365,
|
|
"num_tokens": 1038819.0,
|
|
"step": 1225
|
|
},
|
|
{
|
|
"entropy": 6.105956125259399,
|
|
"epoch": 0.027483883942038052,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0004935612962823645,
|
|
"loss": 6.1433,
|
|
"mean_token_accuracy": 0.1664537325501442,
|
|
"num_tokens": 1043303.0,
|
|
"step": 1230
|
|
},
|
|
{
|
|
"entropy": 5.987173557281494,
|
|
"epoch": 0.027595607047493493,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 0.0004932784842102739,
|
|
"loss": 5.9435,
|
|
"mean_token_accuracy": 0.1737391769886017,
|
|
"num_tokens": 1047451.0,
|
|
"step": 1235
|
|
},
|
|
{
|
|
"entropy": 6.031593132019043,
|
|
"epoch": 0.02770733015294893,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.0004929896879232758,
|
|
"loss": 5.9946,
|
|
"mean_token_accuracy": 0.1745271012187004,
|
|
"num_tokens": 1052370.0,
|
|
"step": 1240
|
|
},
|
|
{
|
|
"entropy": 6.133889484405517,
|
|
"epoch": 0.02781905325840437,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.0004926949153388668,
|
|
"loss": 6.1184,
|
|
"mean_token_accuracy": 0.15884814709424971,
|
|
"num_tokens": 1056755.0,
|
|
"step": 1245
|
|
},
|
|
{
|
|
"entropy": 6.027875137329102,
|
|
"epoch": 0.02793077636385981,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0004923941745383859,
|
|
"loss": 6.0237,
|
|
"mean_token_accuracy": 0.1649158090353012,
|
|
"num_tokens": 1061380.0,
|
|
"step": 1250
|
|
},
|
|
{
|
|
"entropy": 5.859837007522583,
|
|
"epoch": 0.02804249946931525,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.000492087473766794,
|
|
"loss": 5.9049,
|
|
"mean_token_accuracy": 0.1740475058555603,
|
|
"num_tokens": 1065633.0,
|
|
"step": 1255
|
|
},
|
|
{
|
|
"entropy": 5.939010953903198,
|
|
"epoch": 0.02815422257477069,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 0.000491774821432448,
|
|
"loss": 6.043,
|
|
"mean_token_accuracy": 0.1675959825515747,
|
|
"num_tokens": 1070131.0,
|
|
"step": 1260
|
|
},
|
|
{
|
|
"entropy": 6.035466241836548,
|
|
"epoch": 0.028265945680226126,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.0004914562261068693,
|
|
"loss": 5.956,
|
|
"mean_token_accuracy": 0.17018137276172637,
|
|
"num_tokens": 1074465.0,
|
|
"step": 1265
|
|
},
|
|
{
|
|
"entropy": 5.894679594039917,
|
|
"epoch": 0.028377668785681567,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 0.0004911316965245098,
|
|
"loss": 5.8307,
|
|
"mean_token_accuracy": 0.17515944987535476,
|
|
"num_tokens": 1078738.0,
|
|
"step": 1270
|
|
},
|
|
{
|
|
"entropy": 5.887772560119629,
|
|
"epoch": 0.028489391891137007,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.000490801241582512,
|
|
"loss": 5.9601,
|
|
"mean_token_accuracy": 0.1752310276031494,
|
|
"num_tokens": 1082529.0,
|
|
"step": 1275
|
|
},
|
|
{
|
|
"entropy": 6.000855207443237,
|
|
"epoch": 0.028601114996592444,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.000490464870340465,
|
|
"loss": 6.0392,
|
|
"mean_token_accuracy": 0.16873422712087632,
|
|
"num_tokens": 1087043.0,
|
|
"step": 1280
|
|
},
|
|
{
|
|
"entropy": 5.8165771007537845,
|
|
"epoch": 0.028712838102047885,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 0.0004901225920201563,
|
|
"loss": 5.9401,
|
|
"mean_token_accuracy": 0.18170406371355058,
|
|
"num_tokens": 1091045.0,
|
|
"step": 1285
|
|
},
|
|
{
|
|
"entropy": 6.046920347213745,
|
|
"epoch": 0.028824561207503322,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 0.000489774416005319,
|
|
"loss": 5.911,
|
|
"mean_token_accuracy": 0.17549529373645784,
|
|
"num_tokens": 1094949.0,
|
|
"step": 1290
|
|
},
|
|
{
|
|
"entropy": 5.849216651916504,
|
|
"epoch": 0.028936284312958763,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 0.0004894203518413742,
|
|
"loss": 5.9407,
|
|
"mean_token_accuracy": 0.1647828571498394,
|
|
"num_tokens": 1099131.0,
|
|
"step": 1295
|
|
},
|
|
{
|
|
"entropy": 5.80899658203125,
|
|
"epoch": 0.029048007418414203,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.0004890604092351701,
|
|
"loss": 5.8729,
|
|
"mean_token_accuracy": 0.17930638641119004,
|
|
"num_tokens": 1102967.0,
|
|
"step": 1300
|
|
},
|
|
{
|
|
"entropy": 5.98758225440979,
|
|
"epoch": 0.02915973052386964,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.000488694598054715,
|
|
"loss": 5.9128,
|
|
"mean_token_accuracy": 0.17649397403001785,
|
|
"num_tokens": 1106777.0,
|
|
"step": 1305
|
|
},
|
|
{
|
|
"entropy": 5.834689521789551,
|
|
"epoch": 0.02927145362932508,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.0004883229283289071,
|
|
"loss": 5.9249,
|
|
"mean_token_accuracy": 0.1766020655632019,
|
|
"num_tokens": 1111184.0,
|
|
"step": 1310
|
|
},
|
|
{
|
|
"entropy": 6.064471244812012,
|
|
"epoch": 0.02938317673478052,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.00048794541024725993,
|
|
"loss": 5.9235,
|
|
"mean_token_accuracy": 0.17901833802461625,
|
|
"num_tokens": 1115651.0,
|
|
"step": 1315
|
|
},
|
|
{
|
|
"entropy": 6.012405014038086,
|
|
"epoch": 0.02949489984023596,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.0004875620541596221,
|
|
"loss": 5.956,
|
|
"mean_token_accuracy": 0.1633922666311264,
|
|
"num_tokens": 1120012.0,
|
|
"step": 1320
|
|
},
|
|
{
|
|
"entropy": 5.800662183761597,
|
|
"epoch": 0.0296066229456914,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.00048717287057589454,
|
|
"loss": 5.859,
|
|
"mean_token_accuracy": 0.18077441900968552,
|
|
"num_tokens": 1124549.0,
|
|
"step": 1325
|
|
},
|
|
{
|
|
"entropy": 5.966872882843018,
|
|
"epoch": 0.029718346051146836,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.0004867778701657417,
|
|
"loss": 5.9441,
|
|
"mean_token_accuracy": 0.17887909561395646,
|
|
"num_tokens": 1128968.0,
|
|
"step": 1330
|
|
},
|
|
{
|
|
"entropy": 5.908768558502198,
|
|
"epoch": 0.029830069156602277,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00048637706375829955,
|
|
"loss": 6.0053,
|
|
"mean_token_accuracy": 0.17145692855119704,
|
|
"num_tokens": 1133396.0,
|
|
"step": 1335
|
|
},
|
|
{
|
|
"entropy": 6.086913299560547,
|
|
"epoch": 0.029941792262057718,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.000485970462341878,
|
|
"loss": 5.9048,
|
|
"mean_token_accuracy": 0.17118549942970276,
|
|
"num_tokens": 1137577.0,
|
|
"step": 1340
|
|
},
|
|
{
|
|
"entropy": 5.773934030532837,
|
|
"epoch": 0.030053515367513155,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.00048555807706366044,
|
|
"loss": 5.8044,
|
|
"mean_token_accuracy": 0.17675584256649018,
|
|
"num_tokens": 1141551.0,
|
|
"step": 1345
|
|
},
|
|
{
|
|
"entropy": 5.967331218719482,
|
|
"epoch": 0.030165238472968595,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 0.00048513991922939756,
|
|
"loss": 5.8297,
|
|
"mean_token_accuracy": 0.17623057961463928,
|
|
"num_tokens": 1145295.0,
|
|
"step": 1350
|
|
},
|
|
{
|
|
"entropy": 5.916956901550293,
|
|
"epoch": 0.030276961578424032,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.00048471600030309744,
|
|
"loss": 6.0757,
|
|
"mean_token_accuracy": 0.16089194118976594,
|
|
"num_tokens": 1149946.0,
|
|
"step": 1355
|
|
},
|
|
{
|
|
"entropy": 5.946835136413574,
|
|
"epoch": 0.030388684683879473,
|
|
"grad_norm": 1.625,
|
|
"learning_rate": 0.00048428633190671186,
|
|
"loss": 5.9579,
|
|
"mean_token_accuracy": 0.17933496832847595,
|
|
"num_tokens": 1153955.0,
|
|
"step": 1360
|
|
},
|
|
{
|
|
"entropy": 5.879767847061157,
|
|
"epoch": 0.030500407789334914,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.0004838509258198167,
|
|
"loss": 5.8494,
|
|
"mean_token_accuracy": 0.17823615819215774,
|
|
"num_tokens": 1158568.0,
|
|
"step": 1365
|
|
},
|
|
{
|
|
"entropy": 5.8853926181793215,
|
|
"epoch": 0.03061213089479035,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 0.00048340979397929,
|
|
"loss": 5.8017,
|
|
"mean_token_accuracy": 0.17762538492679597,
|
|
"num_tokens": 1162217.0,
|
|
"step": 1370
|
|
},
|
|
{
|
|
"entropy": 5.819405508041382,
|
|
"epoch": 0.03072385400024579,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.00048296294847898386,
|
|
"loss": 5.8786,
|
|
"mean_token_accuracy": 0.17288521826267242,
|
|
"num_tokens": 1165905.0,
|
|
"step": 1375
|
|
},
|
|
{
|
|
"entropy": 5.949618673324585,
|
|
"epoch": 0.03083557710570123,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 0.0004825104015693934,
|
|
"loss": 5.8634,
|
|
"mean_token_accuracy": 0.1736384004354477,
|
|
"num_tokens": 1170352.0,
|
|
"step": 1380
|
|
},
|
|
{
|
|
"entropy": 5.8261597633361815,
|
|
"epoch": 0.03094730021115667,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.0004820521656573208,
|
|
"loss": 5.9311,
|
|
"mean_token_accuracy": 0.17523645013570785,
|
|
"num_tokens": 1174560.0,
|
|
"step": 1385
|
|
},
|
|
{
|
|
"entropy": 5.963821220397949,
|
|
"epoch": 0.03105902331661211,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.00048158825330553505,
|
|
"loss": 5.8748,
|
|
"mean_token_accuracy": 0.1740915596485138,
|
|
"num_tokens": 1178544.0,
|
|
"step": 1390
|
|
},
|
|
{
|
|
"entropy": 5.702950811386108,
|
|
"epoch": 0.031170746422067547,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 0.00048111867723242763,
|
|
"loss": 5.7173,
|
|
"mean_token_accuracy": 0.19122548252344132,
|
|
"num_tokens": 1182541.0,
|
|
"step": 1395
|
|
},
|
|
{
|
|
"entropy": 5.825392246246338,
|
|
"epoch": 0.031282469527522984,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0004806434503116637,
|
|
"loss": 5.803,
|
|
"mean_token_accuracy": 0.18207484632730483,
|
|
"num_tokens": 1186684.0,
|
|
"step": 1400
|
|
},
|
|
{
|
|
"entropy": 5.852643585205078,
|
|
"epoch": 0.03139419263297843,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.0004801625855718296,
|
|
"loss": 5.7638,
|
|
"mean_token_accuracy": 0.1846204161643982,
|
|
"num_tokens": 1190785.0,
|
|
"step": 1405
|
|
},
|
|
{
|
|
"entropy": 5.6974410057067875,
|
|
"epoch": 0.031505915738433865,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00047967609619607477,
|
|
"loss": 5.7808,
|
|
"mean_token_accuracy": 0.17526547610759735,
|
|
"num_tokens": 1194940.0,
|
|
"step": 1410
|
|
},
|
|
{
|
|
"entropy": 5.803288984298706,
|
|
"epoch": 0.0316176388438893,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.0004791839955217513,
|
|
"loss": 5.7563,
|
|
"mean_token_accuracy": 0.18818183541297911,
|
|
"num_tokens": 1198765.0,
|
|
"step": 1415
|
|
},
|
|
{
|
|
"entropy": 5.929622077941895,
|
|
"epoch": 0.031729361949344746,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.00047868629704004786,
|
|
"loss": 5.979,
|
|
"mean_token_accuracy": 0.1642720431089401,
|
|
"num_tokens": 1203266.0,
|
|
"step": 1420
|
|
},
|
|
{
|
|
"entropy": 5.848547840118409,
|
|
"epoch": 0.03184108505480018,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.00047818301439561965,
|
|
"loss": 5.7561,
|
|
"mean_token_accuracy": 0.1881594181060791,
|
|
"num_tokens": 1207414.0,
|
|
"step": 1425
|
|
},
|
|
{
|
|
"entropy": 5.699222707748413,
|
|
"epoch": 0.03195280816025562,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.00047767416138621454,
|
|
"loss": 5.7656,
|
|
"mean_token_accuracy": 0.18503025323152542,
|
|
"num_tokens": 1211325.0,
|
|
"step": 1430
|
|
},
|
|
{
|
|
"entropy": 5.976352977752685,
|
|
"epoch": 0.032064531265711065,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.000477159751962295,
|
|
"loss": 5.801,
|
|
"mean_token_accuracy": 0.18062487095594407,
|
|
"num_tokens": 1215866.0,
|
|
"step": 1435
|
|
},
|
|
{
|
|
"entropy": 5.736691808700561,
|
|
"epoch": 0.0321762543711665,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.00047663980022665507,
|
|
"loss": 5.861,
|
|
"mean_token_accuracy": 0.17032930999994278,
|
|
"num_tokens": 1220241.0,
|
|
"step": 1440
|
|
},
|
|
{
|
|
"entropy": 5.691896915435791,
|
|
"epoch": 0.03228797747662194,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 0.00047611432043403437,
|
|
"loss": 5.8819,
|
|
"mean_token_accuracy": 0.1773227259516716,
|
|
"num_tokens": 1224330.0,
|
|
"step": 1445
|
|
},
|
|
{
|
|
"entropy": 6.069937038421631,
|
|
"epoch": 0.03239970058207738,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.0004755833269907267,
|
|
"loss": 5.9618,
|
|
"mean_token_accuracy": 0.16885158866643907,
|
|
"num_tokens": 1228518.0,
|
|
"step": 1450
|
|
},
|
|
{
|
|
"entropy": 5.808598423004151,
|
|
"epoch": 0.03251142368753282,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.0004750468344541857,
|
|
"loss": 5.8203,
|
|
"mean_token_accuracy": 0.18446521908044816,
|
|
"num_tokens": 1232556.0,
|
|
"step": 1455
|
|
},
|
|
{
|
|
"entropy": 5.866223096847534,
|
|
"epoch": 0.03262314679298826,
|
|
"grad_norm": 1.6015625,
|
|
"learning_rate": 0.00047450485753262525,
|
|
"loss": 5.8594,
|
|
"mean_token_accuracy": 0.1794953316450119,
|
|
"num_tokens": 1236464.0,
|
|
"step": 1460
|
|
},
|
|
{
|
|
"entropy": 5.9548783779144285,
|
|
"epoch": 0.032734869898443694,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.00047395741108461633,
|
|
"loss": 5.8455,
|
|
"mean_token_accuracy": 0.18512817621231079,
|
|
"num_tokens": 1240478.0,
|
|
"step": 1465
|
|
},
|
|
{
|
|
"entropy": 5.9534708023071286,
|
|
"epoch": 0.03284659300389914,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.00047340451011867985,
|
|
"loss": 5.906,
|
|
"mean_token_accuracy": 0.1799629181623459,
|
|
"num_tokens": 1244149.0,
|
|
"step": 1470
|
|
},
|
|
{
|
|
"entropy": 6.017633962631225,
|
|
"epoch": 0.032958316109354575,
|
|
"grad_norm": 1.671875,
|
|
"learning_rate": 0.00047284616979287515,
|
|
"loss": 5.9879,
|
|
"mean_token_accuracy": 0.16287869438529015,
|
|
"num_tokens": 1248617.0,
|
|
"step": 1475
|
|
},
|
|
{
|
|
"entropy": 5.903649234771729,
|
|
"epoch": 0.03307003921481001,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.00047228240541438433,
|
|
"loss": 5.8795,
|
|
"mean_token_accuracy": 0.1832698553800583,
|
|
"num_tokens": 1253044.0,
|
|
"step": 1480
|
|
},
|
|
{
|
|
"entropy": 5.926074123382568,
|
|
"epoch": 0.03318176232026546,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.00047171323243909257,
|
|
"loss": 5.9143,
|
|
"mean_token_accuracy": 0.1814977541565895,
|
|
"num_tokens": 1257651.0,
|
|
"step": 1485
|
|
},
|
|
{
|
|
"entropy": 5.833726167678833,
|
|
"epoch": 0.033293485425720894,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.00047113866647116457,
|
|
"loss": 5.8368,
|
|
"mean_token_accuracy": 0.18549390286207199,
|
|
"num_tokens": 1261805.0,
|
|
"step": 1490
|
|
},
|
|
{
|
|
"entropy": 5.733483600616455,
|
|
"epoch": 0.03340520853117633,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.0004705587232626164,
|
|
"loss": 5.6356,
|
|
"mean_token_accuracy": 0.19865250885486602,
|
|
"num_tokens": 1265625.0,
|
|
"step": 1495
|
|
},
|
|
{
|
|
"entropy": 5.693035125732422,
|
|
"epoch": 0.033516931636631775,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.00046997341871288424,
|
|
"loss": 5.8133,
|
|
"mean_token_accuracy": 0.18278113082051278,
|
|
"num_tokens": 1269876.0,
|
|
"step": 1500
|
|
},
|
|
{
|
|
"entropy": 5.831105041503906,
|
|
"epoch": 0.03362865474208721,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.0004693827688683879,
|
|
"loss": 5.7775,
|
|
"mean_token_accuracy": 0.18592336624860764,
|
|
"num_tokens": 1273821.0,
|
|
"step": 1505
|
|
},
|
|
{
|
|
"entropy": 5.830736303329468,
|
|
"epoch": 0.03374037784754265,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.0004687867899220914,
|
|
"loss": 5.8811,
|
|
"mean_token_accuracy": 0.18795700818300248,
|
|
"num_tokens": 1278574.0,
|
|
"step": 1510
|
|
},
|
|
{
|
|
"entropy": 5.784781217575073,
|
|
"epoch": 0.033852100952998086,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.00046818549821305846,
|
|
"loss": 5.8642,
|
|
"mean_token_accuracy": 0.17242854535579683,
|
|
"num_tokens": 1283271.0,
|
|
"step": 1515
|
|
},
|
|
{
|
|
"entropy": 5.806311082839966,
|
|
"epoch": 0.03396382405845353,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.00046757891022600494,
|
|
"loss": 5.6992,
|
|
"mean_token_accuracy": 0.1895821213722229,
|
|
"num_tokens": 1287633.0,
|
|
"step": 1520
|
|
},
|
|
{
|
|
"entropy": 5.612684059143066,
|
|
"epoch": 0.03407554716390897,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 0.0004669670425908471,
|
|
"loss": 5.7039,
|
|
"mean_token_accuracy": 0.19535314291715622,
|
|
"num_tokens": 1291796.0,
|
|
"step": 1525
|
|
},
|
|
{
|
|
"entropy": 5.7842505931854244,
|
|
"epoch": 0.034187270269364405,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.0004663499120822451,
|
|
"loss": 5.8562,
|
|
"mean_token_accuracy": 0.18286515921354293,
|
|
"num_tokens": 1296086.0,
|
|
"step": 1530
|
|
},
|
|
{
|
|
"entropy": 5.824511098861694,
|
|
"epoch": 0.03429899337481985,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.0004657275356191437,
|
|
"loss": 5.8006,
|
|
"mean_token_accuracy": 0.17361746430397035,
|
|
"num_tokens": 1300651.0,
|
|
"step": 1535
|
|
},
|
|
{
|
|
"entropy": 5.797060441970825,
|
|
"epoch": 0.034410716480275286,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 0.00046509993026430804,
|
|
"loss": 5.7873,
|
|
"mean_token_accuracy": 0.18226577788591386,
|
|
"num_tokens": 1304930.0,
|
|
"step": 1540
|
|
},
|
|
{
|
|
"entropy": 5.883905744552612,
|
|
"epoch": 0.03452243958573072,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.0004644671132238558,
|
|
"loss": 5.9211,
|
|
"mean_token_accuracy": 0.16726794093847275,
|
|
"num_tokens": 1308950.0,
|
|
"step": 1545
|
|
},
|
|
{
|
|
"entropy": 5.8268091678619385,
|
|
"epoch": 0.03463416269118617,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 0.00046382910184678585,
|
|
"loss": 5.7787,
|
|
"mean_token_accuracy": 0.17821203917264938,
|
|
"num_tokens": 1313059.0,
|
|
"step": 1550
|
|
},
|
|
{
|
|
"entropy": 5.647272109985352,
|
|
"epoch": 0.034745885796641604,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.0004631859136245025,
|
|
"loss": 5.6842,
|
|
"mean_token_accuracy": 0.19873663783073425,
|
|
"num_tokens": 1316756.0,
|
|
"step": 1555
|
|
},
|
|
{
|
|
"entropy": 5.785255432128906,
|
|
"epoch": 0.03485760890209704,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.0004625375661903357,
|
|
"loss": 5.9699,
|
|
"mean_token_accuracy": 0.1627327397465706,
|
|
"num_tokens": 1320834.0,
|
|
"step": 1560
|
|
},
|
|
{
|
|
"entropy": 5.799972629547119,
|
|
"epoch": 0.034969332007552485,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.00046188407731905787,
|
|
"loss": 5.747,
|
|
"mean_token_accuracy": 0.17980815470218658,
|
|
"num_tokens": 1324484.0,
|
|
"step": 1565
|
|
},
|
|
{
|
|
"entropy": 5.779381561279297,
|
|
"epoch": 0.03508105511300792,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.00046122546492639643,
|
|
"loss": 5.7251,
|
|
"mean_token_accuracy": 0.18375469446182252,
|
|
"num_tokens": 1328935.0,
|
|
"step": 1570
|
|
},
|
|
{
|
|
"entropy": 5.741471004486084,
|
|
"epoch": 0.03519277821846336,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 0.000460561747068543,
|
|
"loss": 5.7432,
|
|
"mean_token_accuracy": 0.1859663426876068,
|
|
"num_tokens": 1333225.0,
|
|
"step": 1575
|
|
},
|
|
{
|
|
"entropy": 5.714008522033692,
|
|
"epoch": 0.0353045013239188,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 0.0004598929419416578,
|
|
"loss": 5.6636,
|
|
"mean_token_accuracy": 0.18822408169507981,
|
|
"num_tokens": 1337234.0,
|
|
"step": 1580
|
|
},
|
|
{
|
|
"entropy": 5.844133186340332,
|
|
"epoch": 0.03541622442937424,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.00045921906788137123,
|
|
"loss": 5.9233,
|
|
"mean_token_accuracy": 0.1749085620045662,
|
|
"num_tokens": 1341765.0,
|
|
"step": 1585
|
|
},
|
|
{
|
|
"entropy": 5.890516853332519,
|
|
"epoch": 0.03552794753482968,
|
|
"grad_norm": 1.5234375,
|
|
"learning_rate": 0.00045854014336228115,
|
|
"loss": 5.8023,
|
|
"mean_token_accuracy": 0.17829006016254426,
|
|
"num_tokens": 1346087.0,
|
|
"step": 1590
|
|
},
|
|
{
|
|
"entropy": 5.80145001411438,
|
|
"epoch": 0.035639670640285115,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.00045785618699744615,
|
|
"loss": 5.8247,
|
|
"mean_token_accuracy": 0.1754228189587593,
|
|
"num_tokens": 1350443.0,
|
|
"step": 1595
|
|
},
|
|
{
|
|
"entropy": 5.880692100524902,
|
|
"epoch": 0.03575139374574056,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.00045716721753787543,
|
|
"loss": 5.8544,
|
|
"mean_token_accuracy": 0.1789305865764618,
|
|
"num_tokens": 1355092.0,
|
|
"step": 1600
|
|
},
|
|
{
|
|
"entropy": 5.775990533828735,
|
|
"epoch": 0.035863116851195996,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.0004564732538720148,
|
|
"loss": 5.8375,
|
|
"mean_token_accuracy": 0.18685502409934998,
|
|
"num_tokens": 1359379.0,
|
|
"step": 1605
|
|
},
|
|
{
|
|
"entropy": 5.826099395751953,
|
|
"epoch": 0.03597483995665143,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.00045577431502522877,
|
|
"loss": 5.6924,
|
|
"mean_token_accuracy": 0.18221372812986375,
|
|
"num_tokens": 1363346.0,
|
|
"step": 1610
|
|
},
|
|
{
|
|
"entropy": 5.730778932571411,
|
|
"epoch": 0.03608656306210688,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.0004550704201592787,
|
|
"loss": 5.773,
|
|
"mean_token_accuracy": 0.17810402810573578,
|
|
"num_tokens": 1367322.0,
|
|
"step": 1615
|
|
},
|
|
{
|
|
"entropy": 5.649629878997803,
|
|
"epoch": 0.036198286167562314,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.0004543615885717981,
|
|
"loss": 5.7162,
|
|
"mean_token_accuracy": 0.1771922379732132,
|
|
"num_tokens": 1371165.0,
|
|
"step": 1620
|
|
},
|
|
{
|
|
"entropy": 5.648497486114502,
|
|
"epoch": 0.03631000927301775,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.00045364783969576296,
|
|
"loss": 5.7376,
|
|
"mean_token_accuracy": 0.1936981901526451,
|
|
"num_tokens": 1375119.0,
|
|
"step": 1625
|
|
},
|
|
{
|
|
"entropy": 5.806000185012818,
|
|
"epoch": 0.03642173237847319,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.0004529291930989592,
|
|
"loss": 5.6657,
|
|
"mean_token_accuracy": 0.18797771632671356,
|
|
"num_tokens": 1379268.0,
|
|
"step": 1630
|
|
},
|
|
{
|
|
"entropy": 5.663884687423706,
|
|
"epoch": 0.03653345548392863,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.0004522056684834464,
|
|
"loss": 5.785,
|
|
"mean_token_accuracy": 0.16971432119607927,
|
|
"num_tokens": 1383477.0,
|
|
"step": 1635
|
|
},
|
|
{
|
|
"entropy": 5.687128686904908,
|
|
"epoch": 0.03664517858938407,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.0004514772856850173,
|
|
"loss": 5.7087,
|
|
"mean_token_accuracy": 0.1861000046133995,
|
|
"num_tokens": 1387138.0,
|
|
"step": 1640
|
|
},
|
|
{
|
|
"entropy": 5.865495443344116,
|
|
"epoch": 0.03675690169483951,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.0004507440646726542,
|
|
"loss": 5.835,
|
|
"mean_token_accuracy": 0.16652033776044844,
|
|
"num_tokens": 1391758.0,
|
|
"step": 1645
|
|
},
|
|
{
|
|
"entropy": 5.652455759048462,
|
|
"epoch": 0.03686862480029495,
|
|
"grad_norm": 1.484375,
|
|
"learning_rate": 0.0004500060255479818,
|
|
"loss": 5.7357,
|
|
"mean_token_accuracy": 0.17026722058653831,
|
|
"num_tokens": 1395517.0,
|
|
"step": 1650
|
|
},
|
|
{
|
|
"entropy": 5.77337498664856,
|
|
"epoch": 0.03698034790575039,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.0004492631885447151,
|
|
"loss": 5.7572,
|
|
"mean_token_accuracy": 0.1879993349313736,
|
|
"num_tokens": 1400021.0,
|
|
"step": 1655
|
|
},
|
|
{
|
|
"entropy": 5.89671516418457,
|
|
"epoch": 0.037092071011205825,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.00044851557402810616,
|
|
"loss": 5.918,
|
|
"mean_token_accuracy": 0.17465589046478272,
|
|
"num_tokens": 1404456.0,
|
|
"step": 1660
|
|
},
|
|
{
|
|
"entropy": 5.753234243392944,
|
|
"epoch": 0.03720379411666127,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.00044776320249438444,
|
|
"loss": 5.8144,
|
|
"mean_token_accuracy": 0.17828851491212844,
|
|
"num_tokens": 1408612.0,
|
|
"step": 1665
|
|
},
|
|
{
|
|
"entropy": 5.80224084854126,
|
|
"epoch": 0.037315517222116706,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.00044700609457019565,
|
|
"loss": 5.7924,
|
|
"mean_token_accuracy": 0.18226661533117294,
|
|
"num_tokens": 1412688.0,
|
|
"step": 1670
|
|
},
|
|
{
|
|
"entropy": 5.919682836532592,
|
|
"epoch": 0.037427240327572144,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.0004462442710120359,
|
|
"loss": 5.8297,
|
|
"mean_token_accuracy": 0.17643865197896957,
|
|
"num_tokens": 1417273.0,
|
|
"step": 1675
|
|
},
|
|
{
|
|
"entropy": 5.83403468132019,
|
|
"epoch": 0.03753896343302759,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.000445477752705683,
|
|
"loss": 5.7559,
|
|
"mean_token_accuracy": 0.17946113049983978,
|
|
"num_tokens": 1421737.0,
|
|
"step": 1680
|
|
},
|
|
{
|
|
"entropy": 5.653639125823974,
|
|
"epoch": 0.037650686538483025,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00044470656066562336,
|
|
"loss": 5.528,
|
|
"mean_token_accuracy": 0.20140709578990937,
|
|
"num_tokens": 1426154.0,
|
|
"step": 1685
|
|
},
|
|
{
|
|
"entropy": 5.726971578598023,
|
|
"epoch": 0.03776240964393846,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.0004439307160344765,
|
|
"loss": 5.7815,
|
|
"mean_token_accuracy": 0.17367986142635344,
|
|
"num_tokens": 1430270.0,
|
|
"step": 1690
|
|
},
|
|
{
|
|
"entropy": 5.809293508529663,
|
|
"epoch": 0.0378741327493939,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.00044315024008241473,
|
|
"loss": 5.6611,
|
|
"mean_token_accuracy": 0.1972547873854637,
|
|
"num_tokens": 1434877.0,
|
|
"step": 1695
|
|
},
|
|
{
|
|
"entropy": 5.65094747543335,
|
|
"epoch": 0.03798585585484934,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.0004423651542065806,
|
|
"loss": 5.7318,
|
|
"mean_token_accuracy": 0.19258931577205657,
|
|
"num_tokens": 1438690.0,
|
|
"step": 1700
|
|
},
|
|
{
|
|
"entropy": 5.721880674362183,
|
|
"epoch": 0.03809757896030478,
|
|
"grad_norm": 1.53125,
|
|
"learning_rate": 0.00044157547993050006,
|
|
"loss": 5.6646,
|
|
"mean_token_accuracy": 0.19387782514095306,
|
|
"num_tokens": 1443085.0,
|
|
"step": 1705
|
|
},
|
|
{
|
|
"entropy": 5.580819368362427,
|
|
"epoch": 0.03820930206576022,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 0.00044078123890349227,
|
|
"loss": 5.5564,
|
|
"mean_token_accuracy": 0.20745181739330293,
|
|
"num_tokens": 1447280.0,
|
|
"step": 1710
|
|
},
|
|
{
|
|
"entropy": 5.709810638427735,
|
|
"epoch": 0.03832102517121566,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.00043998245290007606,
|
|
"loss": 5.7408,
|
|
"mean_token_accuracy": 0.1841261014342308,
|
|
"num_tokens": 1451639.0,
|
|
"step": 1715
|
|
},
|
|
{
|
|
"entropy": 5.823773813247681,
|
|
"epoch": 0.0384327482766711,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.00043917914381937323,
|
|
"loss": 5.7155,
|
|
"mean_token_accuracy": 0.18863430619239807,
|
|
"num_tokens": 1455473.0,
|
|
"step": 1720
|
|
},
|
|
{
|
|
"entropy": 5.738099813461304,
|
|
"epoch": 0.038544471382126536,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.00043837133368450815,
|
|
"loss": 5.7901,
|
|
"mean_token_accuracy": 0.1780441515147686,
|
|
"num_tokens": 1460332.0,
|
|
"step": 1725
|
|
},
|
|
{
|
|
"entropy": 5.72046480178833,
|
|
"epoch": 0.03865619448758198,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.0004375590446420037,
|
|
"loss": 5.6767,
|
|
"mean_token_accuracy": 0.19585542976856232,
|
|
"num_tokens": 1465013.0,
|
|
"step": 1730
|
|
},
|
|
{
|
|
"entropy": 5.666249418258667,
|
|
"epoch": 0.03876791759303742,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.0004367422989611743,
|
|
"loss": 5.5938,
|
|
"mean_token_accuracy": 0.1894269183278084,
|
|
"num_tokens": 1469070.0,
|
|
"step": 1735
|
|
},
|
|
{
|
|
"entropy": 5.807224321365356,
|
|
"epoch": 0.038879640698492854,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.0004359211190335153,
|
|
"loss": 5.6786,
|
|
"mean_token_accuracy": 0.1812390998005867,
|
|
"num_tokens": 1473426.0,
|
|
"step": 1740
|
|
},
|
|
{
|
|
"entropy": 5.647406721115113,
|
|
"epoch": 0.0389913638039483,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.00043509552737208923,
|
|
"loss": 5.7754,
|
|
"mean_token_accuracy": 0.18348126858472824,
|
|
"num_tokens": 1477631.0,
|
|
"step": 1745
|
|
},
|
|
{
|
|
"entropy": 5.6247352123260494,
|
|
"epoch": 0.039103086909403735,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.00043426554661090853,
|
|
"loss": 5.8263,
|
|
"mean_token_accuracy": 0.17752304524183274,
|
|
"num_tokens": 1481921.0,
|
|
"step": 1750
|
|
},
|
|
{
|
|
"entropy": 5.883453607559204,
|
|
"epoch": 0.03921481001485917,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00043343119950431516,
|
|
"loss": 5.7789,
|
|
"mean_token_accuracy": 0.18748508542776107,
|
|
"num_tokens": 1486571.0,
|
|
"step": 1755
|
|
},
|
|
{
|
|
"entropy": 5.773577117919922,
|
|
"epoch": 0.03932653312031461,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.00043259250892635644,
|
|
"loss": 5.7668,
|
|
"mean_token_accuracy": 0.17757585793733596,
|
|
"num_tokens": 1490952.0,
|
|
"step": 1760
|
|
},
|
|
{
|
|
"entropy": 5.652618646621704,
|
|
"epoch": 0.03943825622577005,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.0004317494978701582,
|
|
"loss": 5.6286,
|
|
"mean_token_accuracy": 0.1940794676542282,
|
|
"num_tokens": 1495144.0,
|
|
"step": 1765
|
|
},
|
|
{
|
|
"entropy": 5.822029495239258,
|
|
"epoch": 0.03954997933122549,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.0004309021894472943,
|
|
"loss": 5.7134,
|
|
"mean_token_accuracy": 0.18479356169700623,
|
|
"num_tokens": 1499386.0,
|
|
"step": 1770
|
|
},
|
|
{
|
|
"entropy": 5.70394229888916,
|
|
"epoch": 0.03966170243668093,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 0.0004300506068871534,
|
|
"loss": 5.6437,
|
|
"mean_token_accuracy": 0.19557296335697175,
|
|
"num_tokens": 1503583.0,
|
|
"step": 1775
|
|
},
|
|
{
|
|
"entropy": 5.668413591384888,
|
|
"epoch": 0.03977342554213637,
|
|
"grad_norm": 1.46875,
|
|
"learning_rate": 0.00042919477353630135,
|
|
"loss": 5.6645,
|
|
"mean_token_accuracy": 0.184171424806118,
|
|
"num_tokens": 1507680.0,
|
|
"step": 1780
|
|
},
|
|
{
|
|
"entropy": 5.8005297660827635,
|
|
"epoch": 0.03988514864759181,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.000428334712857842,
|
|
"loss": 5.7538,
|
|
"mean_token_accuracy": 0.18311063945293427,
|
|
"num_tokens": 1511856.0,
|
|
"step": 1785
|
|
},
|
|
{
|
|
"entropy": 5.912327814102173,
|
|
"epoch": 0.039996871753047246,
|
|
"grad_norm": 1.234375,
|
|
"learning_rate": 0.00042747044843077304,
|
|
"loss": 5.7356,
|
|
"mean_token_accuracy": 0.17745298892259598,
|
|
"num_tokens": 1516176.0,
|
|
"step": 1790
|
|
},
|
|
{
|
|
"entropy": 5.7352742671966555,
|
|
"epoch": 0.04010859485850269,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.00042660200394934047,
|
|
"loss": 5.635,
|
|
"mean_token_accuracy": 0.19073558300733567,
|
|
"num_tokens": 1520060.0,
|
|
"step": 1795
|
|
},
|
|
{
|
|
"entropy": 5.495400428771973,
|
|
"epoch": 0.04022031796395813,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.00042572940322238844,
|
|
"loss": 5.7122,
|
|
"mean_token_accuracy": 0.18623971492052077,
|
|
"num_tokens": 1524165.0,
|
|
"step": 1800
|
|
},
|
|
{
|
|
"entropy": 5.868027305603027,
|
|
"epoch": 0.040332041069413564,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.00042485267017270664,
|
|
"loss": 5.8853,
|
|
"mean_token_accuracy": 0.1717434585094452,
|
|
"num_tokens": 1528065.0,
|
|
"step": 1805
|
|
},
|
|
{
|
|
"entropy": 5.821336555480957,
|
|
"epoch": 0.040443764174869,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0004239718288363745,
|
|
"loss": 5.6691,
|
|
"mean_token_accuracy": 0.19097715318202974,
|
|
"num_tokens": 1532618.0,
|
|
"step": 1810
|
|
},
|
|
{
|
|
"entropy": 5.592393159866333,
|
|
"epoch": 0.040555487280324445,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.0004230869033621023,
|
|
"loss": 5.6987,
|
|
"mean_token_accuracy": 0.1828291967511177,
|
|
"num_tokens": 1536965.0,
|
|
"step": 1815
|
|
},
|
|
{
|
|
"entropy": 5.767320680618286,
|
|
"epoch": 0.04066721038577988,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.0004221979180105688,
|
|
"loss": 5.8283,
|
|
"mean_token_accuracy": 0.17617318630218506,
|
|
"num_tokens": 1541332.0,
|
|
"step": 1820
|
|
},
|
|
{
|
|
"entropy": 5.672343111038208,
|
|
"epoch": 0.04077893349123532,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.00042130489715375645,
|
|
"loss": 5.5244,
|
|
"mean_token_accuracy": 0.2023320123553276,
|
|
"num_tokens": 1545324.0,
|
|
"step": 1825
|
|
},
|
|
{
|
|
"entropy": 5.767176914215088,
|
|
"epoch": 0.040890656596690764,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 0.00042040786527428335,
|
|
"loss": 5.7512,
|
|
"mean_token_accuracy": 0.1897562712430954,
|
|
"num_tokens": 1549409.0,
|
|
"step": 1830
|
|
},
|
|
{
|
|
"entropy": 5.759979391098023,
|
|
"epoch": 0.0410023797021462,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.0004195068469647315,
|
|
"loss": 5.7413,
|
|
"mean_token_accuracy": 0.1818301036953926,
|
|
"num_tokens": 1553526.0,
|
|
"step": 1835
|
|
},
|
|
{
|
|
"entropy": 5.775640344619751,
|
|
"epoch": 0.04111410280760164,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.00041860186692697297,
|
|
"loss": 5.8355,
|
|
"mean_token_accuracy": 0.1880885362625122,
|
|
"num_tokens": 1558475.0,
|
|
"step": 1840
|
|
},
|
|
{
|
|
"entropy": 5.726383304595947,
|
|
"epoch": 0.04122582591305708,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.00041769294997149264,
|
|
"loss": 5.6429,
|
|
"mean_token_accuracy": 0.18811526596546174,
|
|
"num_tokens": 1562702.0,
|
|
"step": 1845
|
|
},
|
|
{
|
|
"entropy": 5.63990626335144,
|
|
"epoch": 0.04133754901851252,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.0004167801210167081,
|
|
"loss": 5.6439,
|
|
"mean_token_accuracy": 0.1975985050201416,
|
|
"num_tokens": 1566706.0,
|
|
"step": 1850
|
|
},
|
|
{
|
|
"entropy": 5.644278717041016,
|
|
"epoch": 0.041449272123967956,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.0004158634050882861,
|
|
"loss": 5.6656,
|
|
"mean_token_accuracy": 0.1993358850479126,
|
|
"num_tokens": 1570843.0,
|
|
"step": 1855
|
|
},
|
|
{
|
|
"entropy": 5.706397962570191,
|
|
"epoch": 0.0415609952294234,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.0004149428273184569,
|
|
"loss": 5.7084,
|
|
"mean_token_accuracy": 0.18736132681369783,
|
|
"num_tokens": 1575308.0,
|
|
"step": 1860
|
|
},
|
|
{
|
|
"entropy": 5.728898191452027,
|
|
"epoch": 0.04167271833487884,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.0004140184129453253,
|
|
"loss": 5.8393,
|
|
"mean_token_accuracy": 0.17313541173934938,
|
|
"num_tokens": 1579397.0,
|
|
"step": 1865
|
|
},
|
|
{
|
|
"entropy": 5.7663342475891115,
|
|
"epoch": 0.041784441440334275,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.000413090187312178,
|
|
"loss": 5.6239,
|
|
"mean_token_accuracy": 0.18672820925712585,
|
|
"num_tokens": 1583464.0,
|
|
"step": 1870
|
|
},
|
|
{
|
|
"entropy": 5.718973779678345,
|
|
"epoch": 0.04189616454578971,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.0004121581758667898,
|
|
"loss": 5.6461,
|
|
"mean_token_accuracy": 0.18322024047374724,
|
|
"num_tokens": 1587948.0,
|
|
"step": 1875
|
|
},
|
|
{
|
|
"entropy": 5.578612995147705,
|
|
"epoch": 0.042007887651245156,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.00041122240416072533,
|
|
"loss": 5.5589,
|
|
"mean_token_accuracy": 0.21150048673152924,
|
|
"num_tokens": 1591855.0,
|
|
"step": 1880
|
|
},
|
|
{
|
|
"entropy": 5.71712212562561,
|
|
"epoch": 0.04211961075670059,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.0004102828978486385,
|
|
"loss": 5.6911,
|
|
"mean_token_accuracy": 0.19235271066427231,
|
|
"num_tokens": 1595951.0,
|
|
"step": 1885
|
|
},
|
|
{
|
|
"entropy": 5.7032770156860355,
|
|
"epoch": 0.04223133386215603,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.0004093396826875695,
|
|
"loss": 5.6402,
|
|
"mean_token_accuracy": 0.1927582010626793,
|
|
"num_tokens": 1600091.0,
|
|
"step": 1890
|
|
},
|
|
{
|
|
"entropy": 5.731023693084717,
|
|
"epoch": 0.042343056967611474,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.00040839278453623837,
|
|
"loss": 5.706,
|
|
"mean_token_accuracy": 0.1776691198348999,
|
|
"num_tokens": 1604093.0,
|
|
"step": 1895
|
|
},
|
|
{
|
|
"entropy": 5.688064241409302,
|
|
"epoch": 0.04245478007306691,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.0004074422293543363,
|
|
"loss": 5.6452,
|
|
"mean_token_accuracy": 0.19756510853767395,
|
|
"num_tokens": 1608254.0,
|
|
"step": 1900
|
|
},
|
|
{
|
|
"entropy": 5.602776527404785,
|
|
"epoch": 0.04256650317852235,
|
|
"grad_norm": 1.2890625,
|
|
"learning_rate": 0.0004064880432018137,
|
|
"loss": 5.7254,
|
|
"mean_token_accuracy": 0.18962162733078003,
|
|
"num_tokens": 1612064.0,
|
|
"step": 1905
|
|
},
|
|
{
|
|
"entropy": 5.6795135021209715,
|
|
"epoch": 0.04267822628397779,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.00040553025223816615,
|
|
"loss": 5.7919,
|
|
"mean_token_accuracy": 0.18375196754932405,
|
|
"num_tokens": 1616494.0,
|
|
"step": 1910
|
|
},
|
|
{
|
|
"entropy": 5.72490439414978,
|
|
"epoch": 0.04278994938943323,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.00040456888272171653,
|
|
"loss": 5.4819,
|
|
"mean_token_accuracy": 0.20083572566509247,
|
|
"num_tokens": 1620432.0,
|
|
"step": 1915
|
|
},
|
|
{
|
|
"entropy": 5.752138948440551,
|
|
"epoch": 0.04290167249488867,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.00040360396100889577,
|
|
"loss": 5.7682,
|
|
"mean_token_accuracy": 0.17632409781217576,
|
|
"num_tokens": 1625170.0,
|
|
"step": 1920
|
|
},
|
|
{
|
|
"entropy": 5.650776195526123,
|
|
"epoch": 0.043013395600344104,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.0004026355135535202,
|
|
"loss": 5.6995,
|
|
"mean_token_accuracy": 0.17991890013217926,
|
|
"num_tokens": 1629487.0,
|
|
"step": 1925
|
|
},
|
|
{
|
|
"entropy": 5.6605809211730955,
|
|
"epoch": 0.04312511870579955,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.000401663566906066,
|
|
"loss": 5.6162,
|
|
"mean_token_accuracy": 0.19089022427797317,
|
|
"num_tokens": 1633868.0,
|
|
"step": 1930
|
|
},
|
|
{
|
|
"entropy": 5.6820018768310545,
|
|
"epoch": 0.043236841811254985,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.00040068814771294134,
|
|
"loss": 5.7443,
|
|
"mean_token_accuracy": 0.17744273394346238,
|
|
"num_tokens": 1638153.0,
|
|
"step": 1935
|
|
},
|
|
{
|
|
"entropy": 5.644814300537109,
|
|
"epoch": 0.04334856491671042,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.0003997092827157562,
|
|
"loss": 5.5181,
|
|
"mean_token_accuracy": 0.22187568694353105,
|
|
"num_tokens": 1642365.0,
|
|
"step": 1940
|
|
},
|
|
{
|
|
"entropy": 5.932576513290405,
|
|
"epoch": 0.043460288022165866,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 0.000398726998750589,
|
|
"loss": 5.6869,
|
|
"mean_token_accuracy": 0.1798103779554367,
|
|
"num_tokens": 1646202.0,
|
|
"step": 1945
|
|
},
|
|
{
|
|
"entropy": 5.623147630691529,
|
|
"epoch": 0.0435720111276213,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00039774132274725076,
|
|
"loss": 5.566,
|
|
"mean_token_accuracy": 0.198709699511528,
|
|
"num_tokens": 1650542.0,
|
|
"step": 1950
|
|
},
|
|
{
|
|
"entropy": 5.543530559539795,
|
|
"epoch": 0.04368373423307674,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 0.00039675228172854707,
|
|
"loss": 5.5123,
|
|
"mean_token_accuracy": 0.1991402894258499,
|
|
"num_tokens": 1654386.0,
|
|
"step": 1955
|
|
},
|
|
{
|
|
"entropy": 5.532556343078613,
|
|
"epoch": 0.043795457338532184,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.0003957599028095371,
|
|
"loss": 5.5337,
|
|
"mean_token_accuracy": 0.19989801347255706,
|
|
"num_tokens": 1658211.0,
|
|
"step": 1960
|
|
},
|
|
{
|
|
"entropy": 5.6866264820098875,
|
|
"epoch": 0.04390718044398762,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 0.00039476421319679017,
|
|
"loss": 5.684,
|
|
"mean_token_accuracy": 0.18525188714265822,
|
|
"num_tokens": 1662757.0,
|
|
"step": 1965
|
|
},
|
|
{
|
|
"entropy": 5.693357276916504,
|
|
"epoch": 0.04401890354944306,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.00039376524018764,
|
|
"loss": 5.667,
|
|
"mean_token_accuracy": 0.18190956711769105,
|
|
"num_tokens": 1666714.0,
|
|
"step": 1970
|
|
},
|
|
{
|
|
"entropy": 5.652660083770752,
|
|
"epoch": 0.0441306266548985,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.00039276301116943616,
|
|
"loss": 5.6781,
|
|
"mean_token_accuracy": 0.18582605719566345,
|
|
"num_tokens": 1671029.0,
|
|
"step": 1975
|
|
},
|
|
{
|
|
"entropy": 5.742524433135986,
|
|
"epoch": 0.04424234976035394,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.0003917575536187936,
|
|
"loss": 5.6653,
|
|
"mean_token_accuracy": 0.18949595242738723,
|
|
"num_tokens": 1675366.0,
|
|
"step": 1980
|
|
},
|
|
{
|
|
"entropy": 5.644546365737915,
|
|
"epoch": 0.04435407286580938,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 0.00039074889510083894,
|
|
"loss": 5.5579,
|
|
"mean_token_accuracy": 0.2034739837050438,
|
|
"num_tokens": 1679657.0,
|
|
"step": 1985
|
|
},
|
|
{
|
|
"entropy": 5.530533266067505,
|
|
"epoch": 0.044465795971264814,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.00038973706326845495,
|
|
"loss": 5.5582,
|
|
"mean_token_accuracy": 0.19022609293460846,
|
|
"num_tokens": 1683676.0,
|
|
"step": 1990
|
|
},
|
|
{
|
|
"entropy": 5.533994388580322,
|
|
"epoch": 0.04457751907672026,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.0003887220858615225,
|
|
"loss": 5.5562,
|
|
"mean_token_accuracy": 0.20196932703256607,
|
|
"num_tokens": 1687718.0,
|
|
"step": 1995
|
|
},
|
|
{
|
|
"entropy": 5.549967861175537,
|
|
"epoch": 0.044689242182175695,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 0.0003877039907061597,
|
|
"loss": 5.598,
|
|
"mean_token_accuracy": 0.18386923521757126,
|
|
"num_tokens": 1691434.0,
|
|
"step": 2000
|
|
},
|
|
{
|
|
"entropy": 5.598109769821167,
|
|
"epoch": 0.04480096528763113,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.0003866828057139598,
|
|
"loss": 5.6704,
|
|
"mean_token_accuracy": 0.1848026692867279,
|
|
"num_tokens": 1695714.0,
|
|
"step": 2005
|
|
},
|
|
{
|
|
"entropy": 5.667990732192993,
|
|
"epoch": 0.044912688393086576,
|
|
"grad_norm": 1.4921875,
|
|
"learning_rate": 0.00038565855888122503,
|
|
"loss": 5.4512,
|
|
"mean_token_accuracy": 0.19753458797931672,
|
|
"num_tokens": 1700245.0,
|
|
"step": 2010
|
|
},
|
|
{
|
|
"entropy": 5.413176441192627,
|
|
"epoch": 0.045024411498542014,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.00038463127828819975,
|
|
"loss": 5.5758,
|
|
"mean_token_accuracy": 0.19638268798589706,
|
|
"num_tokens": 1704329.0,
|
|
"step": 2015
|
|
},
|
|
{
|
|
"entropy": 5.654092931747437,
|
|
"epoch": 0.04513613460399745,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.00038360099209830043,
|
|
"loss": 5.601,
|
|
"mean_token_accuracy": 0.1924460396170616,
|
|
"num_tokens": 1708387.0,
|
|
"step": 2020
|
|
},
|
|
{
|
|
"entropy": 5.643488121032715,
|
|
"epoch": 0.045247857709452895,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.0003825677285573433,
|
|
"loss": 5.5818,
|
|
"mean_token_accuracy": 0.190503291785717,
|
|
"num_tokens": 1712620.0,
|
|
"step": 2025
|
|
},
|
|
{
|
|
"entropy": 5.61337571144104,
|
|
"epoch": 0.04535958081490833,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.00038153151599277027,
|
|
"loss": 5.5974,
|
|
"mean_token_accuracy": 0.19394924491643906,
|
|
"num_tokens": 1716862.0,
|
|
"step": 2030
|
|
},
|
|
{
|
|
"entropy": 5.756779384613037,
|
|
"epoch": 0.04547130392036377,
|
|
"grad_norm": 1.4140625,
|
|
"learning_rate": 0.0003804923828128723,
|
|
"loss": 5.6628,
|
|
"mean_token_accuracy": 0.17784910202026366,
|
|
"num_tokens": 1721121.0,
|
|
"step": 2035
|
|
},
|
|
{
|
|
"entropy": 5.6854815006256105,
|
|
"epoch": 0.04558302702581921,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.0003794503575060104,
|
|
"loss": 5.4768,
|
|
"mean_token_accuracy": 0.20856099724769592,
|
|
"num_tokens": 1725307.0,
|
|
"step": 2040
|
|
},
|
|
{
|
|
"entropy": 5.673893165588379,
|
|
"epoch": 0.04569475013127465,
|
|
"grad_norm": 1.2421875,
|
|
"learning_rate": 0.00037840546863983484,
|
|
"loss": 5.7124,
|
|
"mean_token_accuracy": 0.18246976882219315,
|
|
"num_tokens": 1729931.0,
|
|
"step": 2045
|
|
},
|
|
{
|
|
"entropy": 5.563566875457764,
|
|
"epoch": 0.04580647323673009,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.0003773577448605015,
|
|
"loss": 5.6909,
|
|
"mean_token_accuracy": 0.18922285884618759,
|
|
"num_tokens": 1734117.0,
|
|
"step": 2050
|
|
},
|
|
{
|
|
"entropy": 5.712671947479248,
|
|
"epoch": 0.045918196342185524,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.0003763072148918872,
|
|
"loss": 5.7125,
|
|
"mean_token_accuracy": 0.18284738808870316,
|
|
"num_tokens": 1738246.0,
|
|
"step": 2055
|
|
},
|
|
{
|
|
"entropy": 5.79467134475708,
|
|
"epoch": 0.04602991944764097,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.0003752539075348017,
|
|
"loss": 5.6553,
|
|
"mean_token_accuracy": 0.1847241759300232,
|
|
"num_tokens": 1742557.0,
|
|
"step": 2060
|
|
},
|
|
{
|
|
"entropy": 5.5943460941314695,
|
|
"epoch": 0.046141642553096406,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.00037419785166619817,
|
|
"loss": 5.5863,
|
|
"mean_token_accuracy": 0.19431493878364564,
|
|
"num_tokens": 1746639.0,
|
|
"step": 2065
|
|
},
|
|
{
|
|
"entropy": 5.792494869232177,
|
|
"epoch": 0.04625336565855184,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.0003731390762383818,
|
|
"loss": 5.673,
|
|
"mean_token_accuracy": 0.1853930190205574,
|
|
"num_tokens": 1751528.0,
|
|
"step": 2070
|
|
},
|
|
{
|
|
"entropy": 5.629391717910766,
|
|
"epoch": 0.04636508876400729,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.0003720776102782158,
|
|
"loss": 5.4907,
|
|
"mean_token_accuracy": 0.20055108666419982,
|
|
"num_tokens": 1755734.0,
|
|
"step": 2075
|
|
},
|
|
{
|
|
"entropy": 5.545011901855469,
|
|
"epoch": 0.046476811869462724,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00037101348288632555,
|
|
"loss": 5.6238,
|
|
"mean_token_accuracy": 0.19830233454704285,
|
|
"num_tokens": 1760188.0,
|
|
"step": 2080
|
|
},
|
|
{
|
|
"entropy": 5.631131315231324,
|
|
"epoch": 0.04658853497491816,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.0003699467232363012,
|
|
"loss": 5.5733,
|
|
"mean_token_accuracy": 0.19923410564661026,
|
|
"num_tokens": 1764469.0,
|
|
"step": 2085
|
|
},
|
|
{
|
|
"entropy": 5.608803081512451,
|
|
"epoch": 0.046700258080373605,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.0003688773605738973,
|
|
"loss": 5.6552,
|
|
"mean_token_accuracy": 0.18862041234970092,
|
|
"num_tokens": 1768946.0,
|
|
"step": 2090
|
|
},
|
|
{
|
|
"entropy": 5.570499086380005,
|
|
"epoch": 0.04681198118582904,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.00036780542421623134,
|
|
"loss": 5.4629,
|
|
"mean_token_accuracy": 0.21054447591304778,
|
|
"num_tokens": 1772806.0,
|
|
"step": 2095
|
|
},
|
|
{
|
|
"entropy": 5.711140966415405,
|
|
"epoch": 0.04692370429128448,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.0003667309435509802,
|
|
"loss": 5.7695,
|
|
"mean_token_accuracy": 0.17885771691799163,
|
|
"num_tokens": 1777333.0,
|
|
"step": 2100
|
|
},
|
|
{
|
|
"entropy": 5.665574312210083,
|
|
"epoch": 0.047035427396739916,
|
|
"grad_norm": 1.515625,
|
|
"learning_rate": 0.0003656539480355741,
|
|
"loss": 5.5474,
|
|
"mean_token_accuracy": 0.19172714352607728,
|
|
"num_tokens": 1781232.0,
|
|
"step": 2105
|
|
},
|
|
{
|
|
"entropy": 5.638804864883423,
|
|
"epoch": 0.04714715050219536,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.0003645744671963891,
|
|
"loss": 5.6286,
|
|
"mean_token_accuracy": 0.18934958130121232,
|
|
"num_tokens": 1785543.0,
|
|
"step": 2110
|
|
},
|
|
{
|
|
"entropy": 5.59114089012146,
|
|
"epoch": 0.0472588736076508,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0003634925306279376,
|
|
"loss": 5.6293,
|
|
"mean_token_accuracy": 0.19169047027826308,
|
|
"num_tokens": 1789926.0,
|
|
"step": 2115
|
|
},
|
|
{
|
|
"entropy": 5.635175561904907,
|
|
"epoch": 0.047370596713106235,
|
|
"grad_norm": 1.3203125,
|
|
"learning_rate": 0.0003624081679920574,
|
|
"loss": 5.5693,
|
|
"mean_token_accuracy": 0.19705144166946412,
|
|
"num_tokens": 1794307.0,
|
|
"step": 2120
|
|
},
|
|
{
|
|
"entropy": 5.572471618652344,
|
|
"epoch": 0.04748231981856168,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 0.0003613214090170977,
|
|
"loss": 5.4883,
|
|
"mean_token_accuracy": 0.20081721395254135,
|
|
"num_tokens": 1798553.0,
|
|
"step": 2125
|
|
},
|
|
{
|
|
"entropy": 5.60513825416565,
|
|
"epoch": 0.047594042924017116,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.0003602322834971048,
|
|
"loss": 5.4432,
|
|
"mean_token_accuracy": 0.21184519529342652,
|
|
"num_tokens": 1802795.0,
|
|
"step": 2130
|
|
},
|
|
{
|
|
"entropy": 5.560659694671631,
|
|
"epoch": 0.04770576602947255,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.0003591408212910051,
|
|
"loss": 5.4397,
|
|
"mean_token_accuracy": 0.20747311115264894,
|
|
"num_tokens": 1806762.0,
|
|
"step": 2135
|
|
},
|
|
{
|
|
"entropy": 5.505787324905396,
|
|
"epoch": 0.047817489134928,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.0003580470523217863,
|
|
"loss": 5.4877,
|
|
"mean_token_accuracy": 0.20571576803922653,
|
|
"num_tokens": 1811067.0,
|
|
"step": 2140
|
|
},
|
|
{
|
|
"entropy": 5.4739891529083256,
|
|
"epoch": 0.047929212240383434,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.0003569510065756771,
|
|
"loss": 5.4798,
|
|
"mean_token_accuracy": 0.20073288083076476,
|
|
"num_tokens": 1815180.0,
|
|
"step": 2145
|
|
},
|
|
{
|
|
"entropy": 5.503992986679077,
|
|
"epoch": 0.04804093534583887,
|
|
"grad_norm": 1.4296875,
|
|
"learning_rate": 0.0003558527141013254,
|
|
"loss": 5.6137,
|
|
"mean_token_accuracy": 0.18731415420770645,
|
|
"num_tokens": 1819564.0,
|
|
"step": 2150
|
|
},
|
|
{
|
|
"entropy": 5.636356496810913,
|
|
"epoch": 0.048152658451294315,
|
|
"grad_norm": 1.21875,
|
|
"learning_rate": 0.0003547522050089742,
|
|
"loss": 5.5458,
|
|
"mean_token_accuracy": 0.20209146589040755,
|
|
"num_tokens": 1823974.0,
|
|
"step": 2155
|
|
},
|
|
{
|
|
"entropy": 5.744458246231079,
|
|
"epoch": 0.04826438155674975,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.00035364950946963606,
|
|
"loss": 5.627,
|
|
"mean_token_accuracy": 0.19121788069605827,
|
|
"num_tokens": 1828532.0,
|
|
"step": 2160
|
|
},
|
|
{
|
|
"entropy": 5.5946817874908445,
|
|
"epoch": 0.04837610466220519,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.0003525446577142663,
|
|
"loss": 5.5438,
|
|
"mean_token_accuracy": 0.20079231560230254,
|
|
"num_tokens": 1832678.0,
|
|
"step": 2165
|
|
},
|
|
{
|
|
"entropy": 5.55777587890625,
|
|
"epoch": 0.04848782776766063,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.00035143768003293395,
|
|
"loss": 5.5345,
|
|
"mean_token_accuracy": 0.20150283724069595,
|
|
"num_tokens": 1837118.0,
|
|
"step": 2170
|
|
},
|
|
{
|
|
"entropy": 5.531144762039185,
|
|
"epoch": 0.04859955087311607,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.0003503286067739913,
|
|
"loss": 5.4624,
|
|
"mean_token_accuracy": 0.1982642188668251,
|
|
"num_tokens": 1841348.0,
|
|
"step": 2175
|
|
},
|
|
{
|
|
"entropy": 5.564840984344483,
|
|
"epoch": 0.04871127397857151,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 0.00034921746834324193,
|
|
"loss": 5.5966,
|
|
"mean_token_accuracy": 0.19088261425495148,
|
|
"num_tokens": 1845715.0,
|
|
"step": 2180
|
|
},
|
|
{
|
|
"entropy": 5.588362360000611,
|
|
"epoch": 0.048822997084026945,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.0003481042952031072,
|
|
"loss": 5.5474,
|
|
"mean_token_accuracy": 0.18980715721845626,
|
|
"num_tokens": 1850028.0,
|
|
"step": 2185
|
|
},
|
|
{
|
|
"entropy": 5.563361167907715,
|
|
"epoch": 0.04893472018948239,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.0003469891178717911,
|
|
"loss": 5.5563,
|
|
"mean_token_accuracy": 0.19482785314321518,
|
|
"num_tokens": 1854331.0,
|
|
"step": 2190
|
|
},
|
|
{
|
|
"entropy": 5.4609252452850345,
|
|
"epoch": 0.049046443294937826,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.0003458719669224436,
|
|
"loss": 5.5058,
|
|
"mean_token_accuracy": 0.21373057812452317,
|
|
"num_tokens": 1858740.0,
|
|
"step": 2195
|
|
},
|
|
{
|
|
"entropy": 5.689063501358032,
|
|
"epoch": 0.04915816640039326,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0003447528729823221,
|
|
"loss": 5.6488,
|
|
"mean_token_accuracy": 0.19399519264698029,
|
|
"num_tokens": 1863091.0,
|
|
"step": 2200
|
|
},
|
|
{
|
|
"entropy": 5.661524486541748,
|
|
"epoch": 0.04926988950584871,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.0003436318667319525,
|
|
"loss": 5.5638,
|
|
"mean_token_accuracy": 0.1881960764527321,
|
|
"num_tokens": 1867164.0,
|
|
"step": 2205
|
|
},
|
|
{
|
|
"entropy": 5.527868843078613,
|
|
"epoch": 0.049381612611304145,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.00034250897890428716,
|
|
"loss": 5.6453,
|
|
"mean_token_accuracy": 0.18407594561576843,
|
|
"num_tokens": 1871614.0,
|
|
"step": 2210
|
|
},
|
|
{
|
|
"entropy": 5.720584440231323,
|
|
"epoch": 0.04949333571675958,
|
|
"grad_norm": 1.2578125,
|
|
"learning_rate": 0.0003413842402838633,
|
|
"loss": 5.668,
|
|
"mean_token_accuracy": 0.18221641331911087,
|
|
"num_tokens": 1876315.0,
|
|
"step": 2215
|
|
},
|
|
{
|
|
"entropy": 5.667582225799561,
|
|
"epoch": 0.049605058822215026,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.00034025768170595834,
|
|
"loss": 5.6102,
|
|
"mean_token_accuracy": 0.18524401932954787,
|
|
"num_tokens": 1880404.0,
|
|
"step": 2220
|
|
},
|
|
{
|
|
"entropy": 5.604895305633545,
|
|
"epoch": 0.04971678192767046,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 0.0003391293340557446,
|
|
"loss": 5.5866,
|
|
"mean_token_accuracy": 0.18956515491008757,
|
|
"num_tokens": 1885056.0,
|
|
"step": 2225
|
|
},
|
|
{
|
|
"entropy": 5.529463148117065,
|
|
"epoch": 0.0498285050331259,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.0003379992282674431,
|
|
"loss": 5.5078,
|
|
"mean_token_accuracy": 0.19293873012065887,
|
|
"num_tokens": 1889170.0,
|
|
"step": 2230
|
|
},
|
|
{
|
|
"entropy": 5.542259788513183,
|
|
"epoch": 0.04994022813858134,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.0003368673953234749,
|
|
"loss": 5.4402,
|
|
"mean_token_accuracy": 0.20548522472381592,
|
|
"num_tokens": 1893344.0,
|
|
"step": 2235
|
|
},
|
|
{
|
|
"entropy": 5.5274498462677,
|
|
"epoch": 0.05005195124403678,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.00033573386625361176,
|
|
"loss": 5.5269,
|
|
"mean_token_accuracy": 0.20261189937591553,
|
|
"num_tokens": 1897399.0,
|
|
"step": 2240
|
|
},
|
|
{
|
|
"entropy": 5.537028121948242,
|
|
"epoch": 0.05016367434949222,
|
|
"grad_norm": 1.3125,
|
|
"learning_rate": 0.00033459867213412567,
|
|
"loss": 5.6219,
|
|
"mean_token_accuracy": 0.19377635270357133,
|
|
"num_tokens": 1901687.0,
|
|
"step": 2245
|
|
},
|
|
{
|
|
"entropy": 5.4919548511505125,
|
|
"epoch": 0.050275397454947655,
|
|
"grad_norm": 1.5,
|
|
"learning_rate": 0.000333461844086937,
|
|
"loss": 5.3408,
|
|
"mean_token_accuracy": 0.2090111032128334,
|
|
"num_tokens": 1905556.0,
|
|
"step": 2250
|
|
},
|
|
{
|
|
"entropy": 5.511547946929932,
|
|
"epoch": 0.0503871205604031,
|
|
"grad_norm": 1.5546875,
|
|
"learning_rate": 0.00033232341327876097,
|
|
"loss": 5.5062,
|
|
"mean_token_accuracy": 0.19810119569301604,
|
|
"num_tokens": 1909477.0,
|
|
"step": 2255
|
|
},
|
|
{
|
|
"entropy": 5.554897499084473,
|
|
"epoch": 0.05049884366585854,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.0003311834109202531,
|
|
"loss": 5.551,
|
|
"mean_token_accuracy": 0.18918206989765168,
|
|
"num_tokens": 1913593.0,
|
|
"step": 2260
|
|
},
|
|
{
|
|
"entropy": 5.570584964752197,
|
|
"epoch": 0.050610566771313974,
|
|
"grad_norm": 1.578125,
|
|
"learning_rate": 0.00033004186826515416,
|
|
"loss": 5.6149,
|
|
"mean_token_accuracy": 0.19200795739889145,
|
|
"num_tokens": 1917853.0,
|
|
"step": 2265
|
|
},
|
|
{
|
|
"entropy": 5.660935258865356,
|
|
"epoch": 0.05072228987676942,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 0.0003288988166094324,
|
|
"loss": 5.4973,
|
|
"mean_token_accuracy": 0.19082937389612198,
|
|
"num_tokens": 1922075.0,
|
|
"step": 2270
|
|
},
|
|
{
|
|
"entropy": 5.602800416946411,
|
|
"epoch": 0.050834012982224855,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.00032775428729042656,
|
|
"loss": 5.5452,
|
|
"mean_token_accuracy": 0.1940069317817688,
|
|
"num_tokens": 1926595.0,
|
|
"step": 2275
|
|
},
|
|
{
|
|
"entropy": 5.500808811187744,
|
|
"epoch": 0.05094573608768029,
|
|
"grad_norm": 1.2734375,
|
|
"learning_rate": 0.000326608311685986,
|
|
"loss": 5.4548,
|
|
"mean_token_accuracy": 0.20891236662864685,
|
|
"num_tokens": 1930748.0,
|
|
"step": 2280
|
|
},
|
|
{
|
|
"entropy": 5.621628379821777,
|
|
"epoch": 0.05105745919313573,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.0003254609212136108,
|
|
"loss": 5.6094,
|
|
"mean_token_accuracy": 0.19751961529254913,
|
|
"num_tokens": 1935243.0,
|
|
"step": 2285
|
|
},
|
|
{
|
|
"entropy": 5.719082927703857,
|
|
"epoch": 0.05116918229859117,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.00032431214732959036,
|
|
"loss": 5.6297,
|
|
"mean_token_accuracy": 0.1876582086086273,
|
|
"num_tokens": 1939932.0,
|
|
"step": 2290
|
|
},
|
|
{
|
|
"entropy": 5.7131959915161135,
|
|
"epoch": 0.05128090540404661,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.000323162021528141,
|
|
"loss": 5.5889,
|
|
"mean_token_accuracy": 0.18834880739450455,
|
|
"num_tokens": 1944021.0,
|
|
"step": 2295
|
|
},
|
|
{
|
|
"entropy": 5.4991027355194095,
|
|
"epoch": 0.05139262850950205,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.00032201057534054264,
|
|
"loss": 5.5017,
|
|
"mean_token_accuracy": 0.18835568577051162,
|
|
"num_tokens": 1948339.0,
|
|
"step": 2300
|
|
},
|
|
{
|
|
"entropy": 5.584790372848511,
|
|
"epoch": 0.05150435161495749,
|
|
"grad_norm": 1.4375,
|
|
"learning_rate": 0.00032085784033427414,
|
|
"loss": 5.6607,
|
|
"mean_token_accuracy": 0.1897027984261513,
|
|
"num_tokens": 1952489.0,
|
|
"step": 2305
|
|
},
|
|
{
|
|
"entropy": 5.5921056270599365,
|
|
"epoch": 0.05161607472041293,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.0003197038481121478,
|
|
"loss": 5.4742,
|
|
"mean_token_accuracy": 0.19483380615711213,
|
|
"num_tokens": 1956765.0,
|
|
"step": 2310
|
|
},
|
|
{
|
|
"entropy": 5.654494667053223,
|
|
"epoch": 0.051727797825868366,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.0003185486303114436,
|
|
"loss": 5.5262,
|
|
"mean_token_accuracy": 0.19184301048517227,
|
|
"num_tokens": 1960490.0,
|
|
"step": 2315
|
|
},
|
|
{
|
|
"entropy": 5.637120246887207,
|
|
"epoch": 0.05183952093132381,
|
|
"grad_norm": 1.421875,
|
|
"learning_rate": 0.0003173922186030409,
|
|
"loss": 5.5961,
|
|
"mean_token_accuracy": 0.18270457834005355,
|
|
"num_tokens": 1964452.0,
|
|
"step": 2320
|
|
},
|
|
{
|
|
"entropy": 5.681508779525757,
|
|
"epoch": 0.05195124403677925,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.000316234644690551,
|
|
"loss": 5.556,
|
|
"mean_token_accuracy": 0.20058944523334504,
|
|
"num_tokens": 1968765.0,
|
|
"step": 2325
|
|
},
|
|
{
|
|
"entropy": 5.449357986450195,
|
|
"epoch": 0.052062967142234684,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.0003150759403094473,
|
|
"loss": 5.3411,
|
|
"mean_token_accuracy": 0.20803572088479996,
|
|
"num_tokens": 1972696.0,
|
|
"step": 2330
|
|
},
|
|
{
|
|
"entropy": 5.553472948074341,
|
|
"epoch": 0.05217469024769013,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.00031391613722619587,
|
|
"loss": 5.6746,
|
|
"mean_token_accuracy": 0.16769066452980042,
|
|
"num_tokens": 1976873.0,
|
|
"step": 2335
|
|
},
|
|
{
|
|
"entropy": 5.63549394607544,
|
|
"epoch": 0.052286413353145565,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.000312755267237384,
|
|
"loss": 5.4565,
|
|
"mean_token_accuracy": 0.202728009223938,
|
|
"num_tokens": 1981203.0,
|
|
"step": 2340
|
|
},
|
|
{
|
|
"entropy": 5.582929229736328,
|
|
"epoch": 0.052398136458601,
|
|
"grad_norm": 1.296875,
|
|
"learning_rate": 0.0003115933621688488,
|
|
"loss": 5.5184,
|
|
"mean_token_accuracy": 0.1903327912092209,
|
|
"num_tokens": 1985311.0,
|
|
"step": 2345
|
|
},
|
|
{
|
|
"entropy": 5.554009914398193,
|
|
"epoch": 0.05250985956405644,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.00031043045387480487,
|
|
"loss": 5.498,
|
|
"mean_token_accuracy": 0.19480977058410645,
|
|
"num_tokens": 1989525.0,
|
|
"step": 2350
|
|
},
|
|
{
|
|
"entropy": 5.5623664379119875,
|
|
"epoch": 0.052621582669511884,
|
|
"grad_norm": 1.3984375,
|
|
"learning_rate": 0.0003092665742369703,
|
|
"loss": 5.5656,
|
|
"mean_token_accuracy": 0.19292740523815155,
|
|
"num_tokens": 1993322.0,
|
|
"step": 2355
|
|
},
|
|
{
|
|
"entropy": 5.655750799179077,
|
|
"epoch": 0.05273330577496732,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00030810175516369343,
|
|
"loss": 5.6241,
|
|
"mean_token_accuracy": 0.18445832878351212,
|
|
"num_tokens": 1997360.0,
|
|
"step": 2360
|
|
},
|
|
{
|
|
"entropy": 5.713097095489502,
|
|
"epoch": 0.05284502888042276,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.0003069360285890775,
|
|
"loss": 5.4548,
|
|
"mean_token_accuracy": 0.19537921547889708,
|
|
"num_tokens": 2001867.0,
|
|
"step": 2365
|
|
},
|
|
{
|
|
"entropy": 5.628135013580322,
|
|
"epoch": 0.0529567519858782,
|
|
"grad_norm": 1.3671875,
|
|
"learning_rate": 0.00030576942647210547,
|
|
"loss": 5.6659,
|
|
"mean_token_accuracy": 0.1955395072698593,
|
|
"num_tokens": 2005729.0,
|
|
"step": 2370
|
|
},
|
|
{
|
|
"entropy": 5.563907194137573,
|
|
"epoch": 0.05306847509133364,
|
|
"grad_norm": 1.359375,
|
|
"learning_rate": 0.00030460198079576355,
|
|
"loss": 5.5383,
|
|
"mean_token_accuracy": 0.19072917848825455,
|
|
"num_tokens": 2009839.0,
|
|
"step": 2375
|
|
},
|
|
{
|
|
"entropy": 5.570201206207275,
|
|
"epoch": 0.053180198196789076,
|
|
"grad_norm": 1.453125,
|
|
"learning_rate": 0.0003034337235661648,
|
|
"loss": 5.4604,
|
|
"mean_token_accuracy": 0.19360036700963973,
|
|
"num_tokens": 2013967.0,
|
|
"step": 2380
|
|
},
|
|
{
|
|
"entropy": 5.60407509803772,
|
|
"epoch": 0.05329192130224452,
|
|
"grad_norm": 1.4453125,
|
|
"learning_rate": 0.0003022646868116714,
|
|
"loss": 5.583,
|
|
"mean_token_accuracy": 0.19008321315050125,
|
|
"num_tokens": 2018353.0,
|
|
"step": 2385
|
|
},
|
|
{
|
|
"entropy": 5.69129376411438,
|
|
"epoch": 0.05340364440769996,
|
|
"grad_norm": 1.59375,
|
|
"learning_rate": 0.0003010949025820163,
|
|
"loss": 5.6493,
|
|
"mean_token_accuracy": 0.1880006343126297,
|
|
"num_tokens": 2022171.0,
|
|
"step": 2390
|
|
},
|
|
{
|
|
"entropy": 5.608281373977661,
|
|
"epoch": 0.053515367513155394,
|
|
"grad_norm": 1.5390625,
|
|
"learning_rate": 0.0002999244029474252,
|
|
"loss": 5.5591,
|
|
"mean_token_accuracy": 0.20259366631507875,
|
|
"num_tokens": 2026341.0,
|
|
"step": 2395
|
|
},
|
|
{
|
|
"entropy": 5.683151006698608,
|
|
"epoch": 0.05362709061861083,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.00029875321999773684,
|
|
"loss": 5.5891,
|
|
"mean_token_accuracy": 0.18821725249290466,
|
|
"num_tokens": 2031021.0,
|
|
"step": 2400
|
|
},
|
|
{
|
|
"entropy": 5.754010438919067,
|
|
"epoch": 0.053738813724066276,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.00029758138584152333,
|
|
"loss": 5.7149,
|
|
"mean_token_accuracy": 0.17457710206508636,
|
|
"num_tokens": 2035653.0,
|
|
"step": 2405
|
|
},
|
|
{
|
|
"entropy": 5.656412887573242,
|
|
"epoch": 0.05385053682952171,
|
|
"grad_norm": 1.34375,
|
|
"learning_rate": 0.0002964089326052102,
|
|
"loss": 5.6126,
|
|
"mean_token_accuracy": 0.1949571579694748,
|
|
"num_tokens": 2039728.0,
|
|
"step": 2410
|
|
},
|
|
{
|
|
"entropy": 5.580978488922119,
|
|
"epoch": 0.05396225993497715,
|
|
"grad_norm": 1.4765625,
|
|
"learning_rate": 0.0002952358924321949,
|
|
"loss": 5.4874,
|
|
"mean_token_accuracy": 0.19023145288228988,
|
|
"num_tokens": 2044078.0,
|
|
"step": 2415
|
|
},
|
|
{
|
|
"entropy": 5.599613666534424,
|
|
"epoch": 0.054073983040432594,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.00029406229748196657,
|
|
"loss": 5.3794,
|
|
"mean_token_accuracy": 0.20781491547822953,
|
|
"num_tokens": 2048138.0,
|
|
"step": 2420
|
|
},
|
|
{
|
|
"entropy": 5.596552801132202,
|
|
"epoch": 0.05418570614588803,
|
|
"grad_norm": 1.28125,
|
|
"learning_rate": 0.0002928881799292235,
|
|
"loss": 5.6829,
|
|
"mean_token_accuracy": 0.18918677568435668,
|
|
"num_tokens": 2052837.0,
|
|
"step": 2425
|
|
},
|
|
{
|
|
"entropy": 5.640849542617798,
|
|
"epoch": 0.05429742925134347,
|
|
"grad_norm": 1.375,
|
|
"learning_rate": 0.00029171357196299154,
|
|
"loss": 5.5745,
|
|
"mean_token_accuracy": 0.19951188415288926,
|
|
"num_tokens": 2056715.0,
|
|
"step": 2430
|
|
},
|
|
{
|
|
"entropy": 5.623363733291626,
|
|
"epoch": 0.05440915235679891,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0002905385057857414,
|
|
"loss": 5.5352,
|
|
"mean_token_accuracy": 0.19049001783132552,
|
|
"num_tokens": 2060508.0,
|
|
"step": 2435
|
|
},
|
|
{
|
|
"entropy": 5.679731750488282,
|
|
"epoch": 0.05452087546225435,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.0002893630136125058,
|
|
"loss": 5.5896,
|
|
"mean_token_accuracy": 0.18565710335969926,
|
|
"num_tokens": 2065050.0,
|
|
"step": 2440
|
|
},
|
|
{
|
|
"entropy": 5.565289402008057,
|
|
"epoch": 0.054632598567709786,
|
|
"grad_norm": 1.265625,
|
|
"learning_rate": 0.0002881871276699967,
|
|
"loss": 5.4692,
|
|
"mean_token_accuracy": 0.19635592103004457,
|
|
"num_tokens": 2069488.0,
|
|
"step": 2445
|
|
},
|
|
{
|
|
"entropy": 5.526049280166626,
|
|
"epoch": 0.05474432167316523,
|
|
"grad_norm": 1.40625,
|
|
"learning_rate": 0.00028701088019572114,
|
|
"loss": 5.5709,
|
|
"mean_token_accuracy": 0.19352198988199235,
|
|
"num_tokens": 2073605.0,
|
|
"step": 2450
|
|
},
|
|
{
|
|
"entropy": 5.54777684211731,
|
|
"epoch": 0.05485604477862067,
|
|
"grad_norm": 1.328125,
|
|
"learning_rate": 0.0002858343034370977,
|
|
"loss": 5.4322,
|
|
"mean_token_accuracy": 0.1995716482400894,
|
|
"num_tokens": 2077721.0,
|
|
"step": 2455
|
|
},
|
|
{
|
|
"entropy": 5.556557703018188,
|
|
"epoch": 0.054967767884076105,
|
|
"grad_norm": 1.3828125,
|
|
"learning_rate": 0.00028465742965057267,
|
|
"loss": 5.5684,
|
|
"mean_token_accuracy": 0.18811332881450654,
|
|
"num_tokens": 2081945.0,
|
|
"step": 2460
|
|
},
|
|
{
|
|
"entropy": 5.629191112518311,
|
|
"epoch": 0.05507949098953154,
|
|
"grad_norm": 1.3515625,
|
|
"learning_rate": 0.00028348029110073533,
|
|
"loss": 5.659,
|
|
"mean_token_accuracy": 0.1851786345243454,
|
|
"num_tokens": 2086184.0,
|
|
"step": 2465
|
|
},
|
|
{
|
|
"entropy": 5.602310562133789,
|
|
"epoch": 0.055191214094986986,
|
|
"grad_norm": 1.4609375,
|
|
"learning_rate": 0.00028230292005943365,
|
|
"loss": 5.4623,
|
|
"mean_token_accuracy": 0.20515158921480178,
|
|
"num_tokens": 2090271.0,
|
|
"step": 2470
|
|
},
|
|
{
|
|
"entropy": 5.481738805770874,
|
|
"epoch": 0.05530293720044242,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.00028112534880488945,
|
|
"loss": 5.4505,
|
|
"mean_token_accuracy": 0.19859495311975478,
|
|
"num_tokens": 2094427.0,
|
|
"step": 2475
|
|
},
|
|
{
|
|
"entropy": 5.461277008056641,
|
|
"epoch": 0.05541466030589786,
|
|
"grad_norm": 1.5078125,
|
|
"learning_rate": 0.0002799476096208137,
|
|
"loss": 5.2431,
|
|
"mean_token_accuracy": 0.22603217214345933,
|
|
"num_tokens": 2098795.0,
|
|
"step": 2480
|
|
},
|
|
{
|
|
"entropy": 5.590682172775269,
|
|
"epoch": 0.055526383411353304,
|
|
"grad_norm": 1.390625,
|
|
"learning_rate": 0.00027876973479552087,
|
|
"loss": 5.4535,
|
|
"mean_token_accuracy": 0.2132108509540558,
|
|
"num_tokens": 2103190.0,
|
|
"step": 2485
|
|
},
|
|
{
|
|
"entropy": 5.509807920455932,
|
|
"epoch": 0.05563810651680874,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.00027759175662104424,
|
|
"loss": 5.4031,
|
|
"mean_token_accuracy": 0.20948465317487716,
|
|
"num_tokens": 2107253.0,
|
|
"step": 2490
|
|
},
|
|
{
|
|
"entropy": 5.449338150024414,
|
|
"epoch": 0.05574982962226418,
|
|
"grad_norm": 1.3046875,
|
|
"learning_rate": 0.0002764137073922508,
|
|
"loss": 5.5747,
|
|
"mean_token_accuracy": 0.20230767428874968,
|
|
"num_tokens": 2111361.0,
|
|
"step": 2495
|
|
},
|
|
{
|
|
"entropy": 5.551156520843506,
|
|
"epoch": 0.05586155272771962,
|
|
"grad_norm": 1.3359375,
|
|
"learning_rate": 0.00027523561940595505,
|
|
"loss": 5.3984,
|
|
"mean_token_accuracy": 0.19894679635763168,
|
|
"num_tokens": 2115586.0,
|
|
"step": 2500
|
|
}
|
|
],
|
|
"logging_steps": 5,
|
|
"max_steps": 4000,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 1,
|
|
"save_steps": 500,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": false
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 3061705715712000.0,
|
|
"train_batch_size": 16,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|