| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 100, | |
| "global_step": 385, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 1312.0, | |
| "learning_rate": 1.282051282051282e-07, | |
| "logits/chosen": 88.18099975585938, | |
| "logits/rejected": 88.25153350830078, | |
| "logps/chosen": -29.073104858398438, | |
| "logps/rejected": -26.25731658935547, | |
| "loss": 25.0, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 1168.0, | |
| "learning_rate": 1.282051282051282e-06, | |
| "logits/chosen": 81.08999633789062, | |
| "logits/rejected": 80.79169464111328, | |
| "logps/chosen": -34.18925094604492, | |
| "logps/rejected": -33.03681945800781, | |
| "loss": 24.3423, | |
| "rewards/accuracies": 0.4722222089767456, | |
| "rewards/chosen": 0.005247864406555891, | |
| "rewards/margins": 0.0130887096747756, | |
| "rewards/rejected": -0.007840845733880997, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 1032.0, | |
| "learning_rate": 2.564102564102564e-06, | |
| "logits/chosen": 80.66268157958984, | |
| "logits/rejected": 80.54837799072266, | |
| "logps/chosen": -33.62754821777344, | |
| "logps/rejected": -30.786510467529297, | |
| "loss": 25.2801, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.006235760636627674, | |
| "rewards/margins": 0.007138081826269627, | |
| "rewards/rejected": -0.0009023217717185616, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 1200.0, | |
| "learning_rate": 3.846153846153847e-06, | |
| "logits/chosen": 82.51216888427734, | |
| "logits/rejected": 82.54035949707031, | |
| "logps/chosen": -33.695411682128906, | |
| "logps/rejected": -31.29660415649414, | |
| "loss": 24.6437, | |
| "rewards/accuracies": 0.48750001192092896, | |
| "rewards/chosen": 0.03174129128456116, | |
| "rewards/margins": 0.01901828870177269, | |
| "rewards/rejected": 0.012723001651465893, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 1144.0, | |
| "learning_rate": 4.999896948438434e-06, | |
| "logits/chosen": 81.04322814941406, | |
| "logits/rejected": 81.03800964355469, | |
| "logps/chosen": -32.80583572387695, | |
| "logps/rejected": -33.2043571472168, | |
| "loss": 24.7349, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.03682267293334007, | |
| "rewards/margins": 0.020405994728207588, | |
| "rewards/rejected": 0.016416678205132484, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 948.0, | |
| "learning_rate": 4.987541037542187e-06, | |
| "logits/chosen": 78.70301055908203, | |
| "logits/rejected": 78.71062469482422, | |
| "logps/chosen": -30.72637367248535, | |
| "logps/rejected": -30.724206924438477, | |
| "loss": 26.36, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": 0.04788754880428314, | |
| "rewards/margins": 0.012979410588741302, | |
| "rewards/rejected": 0.03490813449025154, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 968.0, | |
| "learning_rate": 4.954691471941119e-06, | |
| "logits/chosen": 83.19561767578125, | |
| "logits/rejected": 83.24873352050781, | |
| "logps/chosen": -30.971267700195312, | |
| "logps/rejected": -29.551761627197266, | |
| "loss": 25.0468, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.020389318466186523, | |
| "rewards/margins": 0.016513368114829063, | |
| "rewards/rejected": 0.0038759508170187473, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 1552.0, | |
| "learning_rate": 4.901618883413549e-06, | |
| "logits/chosen": 83.78221130371094, | |
| "logits/rejected": 83.81100463867188, | |
| "logps/chosen": -30.479488372802734, | |
| "logps/rejected": -33.11530303955078, | |
| "loss": 24.2453, | |
| "rewards/accuracies": 0.5375000238418579, | |
| "rewards/chosen": 0.01495352853089571, | |
| "rewards/margins": 0.02271350473165512, | |
| "rewards/rejected": -0.007759974803775549, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 1104.0, | |
| "learning_rate": 4.828760511501322e-06, | |
| "logits/chosen": 81.36155700683594, | |
| "logits/rejected": 81.35380554199219, | |
| "logps/chosen": -31.37007713317871, | |
| "logps/rejected": -30.98931884765625, | |
| "loss": 23.0324, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.013486603274941444, | |
| "rewards/margins": 0.03665385767817497, | |
| "rewards/rejected": -0.023167254403233528, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 1120.0, | |
| "learning_rate": 4.7367166013034295e-06, | |
| "logits/chosen": 78.11141967773438, | |
| "logits/rejected": 78.0789794921875, | |
| "logps/chosen": -32.44235610961914, | |
| "logps/rejected": -31.213552474975586, | |
| "loss": 22.8826, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.01958216354250908, | |
| "rewards/margins": 0.0457901768386364, | |
| "rewards/rejected": -0.02620801329612732, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 1056.0, | |
| "learning_rate": 4.626245458345211e-06, | |
| "logits/chosen": 83.38008117675781, | |
| "logits/rejected": 83.412841796875, | |
| "logps/chosen": -34.02827072143555, | |
| "logps/rejected": -31.858572006225586, | |
| "loss": 23.2882, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.027672046795487404, | |
| "rewards/margins": 0.038474611937999725, | |
| "rewards/rejected": -0.010802562348544598, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "eval_logits/chosen": 98.75923156738281, | |
| "eval_logits/rejected": 98.74642181396484, | |
| "eval_logps/chosen": -32.47222137451172, | |
| "eval_logps/rejected": -36.06691360473633, | |
| "eval_loss": 25.671157836914062, | |
| "eval_rewards/accuracies": 0.5228405594825745, | |
| "eval_rewards/chosen": -0.0029044141992926598, | |
| "eval_rewards/margins": 0.0071340943686664104, | |
| "eval_rewards/rejected": -0.010038508102297783, | |
| "eval_runtime": 104.2457, | |
| "eval_samples_per_second": 3.29, | |
| "eval_steps_per_second": 0.412, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 1528.0, | |
| "learning_rate": 4.498257201263691e-06, | |
| "logits/chosen": 83.55766296386719, | |
| "logits/rejected": 83.4551010131836, | |
| "logps/chosen": -32.478965759277344, | |
| "logps/rejected": -32.80836486816406, | |
| "loss": 21.9052, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.05438382551074028, | |
| "rewards/margins": 0.06963126361370087, | |
| "rewards/rejected": -0.015247439965605736, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 1504.0, | |
| "learning_rate": 4.353806263777678e-06, | |
| "logits/chosen": 83.767822265625, | |
| "logits/rejected": 83.86802673339844, | |
| "logps/chosen": -28.277118682861328, | |
| "logps/rejected": -35.63311004638672, | |
| "loss": 20.22, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.06524648517370224, | |
| "rewards/margins": 0.08264312148094177, | |
| "rewards/rejected": -0.01739663816988468, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 956.0, | |
| "learning_rate": 4.1940827077152755e-06, | |
| "logits/chosen": 80.87911224365234, | |
| "logits/rejected": 80.89720153808594, | |
| "logps/chosen": -30.450307846069336, | |
| "logps/rejected": -32.20357131958008, | |
| "loss": 20.9442, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.04570445418357849, | |
| "rewards/margins": 0.07289845496416092, | |
| "rewards/rejected": -0.02719399705529213, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 740.0, | |
| "learning_rate": 4.0204024186666215e-06, | |
| "logits/chosen": 82.11528778076172, | |
| "logits/rejected": 82.12232971191406, | |
| "logps/chosen": -27.089996337890625, | |
| "logps/rejected": -33.013877868652344, | |
| "loss": 20.3465, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.035732947289943695, | |
| "rewards/margins": 0.09478302299976349, | |
| "rewards/rejected": -0.059050071984529495, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 948.0, | |
| "learning_rate": 3.834196265035119e-06, | |
| "logits/chosen": 80.63235473632812, | |
| "logits/rejected": 80.60447692871094, | |
| "logps/chosen": -28.92165756225586, | |
| "logps/rejected": -33.14097213745117, | |
| "loss": 18.7455, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 0.04674536734819412, | |
| "rewards/margins": 0.10419263690710068, | |
| "rewards/rejected": -0.05744727700948715, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 1664.0, | |
| "learning_rate": 3.636998309800573e-06, | |
| "logits/chosen": 82.47834777832031, | |
| "logits/rejected": 82.48927307128906, | |
| "logps/chosen": -33.57959747314453, | |
| "logps/rejected": -30.44466209411621, | |
| "loss": 21.1798, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.055715084075927734, | |
| "rewards/margins": 0.10492750257253647, | |
| "rewards/rejected": -0.049212418496608734, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 1152.0, | |
| "learning_rate": 3.4304331721118078e-06, | |
| "logits/chosen": 83.31620788574219, | |
| "logits/rejected": 83.2592544555664, | |
| "logps/chosen": -30.965845108032227, | |
| "logps/rejected": -32.591552734375, | |
| "loss": 21.522, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.02934456244111061, | |
| "rewards/margins": 0.08649395406246185, | |
| "rewards/rejected": -0.057149387896060944, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 896.0, | |
| "learning_rate": 3.2162026428305436e-06, | |
| "logits/chosen": 80.9317626953125, | |
| "logits/rejected": 80.91111755371094, | |
| "logps/chosen": -30.60970687866211, | |
| "logps/rejected": -31.63262939453125, | |
| "loss": 18.9358, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.058744143694639206, | |
| "rewards/margins": 0.10628656297922134, | |
| "rewards/rejected": -0.04754243046045303, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 620.0, | |
| "learning_rate": 2.996071664294641e-06, | |
| "logits/chosen": 82.61732482910156, | |
| "logits/rejected": 82.6050033569336, | |
| "logps/chosen": -30.360843658447266, | |
| "logps/rejected": -30.71734619140625, | |
| "loss": 23.6279, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.04055742174386978, | |
| "rewards/margins": 0.06548986583948135, | |
| "rewards/rejected": -0.024932442232966423, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 700.0, | |
| "learning_rate": 2.7718537898066833e-06, | |
| "logits/chosen": 78.06134033203125, | |
| "logits/rejected": 78.00863647460938, | |
| "logps/chosen": -33.79378890991211, | |
| "logps/rejected": -32.66465377807617, | |
| "loss": 20.2659, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.09589491784572601, | |
| "rewards/margins": 0.1130196675658226, | |
| "rewards/rejected": -0.017124753445386887, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "eval_logits/chosen": 98.65145874023438, | |
| "eval_logits/rejected": 98.6263427734375, | |
| "eval_logps/chosen": -32.57748031616211, | |
| "eval_logps/rejected": -36.30455017089844, | |
| "eval_loss": 24.828882217407227, | |
| "eval_rewards/accuracies": 0.5340532064437866, | |
| "eval_rewards/chosen": -0.013430174440145493, | |
| "eval_rewards/margins": 0.020371900871396065, | |
| "eval_rewards/rejected": -0.03380206972360611, | |
| "eval_runtime": 104.0575, | |
| "eval_samples_per_second": 3.296, | |
| "eval_steps_per_second": 0.413, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 1320.0, | |
| "learning_rate": 2.5453962426402006e-06, | |
| "logits/chosen": 80.6713638305664, | |
| "logits/rejected": 80.57633972167969, | |
| "logps/chosen": -33.16452407836914, | |
| "logps/rejected": -35.397491455078125, | |
| "loss": 18.2607, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 0.07811959832906723, | |
| "rewards/margins": 0.11925957351922989, | |
| "rewards/rejected": -0.04113996401429176, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 652.0, | |
| "learning_rate": 2.3185646976551794e-06, | |
| "logits/chosen": 82.79257202148438, | |
| "logits/rejected": 82.88256072998047, | |
| "logps/chosen": -30.9622859954834, | |
| "logps/rejected": -31.26416015625, | |
| "loss": 16.5714, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 0.09866676479578018, | |
| "rewards/margins": 0.15218719840049744, | |
| "rewards/rejected": -0.053520433604717255, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 1144.0, | |
| "learning_rate": 2.0932279108998323e-06, | |
| "logits/chosen": 79.95256042480469, | |
| "logits/rejected": 80.00727844238281, | |
| "logps/chosen": -32.26613235473633, | |
| "logps/rejected": -34.29724884033203, | |
| "loss": 21.5779, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.05396091938018799, | |
| "rewards/margins": 0.08247107267379761, | |
| "rewards/rejected": -0.028510143980383873, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 856.0, | |
| "learning_rate": 1.8712423238279358e-06, | |
| "logits/chosen": 82.29564666748047, | |
| "logits/rejected": 82.59175109863281, | |
| "logps/chosen": -30.715564727783203, | |
| "logps/rejected": -31.892749786376953, | |
| "loss": 16.3789, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": 0.0972500815987587, | |
| "rewards/margins": 0.13332059979438782, | |
| "rewards/rejected": -0.03607049956917763, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 1000.0, | |
| "learning_rate": 1.6544367689701824e-06, | |
| "logits/chosen": 80.96902465820312, | |
| "logits/rejected": 81.0352554321289, | |
| "logps/chosen": -26.926239013671875, | |
| "logps/rejected": -30.266109466552734, | |
| "loss": 20.1404, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.066965751349926, | |
| "rewards/margins": 0.10350307077169418, | |
| "rewards/rejected": -0.03653731197118759, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 1064.0, | |
| "learning_rate": 1.4445974030621963e-06, | |
| "logits/chosen": 78.25106048583984, | |
| "logits/rejected": 78.37794494628906, | |
| "logps/chosen": -30.432825088500977, | |
| "logps/rejected": -36.55792999267578, | |
| "loss": 15.8988, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.11281673610210419, | |
| "rewards/margins": 0.158446803689003, | |
| "rewards/rejected": -0.0456300750374794, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 708.0, | |
| "learning_rate": 1.243452991757889e-06, | |
| "logits/chosen": 77.56661224365234, | |
| "logits/rejected": 77.58919525146484, | |
| "logps/chosen": -30.973047256469727, | |
| "logps/rejected": -31.90987777709961, | |
| "loss": 18.1036, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": 0.08058776706457138, | |
| "rewards/margins": 0.1210336685180664, | |
| "rewards/rejected": -0.040445905178785324, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 1056.0, | |
| "learning_rate": 1.0526606671603523e-06, | |
| "logits/chosen": 80.2997817993164, | |
| "logits/rejected": 80.07062530517578, | |
| "logps/chosen": -31.0832576751709, | |
| "logps/rejected": -29.884586334228516, | |
| "loss": 22.0775, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.07266353815793991, | |
| "rewards/margins": 0.0944225937128067, | |
| "rewards/rejected": -0.021759048104286194, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 808.0, | |
| "learning_rate": 8.737922755071455e-07, | |
| "logits/chosen": 80.42820739746094, | |
| "logits/rejected": 80.34810638427734, | |
| "logps/chosen": -33.01291275024414, | |
| "logps/rejected": -32.72394561767578, | |
| "loss": 15.1146, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": 0.10913344472646713, | |
| "rewards/margins": 0.17984186112880707, | |
| "rewards/rejected": -0.07070842385292053, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 1072.0, | |
| "learning_rate": 7.08321427484816e-07, | |
| "logits/chosen": 76.04508209228516, | |
| "logits/rejected": 76.13191986083984, | |
| "logps/chosen": -32.2183952331543, | |
| "logps/rejected": -29.19476890563965, | |
| "loss": 20.0695, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.11873127520084381, | |
| "rewards/margins": 0.12800584733486176, | |
| "rewards/rejected": -0.009274585172533989, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "eval_logits/chosen": 98.68071746826172, | |
| "eval_logits/rejected": 98.6502685546875, | |
| "eval_logps/chosen": -32.66925811767578, | |
| "eval_logps/rejected": -36.280174255371094, | |
| "eval_loss": 26.15445899963379, | |
| "eval_rewards/accuracies": 0.5307309031486511, | |
| "eval_rewards/chosen": -0.022608023136854172, | |
| "eval_rewards/margins": 0.008756463415920734, | |
| "eval_rewards/rejected": -0.03136448562145233, | |
| "eval_runtime": 103.8063, | |
| "eval_samples_per_second": 3.304, | |
| "eval_steps_per_second": 0.414, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 1136.0, | |
| "learning_rate": 5.576113578589035e-07, | |
| "logits/chosen": 83.16893005371094, | |
| "logits/rejected": 83.19877624511719, | |
| "logps/chosen": -30.013708114624023, | |
| "logps/rejected": -32.592529296875, | |
| "loss": 18.2229, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": 0.08498243242502213, | |
| "rewards/margins": 0.12425950914621353, | |
| "rewards/rejected": -0.039277076721191406, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 596.0, | |
| "learning_rate": 4.229036944380913e-07, | |
| "logits/chosen": 80.65386199951172, | |
| "logits/rejected": 80.6548080444336, | |
| "logps/chosen": -30.399967193603516, | |
| "logps/rejected": -29.175945281982422, | |
| "loss": 16.3256, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": 0.11985927820205688, | |
| "rewards/margins": 0.1455042064189911, | |
| "rewards/rejected": -0.025644922628998756, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 604.0, | |
| "learning_rate": 3.053082288996112e-07, | |
| "logits/chosen": 77.81417846679688, | |
| "logits/rejected": 77.8701171875, | |
| "logps/chosen": -29.092737197875977, | |
| "logps/rejected": -33.01492691040039, | |
| "loss": 15.6194, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 0.12598693370819092, | |
| "rewards/margins": 0.16399319469928741, | |
| "rewards/rejected": -0.0380062535405159, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 1384.0, | |
| "learning_rate": 2.0579377374915805e-07, | |
| "logits/chosen": 82.1661376953125, | |
| "logits/rejected": 82.20028686523438, | |
| "logps/chosen": -32.25991439819336, | |
| "logps/rejected": -33.82966995239258, | |
| "loss": 18.1168, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.09522315859794617, | |
| "rewards/margins": 0.1413211077451706, | |
| "rewards/rejected": -0.04609795659780502, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 736.0, | |
| "learning_rate": 1.2518018074041684e-07, | |
| "logits/chosen": 81.15172576904297, | |
| "logits/rejected": 81.16615295410156, | |
| "logps/chosen": -32.44929504394531, | |
| "logps/rejected": -33.39020919799805, | |
| "loss": 17.9927, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": 0.11944446712732315, | |
| "rewards/margins": 0.14928530156612396, | |
| "rewards/rejected": -0.029840845614671707, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 832.0, | |
| "learning_rate": 6.41315865106129e-08, | |
| "logits/chosen": 82.63230895996094, | |
| "logits/rejected": 82.66060638427734, | |
| "logps/chosen": -28.42384910583496, | |
| "logps/rejected": -31.795475006103516, | |
| "loss": 17.9341, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.11283614486455917, | |
| "rewards/margins": 0.12739871442317963, | |
| "rewards/rejected": -0.014562586322426796, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 900.0, | |
| "learning_rate": 2.3150941078050325e-08, | |
| "logits/chosen": 82.10162353515625, | |
| "logits/rejected": 82.12476348876953, | |
| "logps/chosen": -31.836299896240234, | |
| "logps/rejected": -35.486595153808594, | |
| "loss": 19.9538, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.10398608446121216, | |
| "rewards/margins": 0.14170874655246735, | |
| "rewards/rejected": -0.03772266209125519, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 888.0, | |
| "learning_rate": 2.575864278703266e-09, | |
| "logits/chosen": 76.01589965820312, | |
| "logits/rejected": 75.89103698730469, | |
| "logps/chosen": -29.72897720336914, | |
| "logps/rejected": -28.481863021850586, | |
| "loss": 18.4584, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.08471504598855972, | |
| "rewards/margins": 0.11684386432170868, | |
| "rewards/rejected": -0.03212881088256836, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 385, | |
| "total_flos": 0.0, | |
| "train_loss": 20.413429577319654, | |
| "train_runtime": 2557.2601, | |
| "train_samples_per_second": 1.204, | |
| "train_steps_per_second": 0.151 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 385, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |