| [ | |
| { | |
| "loss": 0.6476, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.02, | |
| "step": 1 | |
| }, | |
| { | |
| "loss": 0.6668, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.04, | |
| "step": 2 | |
| }, | |
| { | |
| "loss": 0.6615, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.05, | |
| "step": 3 | |
| }, | |
| { | |
| "loss": 0.5895, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.07, | |
| "step": 4 | |
| }, | |
| { | |
| "loss": 0.6124, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.09, | |
| "step": 5 | |
| }, | |
| { | |
| "loss": 0.5953, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.11, | |
| "step": 6 | |
| }, | |
| { | |
| "loss": 0.5356, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.12, | |
| "step": 7 | |
| }, | |
| { | |
| "loss": 0.4604, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.14, | |
| "step": 8 | |
| }, | |
| { | |
| "loss": 0.4143, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.16, | |
| "step": 9 | |
| }, | |
| { | |
| "loss": 0.4553, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.18, | |
| "step": 10 | |
| }, | |
| { | |
| "loss": 0.3538, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.2, | |
| "step": 11 | |
| }, | |
| { | |
| "loss": 0.3389, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.21, | |
| "step": 12 | |
| }, | |
| { | |
| "loss": 0.2639, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.23, | |
| "step": 13 | |
| }, | |
| { | |
| "loss": 0.2096, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.25, | |
| "step": 14 | |
| }, | |
| { | |
| "loss": 0.2002, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.27, | |
| "step": 15 | |
| }, | |
| { | |
| "loss": 0.2374, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.29, | |
| "step": 16 | |
| }, | |
| { | |
| "loss": 0.2475, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.3, | |
| "step": 17 | |
| }, | |
| { | |
| "loss": 0.1994, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.32, | |
| "step": 18 | |
| }, | |
| { | |
| "loss": 0.1406, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.34, | |
| "step": 19 | |
| }, | |
| { | |
| "loss": 0.055, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.36, | |
| "step": 20 | |
| }, | |
| { | |
| "loss": 0.1168, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.38, | |
| "step": 21 | |
| }, | |
| { | |
| "loss": 0.0916, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.39, | |
| "step": 22 | |
| }, | |
| { | |
| "loss": 0.1596, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.41, | |
| "step": 23 | |
| }, | |
| { | |
| "loss": 0.1585, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.43, | |
| "step": 24 | |
| }, | |
| { | |
| "loss": 0.184, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.45, | |
| "step": 25 | |
| }, | |
| { | |
| "eval_math_exam_questions_loss": 0.09979354590177536, | |
| "eval_math_exam_questions_score": -0.029867494478821754, | |
| "eval_math_exam_questions_brier_score": 0.029867494478821754, | |
| "eval_math_exam_questions_average_probability": 0.9305303692817688, | |
| "eval_math_exam_questions_accuracy": 0.95, | |
| "eval_math_exam_questions_probabilities": [ | |
| 0.8511927723884583, | |
| 0.8560279011726379, | |
| 0.9999574422836304, | |
| 0.9483713507652283, | |
| 0.2606664001941681, | |
| 0.9442870616912842, | |
| 0.9947615265846252, | |
| 0.9007341265678406, | |
| 0.9449180364608765, | |
| 0.41139501333236694, | |
| 0.9922882318496704, | |
| 0.9992316961288452, | |
| 0.9993873834609985, | |
| 0.9968752861022949, | |
| 0.9976814985275269, | |
| 0.9999994039535522, | |
| 0.9999996423721313, | |
| 0.9999994039535522, | |
| 0.9571589231491089, | |
| 0.995111882686615, | |
| 0.979494571685791, | |
| 0.574057400226593, | |
| 0.7285555601119995, | |
| 0.3839860260486603, | |
| 0.9998674392700195, | |
| 0.9982432126998901, | |
| 0.9999611377716064, | |
| 0.9999970197677612, | |
| 0.9999997615814209, | |
| 1.0, | |
| 0.9638392329216003, | |
| 0.968817949295044, | |
| 0.2049836367368698, | |
| 0.9999809265136719, | |
| 0.9998273253440857, | |
| 0.9999629259109497, | |
| 0.9595451951026917, | |
| 0.9992743134498596, | |
| 0.9850507974624634, | |
| 0.9250211715698242, | |
| 0.840378999710083, | |
| 0.8428780436515808, | |
| 0.9994524121284485, | |
| 0.9991476535797119, | |
| 0.9997357726097107, | |
| 0.9995108842849731, | |
| 0.9998007416725159, | |
| 0.9996703863143921, | |
| 0.9999984502792358, | |
| 0.9999967813491821, | |
| 0.999998927116394, | |
| 0.9999979734420776, | |
| 0.9999997615814209, | |
| 0.9999996423721313, | |
| 0.9957982897758484, | |
| 0.9765607714653015, | |
| 0.996933102607727, | |
| 0.8950393795967102, | |
| 0.9991758465766907, | |
| 0.9990474581718445, | |
| 0.7260539531707764, | |
| 0.8293086290359497, | |
| 0.381984144449234, | |
| 0.9996906518936157, | |
| 0.9999486207962036, | |
| 0.9999747276306152, | |
| 0.994877815246582, | |
| 0.991081714630127, | |
| 0.9968804121017456, | |
| 0.9998242259025574, | |
| 0.9999198913574219, | |
| 0.999997615814209, | |
| 0.9805970788002014, | |
| 0.9670814871788025, | |
| 0.8692526817321777, | |
| 0.9998607635498047, | |
| 0.9999896287918091, | |
| 0.999993085861206, | |
| 0.9961829781532288, | |
| 0.9215685129165649, | |
| 0.998610258102417, | |
| 0.9952474236488342, | |
| 0.7989624738693237, | |
| 0.9947852492332458, | |
| 0.9998492002487183, | |
| 0.9999179840087891, | |
| 0.9999663829803467, | |
| 0.9998860359191895, | |
| 0.9998941421508789, | |
| 0.9978526830673218, | |
| 0.8546462059020996, | |
| 0.8965560793876648, | |
| 0.6786884069442749, | |
| 0.9975112676620483, | |
| 0.9996652603149414, | |
| 0.9990561604499817, | |
| 0.9986518025398254, | |
| 0.9981924891471863, | |
| 0.998735249042511, | |
| 0.9286666512489319 | |
| ], | |
| "eval_math_exam_questions_runtime": 58.7461, | |
| "eval_math_exam_questions_samples_per_second": 1.702, | |
| "eval_math_exam_questions_steps_per_second": 0.068, | |
| "epoch": 0.45, | |
| "step": 25 | |
| }, | |
| { | |
| "loss": 0.0599, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.46, | |
| "step": 26 | |
| }, | |
| { | |
| "loss": 0.0445, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.48, | |
| "step": 27 | |
| }, | |
| { | |
| "loss": 0.0625, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.5, | |
| "step": 28 | |
| }, | |
| { | |
| "loss": 0.0125, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.52, | |
| "step": 29 | |
| }, | |
| { | |
| "loss": 0.2316, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.54, | |
| "step": 30 | |
| }, | |
| { | |
| "loss": 0.0746, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.55, | |
| "step": 31 | |
| }, | |
| { | |
| "loss": 0.1032, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.57, | |
| "step": 32 | |
| }, | |
| { | |
| "loss": 0.1413, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.59, | |
| "step": 33 | |
| }, | |
| { | |
| "loss": 0.1345, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.61, | |
| "step": 34 | |
| }, | |
| { | |
| "loss": 0.0688, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.62, | |
| "step": 35 | |
| }, | |
| { | |
| "loss": 0.0762, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.64, | |
| "step": 36 | |
| }, | |
| { | |
| "loss": 0.0552, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.66, | |
| "step": 37 | |
| }, | |
| { | |
| "loss": 0.0356, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.68, | |
| "step": 38 | |
| }, | |
| { | |
| "loss": 0.0543, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.7, | |
| "step": 39 | |
| }, | |
| { | |
| "loss": 0.0279, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.71, | |
| "step": 40 | |
| }, | |
| { | |
| "loss": 0.0704, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.73, | |
| "step": 41 | |
| }, | |
| { | |
| "loss": 0.039, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.75, | |
| "step": 42 | |
| }, | |
| { | |
| "loss": 0.0114, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.77, | |
| "step": 43 | |
| }, | |
| { | |
| "loss": 0.0043, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.79, | |
| "step": 44 | |
| }, | |
| { | |
| "loss": 0.1562, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.8, | |
| "step": 45 | |
| }, | |
| { | |
| "loss": 0.0495, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.82, | |
| "step": 46 | |
| }, | |
| { | |
| "loss": 0.1683, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.84, | |
| "step": 47 | |
| }, | |
| { | |
| "loss": 0.0118, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.86, | |
| "step": 48 | |
| }, | |
| { | |
| "loss": 0.0708, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.88, | |
| "step": 49 | |
| }, | |
| { | |
| "loss": 0.0445, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.89, | |
| "step": 50 | |
| }, | |
| { | |
| "eval_math_exam_questions_loss": 0.11060654371976852, | |
| "eval_math_exam_questions_score": -0.04164460673928261, | |
| "eval_math_exam_questions_brier_score": 0.04164460673928261, | |
| "eval_math_exam_questions_average_probability": 0.9377254247665405, | |
| "eval_math_exam_questions_accuracy": 0.92, | |
| "eval_math_exam_questions_probabilities": [ | |
| 0.9878444075584412, | |
| 0.9649227261543274, | |
| 0.9999626874923706, | |
| 0.9997627139091492, | |
| 0.41481563448905945, | |
| 0.9997966885566711, | |
| 0.9999992847442627, | |
| 0.9999576807022095, | |
| 0.9999995231628418, | |
| 0.13421274721622467, | |
| 0.9997405409812927, | |
| 0.9999984502792358, | |
| 0.9999998807907104, | |
| 0.9999960660934448, | |
| 0.9999959468841553, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999942779541016, | |
| 0.9999998807907104, | |
| 0.9999129772186279, | |
| 0.4325127899646759, | |
| 0.19135379791259766, | |
| 0.2162252813577652, | |
| 1.0, | |
| 0.9999597072601318, | |
| 1.0, | |
| 0.9999998807907104, | |
| 1.0, | |
| 1.0, | |
| 0.9999549388885498, | |
| 0.9999803304672241, | |
| 0.9946361184120178, | |
| 0.9999997615814209, | |
| 0.9998887777328491, | |
| 0.9999957084655762, | |
| 0.9995691180229187, | |
| 1.0, | |
| 0.99983811378479, | |
| 0.9979630708694458, | |
| 0.9983990788459778, | |
| 0.9632781147956848, | |
| 0.9999992847442627, | |
| 0.9999983310699463, | |
| 0.9999996423721313, | |
| 0.9999991655349731, | |
| 0.999947190284729, | |
| 0.9998934268951416, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999369382858276, | |
| 0.9989650249481201, | |
| 0.9998283386230469, | |
| 0.9974669218063354, | |
| 1.0, | |
| 1.0, | |
| 0.7936062216758728, | |
| 0.9307928085327148, | |
| 0.27603790163993835, | |
| 0.9999816417694092, | |
| 0.9999940395355225, | |
| 0.9999995231628418, | |
| 0.9996922016143799, | |
| 0.988324761390686, | |
| 0.9975664615631104, | |
| 0.9999992847442627, | |
| 0.9999995231628418, | |
| 1.0, | |
| 0.9978724718093872, | |
| 0.3203738033771515, | |
| 0.3396102786064148, | |
| 0.9999998807907104, | |
| 1.0, | |
| 1.0, | |
| 0.9998663663864136, | |
| 0.988074004650116, | |
| 0.9999923706054688, | |
| 0.9999964237213135, | |
| 0.9248444437980652, | |
| 0.9999958276748657, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999954700469971, | |
| 0.9937870502471924, | |
| 0.9950788021087646, | |
| 0.9376111030578613, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999998807907104, | |
| 0.9999986886978149, | |
| 0.9999997615814209, | |
| 0.9989570379257202 | |
| ], | |
| "eval_math_exam_questions_runtime": 58.7331, | |
| "eval_math_exam_questions_samples_per_second": 1.703, | |
| "eval_math_exam_questions_steps_per_second": 0.068, | |
| "epoch": 0.89, | |
| "step": 50 | |
| }, | |
| { | |
| "loss": 0.1862, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.91, | |
| "step": 51 | |
| }, | |
| { | |
| "loss": 0.1376, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.93, | |
| "step": 52 | |
| }, | |
| { | |
| "loss": 0.1167, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.95, | |
| "step": 53 | |
| }, | |
| { | |
| "loss": 0.1163, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.96, | |
| "step": 54 | |
| }, | |
| { | |
| "loss": 0.0308, | |
| "learning_rate": 0.0002, | |
| "epoch": 0.98, | |
| "step": 55 | |
| }, | |
| { | |
| "loss": 0.0515, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.0, | |
| "step": 56 | |
| }, | |
| { | |
| "loss": 0.0876, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.02, | |
| "step": 57 | |
| }, | |
| { | |
| "loss": 0.0215, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.04, | |
| "step": 58 | |
| }, | |
| { | |
| "loss": 0.0454, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.05, | |
| "step": 59 | |
| }, | |
| { | |
| "loss": 0.0335, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.07, | |
| "step": 60 | |
| }, | |
| { | |
| "loss": 0.0248, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.09, | |
| "step": 61 | |
| }, | |
| { | |
| "loss": 0.0028, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.11, | |
| "step": 62 | |
| }, | |
| { | |
| "loss": 0.0175, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.12, | |
| "step": 63 | |
| }, | |
| { | |
| "loss": 0.0078, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.14, | |
| "step": 64 | |
| }, | |
| { | |
| "loss": 0.0113, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.16, | |
| "step": 65 | |
| }, | |
| { | |
| "loss": 0.0119, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.18, | |
| "step": 66 | |
| }, | |
| { | |
| "loss": 0.0235, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.2, | |
| "step": 67 | |
| }, | |
| { | |
| "loss": 0.0116, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.21, | |
| "step": 68 | |
| }, | |
| { | |
| "loss": 0.0018, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.23, | |
| "step": 69 | |
| }, | |
| { | |
| "loss": 0.0024, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.25, | |
| "step": 70 | |
| }, | |
| { | |
| "loss": 0.003, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.27, | |
| "step": 71 | |
| }, | |
| { | |
| "loss": 0.0044, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.29, | |
| "step": 72 | |
| }, | |
| { | |
| "loss": 0.0151, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.3, | |
| "step": 73 | |
| }, | |
| { | |
| "loss": 0.0019, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.32, | |
| "step": 74 | |
| }, | |
| { | |
| "loss": 0.0003, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.34, | |
| "step": 75 | |
| }, | |
| { | |
| "eval_math_exam_questions_loss": 0.05752657726407051, | |
| "eval_math_exam_questions_score": -0.019282517954707146, | |
| "eval_math_exam_questions_brier_score": 0.019282517954707146, | |
| "eval_math_exam_questions_average_probability": 0.9707407355308533, | |
| "eval_math_exam_questions_accuracy": 0.97, | |
| "eval_math_exam_questions_probabilities": [ | |
| 0.9951311349868774, | |
| 0.8724949359893799, | |
| 0.9999651908874512, | |
| 1.0, | |
| 0.8811224102973938, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.103812575340271, | |
| 0.9999701976776123, | |
| 0.9999998807907104, | |
| 1.0, | |
| 0.9999998807907104, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999974966049194, | |
| 0.9944435954093933, | |
| 0.9979708790779114, | |
| 0.4370083212852478, | |
| 1.0, | |
| 0.9999988079071045, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999998807907104, | |
| 1.0, | |
| 0.14551357924938202, | |
| 1.0, | |
| 0.999981164932251, | |
| 0.9999678134918213, | |
| 0.9999961853027344, | |
| 0.9987561702728271, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999998807907104, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999978542327881, | |
| 1.0, | |
| 0.999945878982544, | |
| 1.0, | |
| 1.0, | |
| 0.8369179964065552, | |
| 0.9991326928138733, | |
| 0.8580266237258911, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999998807907104, | |
| 0.9998944997787476, | |
| 0.9997386336326599, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999858140945435, | |
| 0.9989979863166809, | |
| 0.9989114999771118, | |
| 1.0, | |
| 1.0, | |
| 0.9999665021896362, | |
| 0.9914140701293945, | |
| 0.9999996423721313, | |
| 1.0, | |
| 0.9750034809112549, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999885559082031, | |
| 0.999998927116394, | |
| 0.9905345439910889, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9994863271713257 | |
| ], | |
| "eval_math_exam_questions_runtime": 58.6753, | |
| "eval_math_exam_questions_samples_per_second": 1.704, | |
| "eval_math_exam_questions_steps_per_second": 0.068, | |
| "epoch": 1.34, | |
| "step": 75 | |
| }, | |
| { | |
| "loss": 0.0139, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.36, | |
| "step": 76 | |
| }, | |
| { | |
| "loss": 0.0121, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.38, | |
| "step": 77 | |
| }, | |
| { | |
| "loss": 0.0001, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.39, | |
| "step": 78 | |
| }, | |
| { | |
| "loss": 0.0117, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.41, | |
| "step": 79 | |
| }, | |
| { | |
| "loss": 0.0046, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.43, | |
| "step": 80 | |
| }, | |
| { | |
| "loss": 0.3396, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.45, | |
| "step": 81 | |
| }, | |
| { | |
| "loss": 0.0, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.46, | |
| "step": 82 | |
| }, | |
| { | |
| "loss": 0.0014, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.48, | |
| "step": 83 | |
| }, | |
| { | |
| "loss": 0.0034, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.5, | |
| "step": 84 | |
| }, | |
| { | |
| "loss": 0.0026, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.52, | |
| "step": 85 | |
| }, | |
| { | |
| "loss": 0.001, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.54, | |
| "step": 86 | |
| }, | |
| { | |
| "loss": 0.0026, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.55, | |
| "step": 87 | |
| }, | |
| { | |
| "loss": 0.006, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.57, | |
| "step": 88 | |
| }, | |
| { | |
| "loss": 0.0001, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.59, | |
| "step": 89 | |
| }, | |
| { | |
| "loss": 0.0024, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.61, | |
| "step": 90 | |
| }, | |
| { | |
| "loss": 0.0031, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.62, | |
| "step": 91 | |
| }, | |
| { | |
| "loss": 0.0005, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.64, | |
| "step": 92 | |
| }, | |
| { | |
| "loss": 0.006, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.66, | |
| "step": 93 | |
| }, | |
| { | |
| "loss": 0.0056, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.68, | |
| "step": 94 | |
| }, | |
| { | |
| "loss": 0.0003, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.7, | |
| "step": 95 | |
| }, | |
| { | |
| "loss": 0.007, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.71, | |
| "step": 96 | |
| }, | |
| { | |
| "loss": 0.1818, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.73, | |
| "step": 97 | |
| }, | |
| { | |
| "loss": 0.0089, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.75, | |
| "step": 98 | |
| }, | |
| { | |
| "loss": 0.0022, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.77, | |
| "step": 99 | |
| }, | |
| { | |
| "loss": 0.0025, | |
| "learning_rate": 0.0002, | |
| "epoch": 1.79, | |
| "step": 100 | |
| }, | |
| { | |
| "eval_math_exam_questions_loss": 0.214411199092865, | |
| "eval_math_exam_questions_score": -0.03251197189092636, | |
| "eval_math_exam_questions_brier_score": 0.03251197189092636, | |
| "eval_math_exam_questions_average_probability": 0.9538560509681702, | |
| "eval_math_exam_questions_accuracy": 0.96, | |
| "eval_math_exam_questions_probabilities": [ | |
| 0.9958956241607666, | |
| 0.8318325877189636, | |
| 0.9999992847442627, | |
| 1.0, | |
| 8.109304872050416e-06, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.29970213770866394, | |
| 0.9999974966049194, | |
| 1.0, | |
| 1.0, | |
| 0.9999998807907104, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999997615814209, | |
| 1.0, | |
| 0.999977707862854, | |
| 0.999778687953949, | |
| 0.9999997615814209, | |
| 0.30764925479888916, | |
| 1.0, | |
| 0.9999996423721313, | |
| 1.0, | |
| 0.9999992847442627, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.6386927366256714, | |
| 1.0, | |
| 0.9961808919906616, | |
| 0.9999927282333374, | |
| 0.9999998807907104, | |
| 0.8549012541770935, | |
| 0.9999998807907104, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999997615814209, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999995231628418, | |
| 1.0, | |
| 0.9937735199928284, | |
| 1.0, | |
| 1.0, | |
| 0.9984048008918762, | |
| 1.0, | |
| 0.9988483190536499, | |
| 0.9999972581863403, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999980926513672, | |
| 0.9998224377632141, | |
| 0.9999912977218628, | |
| 1.0, | |
| 1.0, | |
| 0.7528015971183777, | |
| 0.003164754481986165, | |
| 0.9934700727462769, | |
| 0.9953631162643433, | |
| 1.0, | |
| 1.0, | |
| 0.9982792139053345, | |
| 0.9575594067573547, | |
| 0.9999983310699463, | |
| 1.0, | |
| 0.7891086935997009, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9999996423721313, | |
| 0.9999997615814209, | |
| 0.9999990463256836, | |
| 0.9808980822563171, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 1.0, | |
| 0.9995265007019043 | |
| ], | |
| "eval_math_exam_questions_runtime": 58.6781, | |
| "eval_math_exam_questions_samples_per_second": 1.704, | |
| "eval_math_exam_questions_steps_per_second": 0.068, | |
| "epoch": 1.79, | |
| "step": 100 | |
| }, | |
| { | |
| "train_runtime": 4114.901, | |
| "train_samples_per_second": 0.778, | |
| "train_steps_per_second": 0.024, | |
| "total_flos": 0.0, | |
| "train_loss": 0.11991490689004422, | |
| "epoch": 1.79, | |
| "step": 100 | |
| } | |
| ]] |