| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9994666666666666, | |
| "eval_steps": 500, | |
| "global_step": 937, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 5.319148936170213e-08, | |
| "logits/chosen": 0.1184067577123642, | |
| "logits/rejected": 0.3525714576244354, | |
| "logps/chosen": -429.5767822265625, | |
| "logps/rejected": -514.7810668945312, | |
| "loss": 0.3025, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5.319148936170213e-07, | |
| "logits/chosen": 0.05111883208155632, | |
| "logits/rejected": 0.23624171316623688, | |
| "logps/chosen": -363.744140625, | |
| "logps/rejected": -491.4056091308594, | |
| "loss": 0.3259, | |
| "rewards/accuracies": 0.3819444477558136, | |
| "rewards/chosen": 2.125546416209545e-05, | |
| "rewards/margins": -4.7190667828544974e-05, | |
| "rewards/rejected": 6.844610470579937e-05, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.0638297872340427e-06, | |
| "logits/chosen": 0.0970318466424942, | |
| "logits/rejected": 0.21818876266479492, | |
| "logps/chosen": -342.73223876953125, | |
| "logps/rejected": -533.9749755859375, | |
| "loss": 0.3389, | |
| "rewards/accuracies": 0.5062500238418579, | |
| "rewards/chosen": -0.00010573499457677826, | |
| "rewards/margins": 0.00035456178011372685, | |
| "rewards/rejected": -0.00046029678196646273, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.595744680851064e-06, | |
| "logits/chosen": 0.08977697044610977, | |
| "logits/rejected": 0.20492060482501984, | |
| "logps/chosen": -356.282958984375, | |
| "logps/rejected": -506.893798828125, | |
| "loss": 0.3282, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.0007758208666928113, | |
| "rewards/margins": 0.0015017592813819647, | |
| "rewards/rejected": -0.0022775803226977587, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 2.1276595744680853e-06, | |
| "logits/chosen": 0.14597493410110474, | |
| "logits/rejected": 0.2742587625980377, | |
| "logps/chosen": -373.9236755371094, | |
| "logps/rejected": -478.6368103027344, | |
| "loss": 0.3329, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.0021772016771137714, | |
| "rewards/margins": 0.00360355107113719, | |
| "rewards/rejected": -0.005780753213912249, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.6595744680851065e-06, | |
| "logits/chosen": 0.06719115376472473, | |
| "logits/rejected": 0.20441851019859314, | |
| "logps/chosen": -339.3710021972656, | |
| "logps/rejected": -437.3455505371094, | |
| "loss": 0.3465, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.00426669092848897, | |
| "rewards/margins": 0.009421268478035927, | |
| "rewards/rejected": -0.01368795894086361, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 3.191489361702128e-06, | |
| "logits/chosen": -0.008065430447459221, | |
| "logits/rejected": 0.08582984656095505, | |
| "logps/chosen": -321.49005126953125, | |
| "logps/rejected": -471.43896484375, | |
| "loss": 0.3231, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.008598757907748222, | |
| "rewards/margins": 0.023485153913497925, | |
| "rewards/rejected": -0.032083909958601, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 3.723404255319149e-06, | |
| "logits/chosen": 0.06349015235900879, | |
| "logits/rejected": 0.1338900625705719, | |
| "logps/chosen": -400.74725341796875, | |
| "logps/rejected": -535.9130859375, | |
| "loss": 0.3199, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.02710326947271824, | |
| "rewards/margins": 0.03926965966820717, | |
| "rewards/rejected": -0.06637293100357056, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.255319148936171e-06, | |
| "logits/chosen": -0.012521197088062763, | |
| "logits/rejected": 0.05875171348452568, | |
| "logps/chosen": -416.5260314941406, | |
| "logps/rejected": -619.4620971679688, | |
| "loss": 0.2653, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.04513772577047348, | |
| "rewards/margins": 0.09761229157447815, | |
| "rewards/rejected": -0.14275000989437103, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.787234042553192e-06, | |
| "logits/chosen": -0.06508742272853851, | |
| "logits/rejected": -0.025008535012602806, | |
| "logps/chosen": -398.4359436035156, | |
| "logps/rejected": -709.9359130859375, | |
| "loss": 0.2408, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -0.06391867995262146, | |
| "rewards/margins": 0.18030937016010284, | |
| "rewards/rejected": -0.2442280501127243, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.999375059004058e-06, | |
| "logits/chosen": -0.11634773015975952, | |
| "logits/rejected": -0.04804380610585213, | |
| "logps/chosen": -461.4971618652344, | |
| "logps/rejected": -726.8289794921875, | |
| "loss": 0.2717, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.0884505957365036, | |
| "rewards/margins": 0.17610163986682892, | |
| "rewards/rejected": -0.2645522356033325, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.9955571065548795e-06, | |
| "logits/chosen": -0.07285631448030472, | |
| "logits/rejected": -0.07443198561668396, | |
| "logps/chosen": -428.96588134765625, | |
| "logps/rejected": -709.934326171875, | |
| "loss": 0.2447, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.06827546656131744, | |
| "rewards/margins": 0.1353142112493515, | |
| "rewards/rejected": -0.20358964800834656, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.9882736864879e-06, | |
| "logits/chosen": -0.176306813955307, | |
| "logits/rejected": -0.049706846475601196, | |
| "logps/chosen": -430.32562255859375, | |
| "logps/rejected": -693.2147216796875, | |
| "loss": 0.2456, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.08173463493585587, | |
| "rewards/margins": 0.13715550303459167, | |
| "rewards/rejected": -0.21889011561870575, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.977534912960124e-06, | |
| "logits/chosen": -0.15699656307697296, | |
| "logits/rejected": -0.0577981173992157, | |
| "logps/chosen": -445.72894287109375, | |
| "logps/rejected": -769.3819580078125, | |
| "loss": 0.2081, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.07942022383213043, | |
| "rewards/margins": 0.19334132969379425, | |
| "rewards/rejected": -0.2727615237236023, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.963355698422092e-06, | |
| "logits/chosen": -0.18548890948295593, | |
| "logits/rejected": -0.12828989326953888, | |
| "logps/chosen": -451.9037170410156, | |
| "logps/rejected": -739.3684692382812, | |
| "loss": 0.2257, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.0921342521905899, | |
| "rewards/margins": 0.16765089333057404, | |
| "rewards/rejected": -0.25978511571884155, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.945755732909625e-06, | |
| "logits/chosen": -0.15024694800376892, | |
| "logits/rejected": -0.06675387918949127, | |
| "logps/chosen": -537.3986206054688, | |
| "logps/rejected": -775.892578125, | |
| "loss": 0.2479, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.10632505267858505, | |
| "rewards/margins": 0.16134649515151978, | |
| "rewards/rejected": -0.2676715552806854, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.924759456701167e-06, | |
| "logits/chosen": -0.20865170657634735, | |
| "logits/rejected": -0.10186436027288437, | |
| "logps/chosen": -483.97760009765625, | |
| "logps/rejected": -849.4505004882812, | |
| "loss": 0.2265, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.09985359013080597, | |
| "rewards/margins": 0.2456916868686676, | |
| "rewards/rejected": -0.34554529190063477, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.900396026378671e-06, | |
| "logits/chosen": -0.23064300417900085, | |
| "logits/rejected": -0.12898269295692444, | |
| "logps/chosen": -494.6697692871094, | |
| "logps/rejected": -781.0926513671875, | |
| "loss": 0.2324, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.11788048595190048, | |
| "rewards/margins": 0.17606906592845917, | |
| "rewards/rejected": -0.29394954442977905, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.872699274339169e-06, | |
| "logits/chosen": -0.22916531562805176, | |
| "logits/rejected": -0.14095698297023773, | |
| "logps/chosen": -458.0244140625, | |
| "logps/rejected": -759.4512939453125, | |
| "loss": 0.2393, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.12103551626205444, | |
| "rewards/margins": 0.17601092159748077, | |
| "rewards/rejected": -0.29704639315605164, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.8417076618132434e-06, | |
| "logits/chosen": -0.19613122940063477, | |
| "logits/rejected": -0.19725319743156433, | |
| "logps/chosen": -477.3087463378906, | |
| "logps/rejected": -782.9113159179688, | |
| "loss": 0.2591, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.12908069789409637, | |
| "rewards/margins": 0.18730813264846802, | |
| "rewards/rejected": -0.3163888156414032, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.807464225455655e-06, | |
| "logits/chosen": -0.18979570269584656, | |
| "logits/rejected": -0.12616662681102753, | |
| "logps/chosen": -501.4048767089844, | |
| "logps/rejected": -773.6704711914062, | |
| "loss": 0.2318, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -0.126991406083107, | |
| "rewards/margins": 0.16587643325328827, | |
| "rewards/rejected": -0.29286783933639526, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.770016517582283e-06, | |
| "logits/chosen": -0.19870837032794952, | |
| "logits/rejected": -0.1376388520002365, | |
| "logps/chosen": -476.4922790527344, | |
| "logps/rejected": -797.6036376953125, | |
| "loss": 0.2344, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.12952814996242523, | |
| "rewards/margins": 0.19133572280406952, | |
| "rewards/rejected": -0.32086387276649475, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.7294165401363616e-06, | |
| "logits/chosen": -0.18674388527870178, | |
| "logits/rejected": -0.17149865627288818, | |
| "logps/chosen": -476.748779296875, | |
| "logps/rejected": -736.6768798828125, | |
| "loss": 0.2225, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.1292114555835724, | |
| "rewards/margins": 0.17972734570503235, | |
| "rewards/rejected": -0.30893880128860474, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.68572067247573e-06, | |
| "logits/chosen": -0.2134261131286621, | |
| "logits/rejected": -0.17389492690563202, | |
| "logps/chosen": -556.177490234375, | |
| "logps/rejected": -932.4171752929688, | |
| "loss": 0.2011, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.17576465010643005, | |
| "rewards/margins": 0.23368000984191895, | |
| "rewards/rejected": -0.4094447195529938, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.638989593081364e-06, | |
| "logits/chosen": -0.2312520444393158, | |
| "logits/rejected": -0.17490056157112122, | |
| "logps/chosen": -608.5135498046875, | |
| "logps/rejected": -883.7859497070312, | |
| "loss": 0.2251, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.19390198588371277, | |
| "rewards/margins": 0.20176473259925842, | |
| "rewards/rejected": -0.3956666886806488, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5892881952959015e-06, | |
| "logits/chosen": -0.19589188694953918, | |
| "logits/rejected": -0.16277478635311127, | |
| "logps/chosen": -522.8567504882812, | |
| "logps/rejected": -815.2623291015625, | |
| "loss": 0.2215, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -0.15903013944625854, | |
| "rewards/margins": 0.19178643822669983, | |
| "rewards/rejected": -0.3508165776729584, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.536685497209182e-06, | |
| "logits/chosen": -0.25143852829933167, | |
| "logits/rejected": -0.15717722475528717, | |
| "logps/chosen": -533.4439697265625, | |
| "logps/rejected": -940.82763671875, | |
| "loss": 0.2056, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.18470649421215057, | |
| "rewards/margins": 0.22995992004871368, | |
| "rewards/rejected": -0.41466641426086426, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.481254545815943e-06, | |
| "logits/chosen": -0.23728492856025696, | |
| "logits/rejected": -0.22243991494178772, | |
| "logps/chosen": -552.9218139648438, | |
| "logps/rejected": -903.7254028320312, | |
| "loss": 0.196, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.20833177864551544, | |
| "rewards/margins": 0.22902217507362366, | |
| "rewards/rejected": -0.4373539388179779, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.42307231557875e-06, | |
| "logits/chosen": -0.27826839685440063, | |
| "logits/rejected": -0.200110524892807, | |
| "logps/chosen": -523.31787109375, | |
| "logps/rejected": -899.9382934570312, | |
| "loss": 0.1954, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.20382864773273468, | |
| "rewards/margins": 0.2211592197418213, | |
| "rewards/rejected": -0.42498788237571716, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.3622196015370305e-06, | |
| "logits/chosen": -0.23895792663097382, | |
| "logits/rejected": -0.14843276143074036, | |
| "logps/chosen": -594.8115844726562, | |
| "logps/rejected": -907.5458984375, | |
| "loss": 0.2135, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.19232866168022156, | |
| "rewards/margins": 0.21376939117908478, | |
| "rewards/rejected": -0.40609806776046753, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.298780907110648e-06, | |
| "logits/chosen": -0.24878796935081482, | |
| "logits/rejected": -0.18727460503578186, | |
| "logps/chosen": -580.38671875, | |
| "logps/rejected": -878.1405029296875, | |
| "loss": 0.1938, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.2084374725818634, | |
| "rewards/margins": 0.20776410400867462, | |
| "rewards/rejected": -0.4162015914916992, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.23284432675381e-06, | |
| "logits/chosen": -0.3039627969264984, | |
| "logits/rejected": -0.23918378353118896, | |
| "logps/chosen": -601.3494873046875, | |
| "logps/rejected": -868.8585815429688, | |
| "loss": 0.2041, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.23962783813476562, | |
| "rewards/margins": 0.18101730942726135, | |
| "rewards/rejected": -0.420645147562027, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.164501423622277e-06, | |
| "logits/chosen": -0.3026599884033203, | |
| "logits/rejected": -0.22676105797290802, | |
| "logps/chosen": -559.4996337890625, | |
| "logps/rejected": -871.5550537109375, | |
| "loss": 0.1855, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -0.21141842007637024, | |
| "rewards/margins": 0.21575455367565155, | |
| "rewards/rejected": -0.427172988653183, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.0938471024237355e-06, | |
| "logits/chosen": -0.24148467183113098, | |
| "logits/rejected": -0.19344112277030945, | |
| "logps/chosen": -570.2223510742188, | |
| "logps/rejected": -958.9056396484375, | |
| "loss": 0.1827, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -0.22266192734241486, | |
| "rewards/margins": 0.2378120869398117, | |
| "rewards/rejected": -0.46047407388687134, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.020979477627907e-06, | |
| "logits/chosen": -0.23564183712005615, | |
| "logits/rejected": -0.23679451644420624, | |
| "logps/chosen": -623.2242431640625, | |
| "logps/rejected": -871.82568359375, | |
| "loss": 0.2297, | |
| "rewards/accuracies": 0.7562500238418579, | |
| "rewards/chosen": -0.2473243921995163, | |
| "rewards/margins": 0.19464418292045593, | |
| "rewards/rejected": -0.44196853041648865, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 3.9459997372194105e-06, | |
| "logits/chosen": -0.22112271189689636, | |
| "logits/rejected": -0.24255314469337463, | |
| "logps/chosen": -654.8084716796875, | |
| "logps/rejected": -992.0069580078125, | |
| "loss": 0.2101, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.2653377950191498, | |
| "rewards/margins": 0.21331480145454407, | |
| "rewards/rejected": -0.47865256667137146, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 3.869012002182573e-06, | |
| "logits/chosen": -0.2565579116344452, | |
| "logits/rejected": -0.23455138504505157, | |
| "logps/chosen": -634.1543579101562, | |
| "logps/rejected": -1048.505859375, | |
| "loss": 0.1816, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.246078759431839, | |
| "rewards/margins": 0.2945541739463806, | |
| "rewards/rejected": -0.540632963180542, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 3.7901231819133104e-06, | |
| "logits/chosen": -0.27797406911849976, | |
| "logits/rejected": -0.24355745315551758, | |
| "logps/chosen": -634.2208251953125, | |
| "logps/rejected": -972.6463012695312, | |
| "loss": 0.2072, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.2534112334251404, | |
| "rewards/margins": 0.23360367119312286, | |
| "rewards/rejected": -0.48701491951942444, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 3.709442825758875e-06, | |
| "logits/chosen": -0.2145443856716156, | |
| "logits/rejected": -0.2447199821472168, | |
| "logps/chosen": -630.0811157226562, | |
| "logps/rejected": -895.8094482421875, | |
| "loss": 0.2066, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.2521076202392578, | |
| "rewards/margins": 0.2124219834804535, | |
| "rewards/rejected": -0.4645296037197113, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.6270829708916113e-06, | |
| "logits/chosen": -0.26773589849472046, | |
| "logits/rejected": -0.1856563836336136, | |
| "logps/chosen": -635.20703125, | |
| "logps/rejected": -1008.9361572265625, | |
| "loss": 0.1878, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.2653789818286896, | |
| "rewards/margins": 0.2648240625858307, | |
| "rewards/rejected": -0.5302029848098755, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.543157986727991e-06, | |
| "logits/chosen": -0.24474501609802246, | |
| "logits/rejected": -0.21170318126678467, | |
| "logps/chosen": -730.6715087890625, | |
| "logps/rejected": -1069.3619384765625, | |
| "loss": 0.2077, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.2803260385990143, | |
| "rewards/margins": 0.2544651925563812, | |
| "rewards/rejected": -0.5347911715507507, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 3.4577844161089614e-06, | |
| "logits/chosen": -0.1851538121700287, | |
| "logits/rejected": -0.24047665297985077, | |
| "logps/chosen": -616.1502685546875, | |
| "logps/rejected": -903.6608276367188, | |
| "loss": 0.2123, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -0.24726176261901855, | |
| "rewards/margins": 0.19664901494979858, | |
| "rewards/rejected": -0.44391077756881714, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.3710808134621577e-06, | |
| "logits/chosen": -0.2203509360551834, | |
| "logits/rejected": -0.2040051519870758, | |
| "logps/chosen": -636.1844482421875, | |
| "logps/rejected": -1008.9363403320312, | |
| "loss": 0.1868, | |
| "rewards/accuracies": 0.856249988079071, | |
| "rewards/chosen": -0.2344098538160324, | |
| "rewards/margins": 0.2508249282836914, | |
| "rewards/rejected": -0.485234797000885, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 3.2831675801707126e-06, | |
| "logits/chosen": -0.2882968783378601, | |
| "logits/rejected": -0.28868910670280457, | |
| "logps/chosen": -629.03662109375, | |
| "logps/rejected": -997.5048828125, | |
| "loss": 0.1877, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.2724894881248474, | |
| "rewards/margins": 0.2427108734846115, | |
| "rewards/rejected": -0.5152003765106201, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.194166797377289e-06, | |
| "logits/chosen": -0.2842113971710205, | |
| "logits/rejected": -0.2725662589073181, | |
| "logps/chosen": -629.4954223632812, | |
| "logps/rejected": -921.0350341796875, | |
| "loss": 0.1871, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.2583664059638977, | |
| "rewards/margins": 0.23012156784534454, | |
| "rewards/rejected": -0.48848801851272583, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.104202056455501e-06, | |
| "logits/chosen": -0.23092789947986603, | |
| "logits/rejected": -0.2383890151977539, | |
| "logps/chosen": -603.952392578125, | |
| "logps/rejected": -931.93994140625, | |
| "loss": 0.1852, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.24851222336292267, | |
| "rewards/margins": 0.2311173975467682, | |
| "rewards/rejected": -0.4796296954154968, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 3.013398287384144e-06, | |
| "logits/chosen": -0.25257351994514465, | |
| "logits/rejected": -0.20124118030071259, | |
| "logps/chosen": -720.2821044921875, | |
| "logps/rejected": -1016.345703125, | |
| "loss": 0.2169, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -0.291043221950531, | |
| "rewards/margins": 0.23256464302539825, | |
| "rewards/rejected": -0.5236078500747681, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 2.9218815852625717e-06, | |
| "logits/chosen": -0.280508816242218, | |
| "logits/rejected": -0.27508029341697693, | |
| "logps/chosen": -599.3237915039062, | |
| "logps/rejected": -959.4754028320312, | |
| "loss": 0.1873, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.266213595867157, | |
| "rewards/margins": 0.23423174023628235, | |
| "rewards/rejected": -0.5004453063011169, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.829779035208113e-06, | |
| "logits/chosen": -0.264869749546051, | |
| "logits/rejected": -0.2316662073135376, | |
| "logps/chosen": -632.74853515625, | |
| "logps/rejected": -977.66845703125, | |
| "loss": 0.1933, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.2532680928707123, | |
| "rewards/margins": 0.23631851375102997, | |
| "rewards/rejected": -0.48958665132522583, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 2.737218535878705e-06, | |
| "logits/chosen": -0.26723513007164, | |
| "logits/rejected": -0.2752009928226471, | |
| "logps/chosen": -612.4442749023438, | |
| "logps/rejected": -949.7120971679688, | |
| "loss": 0.1789, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -0.2443923056125641, | |
| "rewards/margins": 0.23941712081432343, | |
| "rewards/rejected": -0.4838094115257263, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 2.64432862186579e-06, | |
| "logits/chosen": -0.3223264515399933, | |
| "logits/rejected": -0.29264402389526367, | |
| "logps/chosen": -628.1448974609375, | |
| "logps/rejected": -997.76123046875, | |
| "loss": 0.1908, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -0.26595020294189453, | |
| "rewards/margins": 0.25043538212776184, | |
| "rewards/rejected": -0.516385555267334, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 2.551238285204126e-06, | |
| "logits/chosen": -0.2382831573486328, | |
| "logits/rejected": -0.2551121711730957, | |
| "logps/chosen": -653.0089111328125, | |
| "logps/rejected": -1023.2013549804688, | |
| "loss": 0.2152, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.2814193367958069, | |
| "rewards/margins": 0.2521519660949707, | |
| "rewards/rejected": -0.5335713028907776, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.4580767962463688e-06, | |
| "logits/chosen": -0.25415441393852234, | |
| "logits/rejected": -0.2852818965911865, | |
| "logps/chosen": -588.615234375, | |
| "logps/rejected": -990.0931396484375, | |
| "loss": 0.1917, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -0.2602744996547699, | |
| "rewards/margins": 0.24789556860923767, | |
| "rewards/rejected": -0.5081701278686523, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 2.3649735241511546e-06, | |
| "logits/chosen": -0.2810899317264557, | |
| "logits/rejected": -0.24050931632518768, | |
| "logps/chosen": -667.4718627929688, | |
| "logps/rejected": -1030.656494140625, | |
| "loss": 0.1788, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -0.2851766049861908, | |
| "rewards/margins": 0.2829527258872986, | |
| "rewards/rejected": -0.5681293606758118, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 2.2720577572339914e-06, | |
| "logits/chosen": -0.29503872990608215, | |
| "logits/rejected": -0.2738192677497864, | |
| "logps/chosen": -669.0242309570312, | |
| "logps/rejected": -1088.0225830078125, | |
| "loss": 0.1644, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.29529038071632385, | |
| "rewards/margins": 0.3003653287887573, | |
| "rewards/rejected": -0.5956557393074036, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 2.1794585234303995e-06, | |
| "logits/chosen": -0.2950688898563385, | |
| "logits/rejected": -0.1541706621646881, | |
| "logps/chosen": -705.3633422851562, | |
| "logps/rejected": -1031.0267333984375, | |
| "loss": 0.1825, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.2902067303657532, | |
| "rewards/margins": 0.284890741109848, | |
| "rewards/rejected": -0.5750974416732788, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.0873044111206407e-06, | |
| "logits/chosen": -0.28339633345603943, | |
| "logits/rejected": -0.22169962525367737, | |
| "logps/chosen": -629.9029541015625, | |
| "logps/rejected": -1066.449462890625, | |
| "loss": 0.1942, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.2589099705219269, | |
| "rewards/margins": 0.2774832844734192, | |
| "rewards/rejected": -0.5363932847976685, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.9957233905648293e-06, | |
| "logits/chosen": -0.24862179160118103, | |
| "logits/rejected": -0.22971436381340027, | |
| "logps/chosen": -636.7313232421875, | |
| "logps/rejected": -972.0720825195312, | |
| "loss": 0.1842, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.2589462399482727, | |
| "rewards/margins": 0.2386230230331421, | |
| "rewards/rejected": -0.4975692629814148, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.904842636196402e-06, | |
| "logits/chosen": -0.2794085443019867, | |
| "logits/rejected": -0.25304803252220154, | |
| "logps/chosen": -655.1509399414062, | |
| "logps/rejected": -1037.134033203125, | |
| "loss": 0.2004, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.2735671103000641, | |
| "rewards/margins": 0.27847912907600403, | |
| "rewards/rejected": -0.5520461797714233, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.814788350020726e-06, | |
| "logits/chosen": -0.2863375246524811, | |
| "logits/rejected": -0.2578074336051941, | |
| "logps/chosen": -644.8294067382812, | |
| "logps/rejected": -978.0472412109375, | |
| "loss": 0.1852, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.27340829372406006, | |
| "rewards/margins": 0.256661593914032, | |
| "rewards/rejected": -0.5300698280334473, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.725685586364051e-06, | |
| "logits/chosen": -0.30925947427749634, | |
| "logits/rejected": -0.26838865876197815, | |
| "logps/chosen": -632.4716796875, | |
| "logps/rejected": -994.3053588867188, | |
| "loss": 0.2151, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.27847468852996826, | |
| "rewards/margins": 0.25020256638526917, | |
| "rewards/rejected": -0.5286772847175598, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.6376580782162172e-06, | |
| "logits/chosen": -0.20261721312999725, | |
| "logits/rejected": -0.221277117729187, | |
| "logps/chosen": -634.4677124023438, | |
| "logps/rejected": -994.3683471679688, | |
| "loss": 0.1934, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -0.25305241346359253, | |
| "rewards/margins": 0.2600497603416443, | |
| "rewards/rejected": -0.5131021738052368, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.550828065408227e-06, | |
| "logits/chosen": -0.293588250875473, | |
| "logits/rejected": -0.250821977853775, | |
| "logps/chosen": -588.5574951171875, | |
| "logps/rejected": -976.6282348632812, | |
| "loss": 0.1876, | |
| "rewards/accuracies": 0.831250011920929, | |
| "rewards/chosen": -0.26154106855392456, | |
| "rewards/margins": 0.25152388215065, | |
| "rewards/rejected": -0.5130649209022522, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.4653161248633053e-06, | |
| "logits/chosen": -0.25644367933273315, | |
| "logits/rejected": -0.2529579699039459, | |
| "logps/chosen": -637.1114501953125, | |
| "logps/rejected": -1041.966064453125, | |
| "loss": 0.177, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.26990941166877747, | |
| "rewards/margins": 0.2800416052341461, | |
| "rewards/rejected": -0.5499509572982788, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.381241003157162e-06, | |
| "logits/chosen": -0.25759977102279663, | |
| "logits/rejected": -0.22868645191192627, | |
| "logps/chosen": -585.88671875, | |
| "logps/rejected": -971.8045043945312, | |
| "loss": 0.1695, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.24511754512786865, | |
| "rewards/margins": 0.2909080982208252, | |
| "rewards/rejected": -0.5360256433486938, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.298719451619979e-06, | |
| "logits/chosen": -0.27405595779418945, | |
| "logits/rejected": -0.25046929717063904, | |
| "logps/chosen": -686.3485107421875, | |
| "logps/rejected": -1069.6209716796875, | |
| "loss": 0.1779, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -0.29048749804496765, | |
| "rewards/margins": 0.3021007776260376, | |
| "rewards/rejected": -0.5925883054733276, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.2178660642091036e-06, | |
| "logits/chosen": -0.258391410112381, | |
| "logits/rejected": -0.2719349265098572, | |
| "logps/chosen": -642.6732177734375, | |
| "logps/rejected": -1080.385498046875, | |
| "loss": 0.1866, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.2909965515136719, | |
| "rewards/margins": 0.3018215596675873, | |
| "rewards/rejected": -0.5928180813789368, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.1387931183775821e-06, | |
| "logits/chosen": -0.29932543635368347, | |
| "logits/rejected": -0.2793118357658386, | |
| "logps/chosen": -649.6714477539062, | |
| "logps/rejected": -1090.2452392578125, | |
| "loss": 0.1646, | |
| "rewards/accuracies": 0.831250011920929, | |
| "rewards/chosen": -0.2679978013038635, | |
| "rewards/margins": 0.30532750487327576, | |
| "rewards/rejected": -0.5733253359794617, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.061610419159532e-06, | |
| "logits/chosen": -0.27962726354599, | |
| "logits/rejected": -0.21005702018737793, | |
| "logps/chosen": -654.9832763671875, | |
| "logps/rejected": -1042.5281982421875, | |
| "loss": 0.195, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.28967055678367615, | |
| "rewards/margins": 0.28377145528793335, | |
| "rewards/rejected": -0.5734419822692871, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.864251466888364e-07, | |
| "logits/chosen": -0.2319578379392624, | |
| "logits/rejected": -0.22523808479309082, | |
| "logps/chosen": -587.6854248046875, | |
| "logps/rejected": -929.84716796875, | |
| "loss": 0.1957, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.2619277834892273, | |
| "rewards/margins": 0.24305231869220734, | |
| "rewards/rejected": -0.5049800276756287, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 9.133417073629288e-07, | |
| "logits/chosen": -0.2581741213798523, | |
| "logits/rejected": -0.22566178441047668, | |
| "logps/chosen": -673.9590454101562, | |
| "logps/rejected": -1043.8779296875, | |
| "loss": 0.1946, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.2741592526435852, | |
| "rewards/margins": 0.2668037712574005, | |
| "rewards/rejected": -0.5409630537033081, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 8.424615888583332e-07, | |
| "logits/chosen": -0.24903810024261475, | |
| "logits/rejected": -0.19368259608745575, | |
| "logps/chosen": -702.1976318359375, | |
| "logps/rejected": -1104.967529296875, | |
| "loss": 0.1768, | |
| "rewards/accuracies": 0.78125, | |
| "rewards/chosen": -0.29275885224342346, | |
| "rewards/margins": 0.24981990456581116, | |
| "rewards/rejected": -0.5425786972045898, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.738832191993092e-07, | |
| "logits/chosen": -0.2902497947216034, | |
| "logits/rejected": -0.27156323194503784, | |
| "logps/chosen": -642.2786865234375, | |
| "logps/rejected": -1030.1949462890625, | |
| "loss": 0.1896, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.296554833650589, | |
| "rewards/margins": 0.2522734999656677, | |
| "rewards/rejected": -0.5488283634185791, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 7.077018300752917e-07, | |
| "logits/chosen": -0.25113141536712646, | |
| "logits/rejected": -0.27631789445877075, | |
| "logps/chosen": -658.340087890625, | |
| "logps/rejected": -1025.861328125, | |
| "loss": 0.1804, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.2932133376598358, | |
| "rewards/margins": 0.24299320578575134, | |
| "rewards/rejected": -0.5362066030502319, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 6.440093245969342e-07, | |
| "logits/chosen": -0.2601129412651062, | |
| "logits/rejected": -0.23299941420555115, | |
| "logps/chosen": -661.2008056640625, | |
| "logps/rejected": -1049.9329833984375, | |
| "loss": 0.1731, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -0.28521472215652466, | |
| "rewards/margins": 0.2829175591468811, | |
| "rewards/rejected": -0.5681322813034058, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 5.828941496744075e-07, | |
| "logits/chosen": -0.2920230031013489, | |
| "logits/rejected": -0.2706466317176819, | |
| "logps/chosen": -614.7430419921875, | |
| "logps/rejected": -1084.22802734375, | |
| "loss": 0.1975, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.26638248562812805, | |
| "rewards/margins": 0.29731154441833496, | |
| "rewards/rejected": -0.5636940598487854, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 5.244411731951671e-07, | |
| "logits/chosen": -0.2530083954334259, | |
| "logits/rejected": -0.2159876525402069, | |
| "logps/chosen": -636.4935913085938, | |
| "logps/rejected": -965.4192504882812, | |
| "loss": 0.1974, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.27532270550727844, | |
| "rewards/margins": 0.23001885414123535, | |
| "rewards/rejected": -0.5053415298461914, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 4.6873156617173594e-07, | |
| "logits/chosen": -0.2676984369754791, | |
| "logits/rejected": -0.2522314190864563, | |
| "logps/chosen": -619.7949829101562, | |
| "logps/rejected": -1006.8516845703125, | |
| "loss": 0.1934, | |
| "rewards/accuracies": 0.7875000238418579, | |
| "rewards/chosen": -0.25617390871047974, | |
| "rewards/margins": 0.25205108523368835, | |
| "rewards/rejected": -0.5082249641418457, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.1584269002318653e-07, | |
| "logits/chosen": -0.276606023311615, | |
| "logits/rejected": -0.18771035969257355, | |
| "logps/chosen": -680.2154541015625, | |
| "logps/rejected": -1099.0855712890625, | |
| "loss": 0.1918, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.2760952413082123, | |
| "rewards/margins": 0.2856435477733612, | |
| "rewards/rejected": -0.5617388486862183, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.658479891468258e-07, | |
| "logits/chosen": -0.2671615183353424, | |
| "logits/rejected": -0.28831931948661804, | |
| "logps/chosen": -652.9794921875, | |
| "logps/rejected": -1049.533447265625, | |
| "loss": 0.1852, | |
| "rewards/accuracies": 0.8125, | |
| "rewards/chosen": -0.2673850953578949, | |
| "rewards/margins": 0.2847273647785187, | |
| "rewards/rejected": -0.5521124601364136, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.18816888929272e-07, | |
| "logits/chosen": -0.2824237644672394, | |
| "logits/rejected": -0.2227623015642166, | |
| "logps/chosen": -643.0541381835938, | |
| "logps/rejected": -957.2637939453125, | |
| "loss": 0.207, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.2690204679965973, | |
| "rewards/margins": 0.2505555748939514, | |
| "rewards/rejected": -0.5195759534835815, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 2.748146993385484e-07, | |
| "logits/chosen": -0.3143889307975769, | |
| "logits/rejected": -0.2759067118167877, | |
| "logps/chosen": -651.569580078125, | |
| "logps/rejected": -1108.028076171875, | |
| "loss": 0.1587, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.2932509779930115, | |
| "rewards/margins": 0.2749115526676178, | |
| "rewards/rejected": -0.5681625604629517, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.3390252423108077e-07, | |
| "logits/chosen": -0.26839888095855713, | |
| "logits/rejected": -0.23119351267814636, | |
| "logps/chosen": -665.0179443359375, | |
| "logps/rejected": -1112.646728515625, | |
| "loss": 0.1658, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.28339672088623047, | |
| "rewards/margins": 0.3101740777492523, | |
| "rewards/rejected": -0.5935708284378052, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.961371764995243e-07, | |
| "logits/chosen": -0.2762928605079651, | |
| "logits/rejected": -0.28436392545700073, | |
| "logps/chosen": -631.9949951171875, | |
| "logps/rejected": -1104.309326171875, | |
| "loss": 0.1821, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.2872810959815979, | |
| "rewards/margins": 0.303080290555954, | |
| "rewards/rejected": -0.5903614163398743, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.61571099179261e-07, | |
| "logits/chosen": -0.2735849916934967, | |
| "logits/rejected": -0.2402714043855667, | |
| "logps/chosen": -652.4813842773438, | |
| "logps/rejected": -1141.485107421875, | |
| "loss": 0.1638, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.2600463032722473, | |
| "rewards/margins": 0.34138140082359314, | |
| "rewards/rejected": -0.6014277338981628, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3025229262312367e-07, | |
| "logits/chosen": -0.20563116669654846, | |
| "logits/rejected": -0.25282037258148193, | |
| "logps/chosen": -693.9833984375, | |
| "logps/rejected": -1084.752197265625, | |
| "loss": 0.183, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -0.2829940617084503, | |
| "rewards/margins": 0.27976807951927185, | |
| "rewards/rejected": -0.5627621412277222, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.0222424784546853e-07, | |
| "logits/chosen": -0.22862930595874786, | |
| "logits/rejected": -0.21136240661144257, | |
| "logps/chosen": -638.4495239257812, | |
| "logps/rejected": -1061.037109375, | |
| "loss": 0.1807, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.2733309268951416, | |
| "rewards/margins": 0.28016480803489685, | |
| "rewards/rejected": -0.5534957647323608, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.752588612816553e-08, | |
| "logits/chosen": -0.2907347083091736, | |
| "logits/rejected": -0.2525694966316223, | |
| "logps/chosen": -626.2022705078125, | |
| "logps/rejected": -999.1546630859375, | |
| "loss": 0.19, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.2700883746147156, | |
| "rewards/margins": 0.2750420868396759, | |
| "rewards/rejected": -0.5451304316520691, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.619150497236991e-08, | |
| "logits/chosen": -0.2824448347091675, | |
| "logits/rejected": -0.21472208201885223, | |
| "logps/chosen": -694.7796630859375, | |
| "logps/rejected": -1101.357666015625, | |
| "loss": 0.1808, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -0.30949831008911133, | |
| "rewards/margins": 0.29203012585639954, | |
| "rewards/rejected": -0.601528525352478, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.825073047112743e-08, | |
| "logits/chosen": -0.2339903563261032, | |
| "logits/rejected": -0.25414004921913147, | |
| "logps/chosen": -677.6926879882812, | |
| "logps/rejected": -1110.830810546875, | |
| "loss": 0.175, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.2908310890197754, | |
| "rewards/margins": 0.3078031539916992, | |
| "rewards/rejected": -0.5986341834068298, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 2.372847616895685e-08, | |
| "logits/chosen": -0.2827141582965851, | |
| "logits/rejected": -0.24739205837249756, | |
| "logps/chosen": -659.048583984375, | |
| "logps/rejected": -1077.357177734375, | |
| "loss": 0.1941, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -0.28590255975723267, | |
| "rewards/margins": 0.280953586101532, | |
| "rewards/rejected": -0.5668561458587646, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.264490846553279e-08, | |
| "logits/chosen": -0.29719191789627075, | |
| "logits/rejected": -0.23341615498065948, | |
| "logps/chosen": -676.8151245117188, | |
| "logps/rejected": -1061.142822265625, | |
| "loss": 0.1835, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -0.2869683802127838, | |
| "rewards/margins": 0.2846153676509857, | |
| "rewards/rejected": -0.57158362865448, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 5.015418611516165e-09, | |
| "logits/chosen": -0.22863967716693878, | |
| "logits/rejected": -0.20865638554096222, | |
| "logps/chosen": -656.0875854492188, | |
| "logps/rejected": -1003.3201904296875, | |
| "loss": 0.1869, | |
| "rewards/accuracies": 0.800000011920929, | |
| "rewards/chosen": -0.2743567228317261, | |
| "rewards/margins": 0.260437935590744, | |
| "rewards/rejected": -0.5347946286201477, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.506013354186993e-10, | |
| "logits/chosen": -0.22728657722473145, | |
| "logits/rejected": -0.21279653906822205, | |
| "logps/chosen": -687.8726806640625, | |
| "logps/rejected": -1077.039306640625, | |
| "loss": 0.1753, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.28496915102005005, | |
| "rewards/margins": 0.2890021502971649, | |
| "rewards/rejected": -0.5739713907241821, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 937, | |
| "total_flos": 0.0, | |
| "train_loss": 0.08644322651809155, | |
| "train_runtime": 5283.9744, | |
| "train_samples_per_second": 5.678, | |
| "train_steps_per_second": 0.177 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 937, | |
| "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 | |
| } | |