| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.12373366328976877, |
| "eval_steps": 300, |
| "global_step": 50, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0024746732657953754, |
| "grad_norm": 11.35000991821289, |
| "learning_rate": 3.278688524590164e-06, |
| "logits/chosen": -0.6587307453155518, |
| "logits/rejected": -0.6631469130516052, |
| "logps/chosen": -13.55949878692627, |
| "logps/rejected": -27.670433044433594, |
| "loss": 1.6191432476043701, |
| "memory(GiB)": 45.42, |
| "nll_loss": 0.9259958863258362, |
| "rewards/accuracies": 0.0, |
| "rewards/chosen": 0.0, |
| "rewards/margins": 0.0, |
| "rewards/rejected": 0.0, |
| "step": 1, |
| "train_speed(iter/s)": 0.005221 |
| }, |
| { |
| "epoch": 0.004949346531590751, |
| "grad_norm": 9.617626190185547, |
| "learning_rate": 6.557377049180328e-06, |
| "logits/chosen": -0.6030034422874451, |
| "logits/rejected": -0.6114212274551392, |
| "logps/chosen": -14.923763275146484, |
| "logps/rejected": -29.194869995117188, |
| "loss": 1.6753138303756714, |
| "memory(GiB)": 45.42, |
| "nll_loss": 0.9821667671203613, |
| "rewards/accuracies": 0.0, |
| "rewards/chosen": 0.0, |
| "rewards/margins": 0.0, |
| "rewards/rejected": 0.0, |
| "step": 2, |
| "train_speed(iter/s)": 0.005275 |
| }, |
| { |
| "epoch": 0.007424019797386126, |
| "grad_norm": 8.504729270935059, |
| "learning_rate": 9.836065573770493e-06, |
| "logits/chosen": -0.5661746859550476, |
| "logits/rejected": -0.5726399421691895, |
| "logps/chosen": -13.571891784667969, |
| "logps/rejected": -26.323240280151367, |
| "loss": 1.6620099544525146, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.9678325057029724, |
| "rewards/accuracies": 0.53125, |
| "rewards/chosen": -0.003509162925183773, |
| "rewards/margins": -0.001862737350165844, |
| "rewards/rejected": -0.0016464253421872854, |
| "step": 3, |
| "train_speed(iter/s)": 0.005301 |
| }, |
| { |
| "epoch": 0.009898693063181502, |
| "grad_norm": 11.376845359802246, |
| "learning_rate": 1.3114754098360657e-05, |
| "logits/chosen": -0.6604664921760559, |
| "logits/rejected": -0.6652082800865173, |
| "logps/chosen": -12.837306022644043, |
| "logps/rejected": -23.095237731933594, |
| "loss": 1.663628339767456, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.975486695766449, |
| "rewards/accuracies": 0.6875, |
| "rewards/chosen": 0.00841558538377285, |
| "rewards/margins": 0.010128158144652843, |
| "rewards/rejected": -0.0017125748563557863, |
| "step": 4, |
| "train_speed(iter/s)": 0.005365 |
| }, |
| { |
| "epoch": 0.012373366328976877, |
| "grad_norm": 5.700821399688721, |
| "learning_rate": 1.6393442622950818e-05, |
| "logits/chosen": -0.6459366083145142, |
| "logits/rejected": -0.6513304710388184, |
| "logps/chosen": -18.565486907958984, |
| "logps/rejected": -23.259275436401367, |
| "loss": 1.594820261001587, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.8972694277763367, |
| "rewards/accuracies": 0.375, |
| "rewards/chosen": 0.00964329019188881, |
| "rewards/margins": -0.008526789955794811, |
| "rewards/rejected": 0.018170079216361046, |
| "step": 5, |
| "train_speed(iter/s)": 0.005387 |
| }, |
| { |
| "epoch": 0.014848039594772253, |
| "grad_norm": 8.467960357666016, |
| "learning_rate": 1.9672131147540985e-05, |
| "logits/chosen": -0.6038874387741089, |
| "logits/rejected": -0.6043834090232849, |
| "logps/chosen": -13.40754508972168, |
| "logps/rejected": -18.13410758972168, |
| "loss": 1.6557857990264893, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.960586428642273, |
| "rewards/accuracies": 0.46875, |
| "rewards/chosen": 0.030818985775113106, |
| "rewards/margins": -0.003026916179805994, |
| "rewards/rejected": 0.03384590521454811, |
| "step": 6, |
| "train_speed(iter/s)": 0.005393 |
| }, |
| { |
| "epoch": 0.017322712860567628, |
| "grad_norm": 10.094538688659668, |
| "learning_rate": 2.295081967213115e-05, |
| "logits/chosen": -0.5412906408309937, |
| "logits/rejected": -0.5506806969642639, |
| "logps/chosen": -12.647178649902344, |
| "logps/rejected": -27.220874786376953, |
| "loss": 1.5276265144348145, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.8624258637428284, |
| "rewards/accuracies": 0.71875, |
| "rewards/chosen": 0.08376072347164154, |
| "rewards/margins": 0.06035517156124115, |
| "rewards/rejected": 0.023405561223626137, |
| "step": 7, |
| "train_speed(iter/s)": 0.005386 |
| }, |
| { |
| "epoch": 0.019797386126363004, |
| "grad_norm": 8.840508460998535, |
| "learning_rate": 2.6229508196721314e-05, |
| "logits/chosen": -0.6647145748138428, |
| "logits/rejected": -0.6669148206710815, |
| "logps/chosen": -14.856410026550293, |
| "logps/rejected": -21.21141815185547, |
| "loss": 1.492944598197937, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.844528317451477, |
| "rewards/accuracies": 0.71875, |
| "rewards/chosen": 0.11737693846225739, |
| "rewards/margins": 0.09855899959802628, |
| "rewards/rejected": 0.01881793886423111, |
| "step": 8, |
| "train_speed(iter/s)": 0.005383 |
| }, |
| { |
| "epoch": 0.02227205939215838, |
| "grad_norm": 6.475947380065918, |
| "learning_rate": 2.9508196721311478e-05, |
| "logits/chosen": -0.6404256224632263, |
| "logits/rejected": -0.6432085037231445, |
| "logps/chosen": -12.697021484375, |
| "logps/rejected": -22.463302612304688, |
| "loss": 1.3507211208343506, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.6943754553794861, |
| "rewards/accuracies": 0.59375, |
| "rewards/chosen": 0.18904930353164673, |
| "rewards/margins": 0.11120466142892838, |
| "rewards/rejected": 0.07784464210271835, |
| "step": 9, |
| "train_speed(iter/s)": 0.005379 |
| }, |
| { |
| "epoch": 0.024746732657953754, |
| "grad_norm": 3.908445119857788, |
| "learning_rate": 3.2786885245901635e-05, |
| "logits/chosen": -0.6233164668083191, |
| "logits/rejected": -0.6281710863113403, |
| "logps/chosen": -12.109004974365234, |
| "logps/rejected": -25.513099670410156, |
| "loss": 1.1971133947372437, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.523026704788208, |
| "rewards/accuracies": 0.5625, |
| "rewards/chosen": 0.1465698629617691, |
| "rewards/margins": 0.09749479591846466, |
| "rewards/rejected": 0.04907506704330444, |
| "step": 10, |
| "train_speed(iter/s)": 0.005374 |
| }, |
| { |
| "epoch": 0.02722140592374913, |
| "grad_norm": 10.216607093811035, |
| "learning_rate": 3.6065573770491806e-05, |
| "logits/chosen": -0.6259695291519165, |
| "logits/rejected": -0.6286618709564209, |
| "logps/chosen": -14.0396146774292, |
| "logps/rejected": -21.92095947265625, |
| "loss": 1.4379217624664307, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.7931011915206909, |
| "rewards/accuracies": 0.65625, |
| "rewards/chosen": 0.10121297836303711, |
| "rewards/margins": 0.2612895369529724, |
| "rewards/rejected": -0.1600765436887741, |
| "step": 11, |
| "train_speed(iter/s)": 0.005372 |
| }, |
| { |
| "epoch": 0.029696079189544505, |
| "grad_norm": 6.620578765869141, |
| "learning_rate": 3.934426229508197e-05, |
| "logits/chosen": -0.6335112452507019, |
| "logits/rejected": -0.638087809085846, |
| "logps/chosen": -9.480979919433594, |
| "logps/rejected": -24.5159912109375, |
| "loss": 1.1956348419189453, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.5891170501708984, |
| "rewards/accuracies": 0.78125, |
| "rewards/chosen": 0.07576582580804825, |
| "rewards/margins": 0.3991776704788208, |
| "rewards/rejected": -0.32341182231903076, |
| "step": 12, |
| "train_speed(iter/s)": 0.005391 |
| }, |
| { |
| "epoch": 0.03217075245533988, |
| "grad_norm": 6.803226947784424, |
| "learning_rate": 4.262295081967213e-05, |
| "logits/chosen": -0.6281187534332275, |
| "logits/rejected": -0.6315115690231323, |
| "logps/chosen": -18.412860870361328, |
| "logps/rejected": -27.436920166015625, |
| "loss": 1.5400699377059937, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.88338702917099, |
| "rewards/accuracies": 0.625, |
| "rewards/chosen": -0.10771963745355606, |
| "rewards/margins": 0.27280330657958984, |
| "rewards/rejected": -0.3805229365825653, |
| "step": 13, |
| "train_speed(iter/s)": 0.005388 |
| }, |
| { |
| "epoch": 0.034645425721135256, |
| "grad_norm": 9.147095680236816, |
| "learning_rate": 4.59016393442623e-05, |
| "logits/chosen": -0.6376056671142578, |
| "logits/rejected": -0.6407235264778137, |
| "logps/chosen": -17.056198120117188, |
| "logps/rejected": -25.44281768798828, |
| "loss": 1.7076380252838135, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.973918616771698, |
| "rewards/accuracies": 0.59375, |
| "rewards/chosen": -0.14579348266124725, |
| "rewards/margins": 0.050162509083747864, |
| "rewards/rejected": -0.19595597684383392, |
| "step": 14, |
| "train_speed(iter/s)": 0.005397 |
| }, |
| { |
| "epoch": 0.03712009898693063, |
| "grad_norm": 4.252878189086914, |
| "learning_rate": 4.918032786885246e-05, |
| "logits/chosen": -0.5979167222976685, |
| "logits/rejected": -0.6013778448104858, |
| "logps/chosen": -12.705180168151855, |
| "logps/rejected": -26.116336822509766, |
| "loss": 1.3297263383865356, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.7305101156234741, |
| "rewards/accuracies": 0.6875, |
| "rewards/chosen": -0.06688955426216125, |
| "rewards/margins": 0.28163066506385803, |
| "rewards/rejected": -0.3485202193260193, |
| "step": 15, |
| "train_speed(iter/s)": 0.005401 |
| }, |
| { |
| "epoch": 0.03959477225272601, |
| "grad_norm": 3.8650903701782227, |
| "learning_rate": 5.245901639344263e-05, |
| "logits/chosen": -0.5886881351470947, |
| "logits/rejected": -0.5908396244049072, |
| "logps/chosen": -22.1767578125, |
| "logps/rejected": -32.695167541503906, |
| "loss": 1.450103521347046, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.7899419069290161, |
| "rewards/accuracies": 0.65625, |
| "rewards/chosen": -0.05856169015169144, |
| "rewards/margins": 0.17948633432388306, |
| "rewards/rejected": -0.2380480319261551, |
| "step": 16, |
| "train_speed(iter/s)": 0.005395 |
| }, |
| { |
| "epoch": 0.04206944551852138, |
| "grad_norm": 3.568068742752075, |
| "learning_rate": 5.5737704918032785e-05, |
| "logits/chosen": -0.557040810585022, |
| "logits/rejected": -0.5580084919929504, |
| "logps/chosen": -17.01874542236328, |
| "logps/rejected": -27.190898895263672, |
| "loss": 1.4369667768478394, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.7801130414009094, |
| "rewards/accuracies": 0.6875, |
| "rewards/chosen": 0.07399038225412369, |
| "rewards/margins": 0.22467321157455444, |
| "rewards/rejected": -0.15068282186985016, |
| "step": 17, |
| "train_speed(iter/s)": 0.005403 |
| }, |
| { |
| "epoch": 0.04454411878431676, |
| "grad_norm": 4.3334150314331055, |
| "learning_rate": 5.9016393442622956e-05, |
| "logits/chosen": -0.5880481600761414, |
| "logits/rejected": -0.5887515544891357, |
| "logps/chosen": -21.928936004638672, |
| "logps/rejected": -22.768033981323242, |
| "loss": 1.6926015615463257, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.9716537594795227, |
| "rewards/accuracies": 0.5, |
| "rewards/chosen": 0.0877460390329361, |
| "rewards/margins": 0.0405881404876709, |
| "rewards/rejected": 0.047157879918813705, |
| "step": 18, |
| "train_speed(iter/s)": 0.005402 |
| }, |
| { |
| "epoch": 0.04701879205011213, |
| "grad_norm": 2.5271899700164795, |
| "learning_rate": 6.229508196721313e-05, |
| "logits/chosen": -0.598438560962677, |
| "logits/rejected": -0.604390025138855, |
| "logps/chosen": -17.247812271118164, |
| "logps/rejected": -31.589086532592773, |
| "loss": 1.248651385307312, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.6379994750022888, |
| "rewards/accuracies": 0.5625, |
| "rewards/chosen": 0.2769991457462311, |
| "rewards/margins": 0.2692391276359558, |
| "rewards/rejected": 0.0077600013464689255, |
| "step": 19, |
| "train_speed(iter/s)": 0.005406 |
| }, |
| { |
| "epoch": 0.04949346531590751, |
| "grad_norm": 3.571847915649414, |
| "learning_rate": 6.557377049180327e-05, |
| "logits/chosen": -0.5842810273170471, |
| "logits/rejected": -0.5847360491752625, |
| "logps/chosen": -15.484521865844727, |
| "logps/rejected": -20.548080444335938, |
| "loss": 1.5169799327850342, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.7671993374824524, |
| "rewards/accuracies": 0.40625, |
| "rewards/chosen": 0.08881673216819763, |
| "rewards/margins": -0.04343637824058533, |
| "rewards/rejected": 0.13225311040878296, |
| "step": 20, |
| "train_speed(iter/s)": 0.00541 |
| }, |
| { |
| "epoch": 0.051968138581702884, |
| "grad_norm": 2.258408546447754, |
| "learning_rate": 6.885245901639344e-05, |
| "logits/chosen": -0.548245906829834, |
| "logits/rejected": -0.551019549369812, |
| "logps/chosen": -12.056645393371582, |
| "logps/rejected": -22.498403549194336, |
| "loss": 1.2092863321304321, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.5635381937026978, |
| "rewards/accuracies": 0.59375, |
| "rewards/chosen": 0.24224215745925903, |
| "rewards/margins": 0.19714026153087616, |
| "rewards/rejected": 0.045101895928382874, |
| "step": 21, |
| "train_speed(iter/s)": 0.005423 |
| }, |
| { |
| "epoch": 0.05444281184749826, |
| "grad_norm": 4.153398036956787, |
| "learning_rate": 7.213114754098361e-05, |
| "logits/chosen": -0.5542203783988953, |
| "logits/rejected": -0.5555762052536011, |
| "logps/chosen": -10.120831489562988, |
| "logps/rejected": -21.4542236328125, |
| "loss": 1.2580621242523193, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.6088542342185974, |
| "rewards/accuracies": 0.625, |
| "rewards/chosen": 0.2806166112422943, |
| "rewards/margins": 0.14948059618473053, |
| "rewards/rejected": 0.1311360001564026, |
| "step": 22, |
| "train_speed(iter/s)": 0.005422 |
| }, |
| { |
| "epoch": 0.056917485113293635, |
| "grad_norm": 2.784181833267212, |
| "learning_rate": 7.540983606557377e-05, |
| "logits/chosen": -0.6017516255378723, |
| "logits/rejected": -0.6026712656021118, |
| "logps/chosen": -14.456979751586914, |
| "logps/rejected": -21.297657012939453, |
| "loss": 1.2428232431411743, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.5972878932952881, |
| "rewards/accuracies": 0.53125, |
| "rewards/chosen": 0.27611616253852844, |
| "rewards/margins": 0.14373204112052917, |
| "rewards/rejected": 0.13238413631916046, |
| "step": 23, |
| "train_speed(iter/s)": 0.005422 |
| }, |
| { |
| "epoch": 0.05939215837908901, |
| "grad_norm": 2.2945520877838135, |
| "learning_rate": 7.868852459016394e-05, |
| "logits/chosen": -0.5793949961662292, |
| "logits/rejected": -0.5800157785415649, |
| "logps/chosen": -9.155426979064941, |
| "logps/rejected": -22.067913055419922, |
| "loss": 1.1942368745803833, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.5604769587516785, |
| "rewards/accuracies": 0.6875, |
| "rewards/chosen": 0.29127371311187744, |
| "rewards/margins": 0.17932376265525818, |
| "rewards/rejected": 0.11194998770952225, |
| "step": 24, |
| "train_speed(iter/s)": 0.005419 |
| }, |
| { |
| "epoch": 0.061866831644884386, |
| "grad_norm": 2.7884368896484375, |
| "learning_rate": 8.19672131147541e-05, |
| "logits/chosen": -0.5844717621803284, |
| "logits/rejected": -0.5868740677833557, |
| "logps/chosen": -13.537481307983398, |
| "logps/rejected": -17.04685401916504, |
| "loss": 1.218743920326233, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.6010019779205322, |
| "rewards/accuracies": 0.65625, |
| "rewards/chosen": 0.38273870944976807, |
| "rewards/margins": 0.2067541480064392, |
| "rewards/rejected": 0.17598456144332886, |
| "step": 25, |
| "train_speed(iter/s)": 0.005422 |
| }, |
| { |
| "epoch": 0.06434150491067976, |
| "grad_norm": 2.3340394496917725, |
| "learning_rate": 8.524590163934426e-05, |
| "logits/chosen": -0.6258128881454468, |
| "logits/rejected": -0.6279971599578857, |
| "logps/chosen": -9.657672882080078, |
| "logps/rejected": -22.88201904296875, |
| "loss": 1.114503264427185, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.5605608820915222, |
| "rewards/accuracies": 0.8125, |
| "rewards/chosen": 0.3312107026576996, |
| "rewards/margins": 0.372702419757843, |
| "rewards/rejected": -0.04149174690246582, |
| "step": 26, |
| "train_speed(iter/s)": 0.005423 |
| }, |
| { |
| "epoch": 0.06681617817647514, |
| "grad_norm": 3.664553165435791, |
| "learning_rate": 8.852459016393443e-05, |
| "logits/chosen": -0.5701621770858765, |
| "logits/rejected": -0.5702242851257324, |
| "logps/chosen": -15.183746337890625, |
| "logps/rejected": -17.559988021850586, |
| "loss": 1.4017894268035889, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.7635535597801208, |
| "rewards/accuracies": 0.65625, |
| "rewards/chosen": 0.21464091539382935, |
| "rewards/margins": 0.1838790327310562, |
| "rewards/rejected": 0.03076188825070858, |
| "step": 27, |
| "train_speed(iter/s)": 0.005417 |
| }, |
| { |
| "epoch": 0.06929085144227051, |
| "grad_norm": 2.2080814838409424, |
| "learning_rate": 9.18032786885246e-05, |
| "logits/chosen": -0.5039485096931458, |
| "logits/rejected": -0.5086482763290405, |
| "logps/chosen": -8.541333198547363, |
| "logps/rejected": -27.416786193847656, |
| "loss": 1.032238483428955, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.42585042119026184, |
| "rewards/accuracies": 0.6875, |
| "rewards/chosen": 0.29641830921173096, |
| "rewards/margins": 0.3008955717086792, |
| "rewards/rejected": -0.004477284848690033, |
| "step": 28, |
| "train_speed(iter/s)": 0.005423 |
| }, |
| { |
| "epoch": 0.0717655247080659, |
| "grad_norm": 2.4612784385681152, |
| "learning_rate": 9.508196721311476e-05, |
| "logits/chosen": -0.5983453989028931, |
| "logits/rejected": -0.6011134386062622, |
| "logps/chosen": -9.658693313598633, |
| "logps/rejected": -25.247703552246094, |
| "loss": 1.0381165742874146, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.46968621015548706, |
| "rewards/accuracies": 0.6875, |
| "rewards/chosen": 0.30931776762008667, |
| "rewards/margins": 0.36611291766166687, |
| "rewards/rejected": -0.05679512768983841, |
| "step": 29, |
| "train_speed(iter/s)": 0.005423 |
| }, |
| { |
| "epoch": 0.07424019797386126, |
| "grad_norm": 3.299984931945801, |
| "learning_rate": 9.836065573770493e-05, |
| "logits/chosen": -0.5566404461860657, |
| "logits/rejected": -0.5608186721801758, |
| "logps/chosen": -11.162762641906738, |
| "logps/rejected": -25.61810874938965, |
| "loss": 1.2033004760742188, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.6197662949562073, |
| "rewards/accuracies": 0.78125, |
| "rewards/chosen": 0.3209737241268158, |
| "rewards/margins": 0.38564494252204895, |
| "rewards/rejected": -0.06467118114233017, |
| "step": 30, |
| "train_speed(iter/s)": 0.005432 |
| }, |
| { |
| "epoch": 0.07671487123965665, |
| "grad_norm": 3.927424430847168, |
| "learning_rate": 0.00010163934426229508, |
| "logits/chosen": -0.5413897037506104, |
| "logits/rejected": -0.5447396636009216, |
| "logps/chosen": -7.564568042755127, |
| "logps/rejected": -23.63496208190918, |
| "loss": 1.1393965482711792, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.5696667432785034, |
| "rewards/accuracies": 0.71875, |
| "rewards/chosen": 0.27428141236305237, |
| "rewards/margins": 0.5005658864974976, |
| "rewards/rejected": -0.2262844443321228, |
| "step": 31, |
| "train_speed(iter/s)": 0.005428 |
| }, |
| { |
| "epoch": 0.07918954450545201, |
| "grad_norm": 3.8795111179351807, |
| "learning_rate": 0.00010491803278688525, |
| "logits/chosen": -0.6150721311569214, |
| "logits/rejected": -0.6170963048934937, |
| "logps/chosen": -15.14966869354248, |
| "logps/rejected": -21.37498664855957, |
| "loss": 1.2894086837768555, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.6919065713882446, |
| "rewards/accuracies": 0.65625, |
| "rewards/chosen": 0.2811221778392792, |
| "rewards/margins": 0.3919064402580261, |
| "rewards/rejected": -0.11078427731990814, |
| "step": 32, |
| "train_speed(iter/s)": 0.005435 |
| }, |
| { |
| "epoch": 0.0816642177712474, |
| "grad_norm": 3.8578360080718994, |
| "learning_rate": 0.00010819672131147543, |
| "logits/chosen": -0.6152044534683228, |
| "logits/rejected": -0.615663468837738, |
| "logps/chosen": -11.967881202697754, |
| "logps/rejected": -20.445411682128906, |
| "loss": 1.1714155673980713, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.5518181324005127, |
| "rewards/accuracies": 0.65625, |
| "rewards/chosen": 0.21441030502319336, |
| "rewards/margins": 0.2466525286436081, |
| "rewards/rejected": -0.03224220499396324, |
| "step": 33, |
| "train_speed(iter/s)": 0.005433 |
| }, |
| { |
| "epoch": 0.08413889103704277, |
| "grad_norm": 3.3592934608459473, |
| "learning_rate": 0.00011147540983606557, |
| "logits/chosen": -0.6198133826255798, |
| "logits/rejected": -0.6197149753570557, |
| "logps/chosen": -11.38640308380127, |
| "logps/rejected": -22.015522003173828, |
| "loss": 1.1110243797302246, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.5177969932556152, |
| "rewards/accuracies": 0.71875, |
| "rewards/chosen": 0.23841583728790283, |
| "rewards/margins": 0.32188504934310913, |
| "rewards/rejected": -0.0834691971540451, |
| "step": 34, |
| "train_speed(iter/s)": 0.00543 |
| }, |
| { |
| "epoch": 0.08661356430283815, |
| "grad_norm": 2.9654669761657715, |
| "learning_rate": 0.00011475409836065574, |
| "logits/chosen": -0.6268936991691589, |
| "logits/rejected": -0.6304693818092346, |
| "logps/chosen": -8.792684555053711, |
| "logps/rejected": -27.867855072021484, |
| "loss": 0.9397363662719727, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.44072026014328003, |
| "rewards/accuracies": 0.75, |
| "rewards/chosen": 0.2782461643218994, |
| "rewards/margins": 0.6700869202613831, |
| "rewards/rejected": -0.39184072613716125, |
| "step": 35, |
| "train_speed(iter/s)": 0.00543 |
| }, |
| { |
| "epoch": 0.08908823756863352, |
| "grad_norm": 4.219470977783203, |
| "learning_rate": 0.00011803278688524591, |
| "logits/chosen": -0.5896088480949402, |
| "logits/rejected": -0.5908181667327881, |
| "logps/chosen": -9.439773559570312, |
| "logps/rejected": -21.788042068481445, |
| "loss": 1.1120812892913818, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.5399542450904846, |
| "rewards/accuracies": 0.625, |
| "rewards/chosen": 0.19946175813674927, |
| "rewards/margins": 0.4053455591201782, |
| "rewards/rejected": -0.20588380098342896, |
| "step": 36, |
| "train_speed(iter/s)": 0.005431 |
| }, |
| { |
| "epoch": 0.0915629108344289, |
| "grad_norm": 4.5739665031433105, |
| "learning_rate": 0.00012131147540983607, |
| "logits/chosen": -0.6280671954154968, |
| "logits/rejected": -0.6325269937515259, |
| "logps/chosen": -8.791489601135254, |
| "logps/rejected": -30.336650848388672, |
| "loss": 1.1410616636276245, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.6490904688835144, |
| "rewards/accuracies": 0.78125, |
| "rewards/chosen": 0.2655602991580963, |
| "rewards/margins": 0.6452851891517639, |
| "rewards/rejected": -0.3797248899936676, |
| "step": 37, |
| "train_speed(iter/s)": 0.005434 |
| }, |
| { |
| "epoch": 0.09403758410022427, |
| "grad_norm": 3.020256280899048, |
| "learning_rate": 0.00012459016393442625, |
| "logits/chosen": -0.6382966637611389, |
| "logits/rejected": -0.6446542143821716, |
| "logps/chosen": -12.288652420043945, |
| "logps/rejected": -31.03972625732422, |
| "loss": 0.9547407031059265, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.574475109577179, |
| "rewards/accuracies": 0.875, |
| "rewards/chosen": 0.3463619351387024, |
| "rewards/margins": 1.0072863101959229, |
| "rewards/rejected": -0.6609243154525757, |
| "step": 38, |
| "train_speed(iter/s)": 0.005438 |
| }, |
| { |
| "epoch": 0.09651225736601965, |
| "grad_norm": 3.258894205093384, |
| "learning_rate": 0.0001278688524590164, |
| "logits/chosen": -0.6787834763526917, |
| "logits/rejected": -0.6786651015281677, |
| "logps/chosen": -12.136661529541016, |
| "logps/rejected": -20.544681549072266, |
| "loss": 1.0752589702606201, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.6294119358062744, |
| "rewards/accuracies": 0.875, |
| "rewards/chosen": 0.34082022309303284, |
| "rewards/margins": 0.8202961683273315, |
| "rewards/rejected": -0.4794759452342987, |
| "step": 39, |
| "train_speed(iter/s)": 0.005436 |
| }, |
| { |
| "epoch": 0.09898693063181502, |
| "grad_norm": 3.8263096809387207, |
| "learning_rate": 0.00013114754098360654, |
| "logits/chosen": -0.6569682359695435, |
| "logits/rejected": -0.6592159271240234, |
| "logps/chosen": -21.227203369140625, |
| "logps/rejected": -24.57542610168457, |
| "loss": 1.3819258213043213, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.8235953450202942, |
| "rewards/accuracies": 0.75, |
| "rewards/chosen": 0.12423999607563019, |
| "rewards/margins": 0.5565845966339111, |
| "rewards/rejected": -0.43234458565711975, |
| "step": 40, |
| "train_speed(iter/s)": 0.005435 |
| }, |
| { |
| "epoch": 0.1014616038976104, |
| "grad_norm": 3.6862781047821045, |
| "learning_rate": 0.00013442622950819673, |
| "logits/chosen": -0.6126083135604858, |
| "logits/rejected": -0.618023157119751, |
| "logps/chosen": -18.38534927368164, |
| "logps/rejected": -31.6680850982666, |
| "loss": 1.0961635112762451, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.6294427514076233, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": 0.16896528005599976, |
| "rewards/margins": 0.7662918567657471, |
| "rewards/rejected": -0.5973266363143921, |
| "step": 41, |
| "train_speed(iter/s)": 0.005439 |
| }, |
| { |
| "epoch": 0.10393627716340577, |
| "grad_norm": 2.924959897994995, |
| "learning_rate": 0.00013770491803278688, |
| "logits/chosen": -0.6756088137626648, |
| "logits/rejected": -0.6791292428970337, |
| "logps/chosen": -12.787935256958008, |
| "logps/rejected": -35.33570861816406, |
| "loss": 0.9558172821998596, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.45665955543518066, |
| "rewards/accuracies": 0.71875, |
| "rewards/chosen": 0.14984729886054993, |
| "rewards/margins": 0.7973157167434692, |
| "rewards/rejected": -0.6474683880805969, |
| "step": 42, |
| "train_speed(iter/s)": 0.005438 |
| }, |
| { |
| "epoch": 0.10641095042920115, |
| "grad_norm": 5.650424480438232, |
| "learning_rate": 0.00014098360655737707, |
| "logits/chosen": -0.6601684093475342, |
| "logits/rejected": -0.667102038860321, |
| "logps/chosen": -12.86988639831543, |
| "logps/rejected": -33.79030227661133, |
| "loss": 0.97264564037323, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.5231238603591919, |
| "rewards/accuracies": 0.78125, |
| "rewards/chosen": 0.10155264288187027, |
| "rewards/margins": 0.8151469230651855, |
| "rewards/rejected": -0.7135943174362183, |
| "step": 43, |
| "train_speed(iter/s)": 0.005438 |
| }, |
| { |
| "epoch": 0.10888562369499652, |
| "grad_norm": 4.125715732574463, |
| "learning_rate": 0.00014426229508196722, |
| "logits/chosen": -0.615486741065979, |
| "logits/rejected": -0.6166262030601501, |
| "logps/chosen": -22.158443450927734, |
| "logps/rejected": -30.26171112060547, |
| "loss": 1.174919843673706, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.67107754945755, |
| "rewards/accuracies": 0.75, |
| "rewards/chosen": 0.12467168271541595, |
| "rewards/margins": 1.025106430053711, |
| "rewards/rejected": -0.9004346132278442, |
| "step": 44, |
| "train_speed(iter/s)": 0.005437 |
| }, |
| { |
| "epoch": 0.1113602969607919, |
| "grad_norm": 4.247290134429932, |
| "learning_rate": 0.00014754098360655738, |
| "logits/chosen": -0.6652258634567261, |
| "logits/rejected": -0.672545313835144, |
| "logps/chosen": -9.705460548400879, |
| "logps/rejected": -31.59243392944336, |
| "loss": 0.9555923342704773, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.5207614302635193, |
| "rewards/accuracies": 0.75, |
| "rewards/chosen": 0.2395530790090561, |
| "rewards/margins": 1.0814993381500244, |
| "rewards/rejected": -0.8419461846351624, |
| "step": 45, |
| "train_speed(iter/s)": 0.005438 |
| }, |
| { |
| "epoch": 0.11383497022658727, |
| "grad_norm": 3.2860352993011475, |
| "learning_rate": 0.00015081967213114754, |
| "logits/chosen": -0.614319920539856, |
| "logits/rejected": -0.619287371635437, |
| "logps/chosen": -13.63390064239502, |
| "logps/rejected": -35.74215316772461, |
| "loss": 0.9872268438339233, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.6003797650337219, |
| "rewards/accuracies": 0.8125, |
| "rewards/chosen": 0.28653809428215027, |
| "rewards/margins": 1.200305461883545, |
| "rewards/rejected": -0.9137674570083618, |
| "step": 46, |
| "train_speed(iter/s)": 0.005439 |
| }, |
| { |
| "epoch": 0.11630964349238265, |
| "grad_norm": 5.599460601806641, |
| "learning_rate": 0.0001540983606557377, |
| "logits/chosen": -0.6212424039840698, |
| "logits/rejected": -0.6217514276504517, |
| "logps/chosen": -17.058250427246094, |
| "logps/rejected": -29.002065658569336, |
| "loss": 1.1983596086502075, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.7794193029403687, |
| "rewards/accuracies": 0.8125, |
| "rewards/chosen": 0.22204741835594177, |
| "rewards/margins": 0.9237122535705566, |
| "rewards/rejected": -0.7016648054122925, |
| "step": 47, |
| "train_speed(iter/s)": 0.005439 |
| }, |
| { |
| "epoch": 0.11878431675817802, |
| "grad_norm": 4.221462726593018, |
| "learning_rate": 0.00015737704918032788, |
| "logits/chosen": -0.6427355408668518, |
| "logits/rejected": -0.6452814936637878, |
| "logps/chosen": -13.284379005432129, |
| "logps/rejected": -28.47121238708496, |
| "loss": 0.9505041241645813, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.5451131463050842, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": 0.22931885719299316, |
| "rewards/margins": 0.9360889792442322, |
| "rewards/rejected": -0.706770122051239, |
| "step": 48, |
| "train_speed(iter/s)": 0.005439 |
| }, |
| { |
| "epoch": 0.1212589900239734, |
| "grad_norm": 4.555298328399658, |
| "learning_rate": 0.00016065573770491804, |
| "logits/chosen": -0.6409769058227539, |
| "logits/rejected": -0.6388667225837708, |
| "logps/chosen": -12.98600959777832, |
| "logps/rejected": -19.881681442260742, |
| "loss": 1.2290067672729492, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.5956090688705444, |
| "rewards/accuracies": 0.59375, |
| "rewards/chosen": 0.07330272346735, |
| "rewards/margins": 0.5512866973876953, |
| "rewards/rejected": -0.47798389196395874, |
| "step": 49, |
| "train_speed(iter/s)": 0.00544 |
| }, |
| { |
| "epoch": 0.12373366328976877, |
| "grad_norm": 3.2366769313812256, |
| "learning_rate": 0.0001639344262295082, |
| "logits/chosen": -0.6371148228645325, |
| "logits/rejected": -0.6468245983123779, |
| "logps/chosen": -20.83090591430664, |
| "logps/rejected": -45.217796325683594, |
| "loss": 0.996230959892273, |
| "memory(GiB)": 46.09, |
| "nll_loss": 0.6383583545684814, |
| "rewards/accuracies": 0.75, |
| "rewards/chosen": 0.13887403905391693, |
| "rewards/margins": 1.3918437957763672, |
| "rewards/rejected": -1.252969741821289, |
| "step": 50, |
| "train_speed(iter/s)": 0.005439 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 1212, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 50, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 3.492565635857121e+17, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
