diff --git "a/model/trainer_state.json" "b/model/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/model/trainer_state.json"
@@ -0,0 +1,16262 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 100,
+  "global_step": 18046,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.000554139421478444,
+      "grad_norm": 8.061482429504395,
+      "learning_rate": 4.972375690607734e-08,
+      "loss": 1.6338,
+      "step": 10
+    },
+    {
+      "epoch": 0.001108278842956888,
+      "grad_norm": 7.974089622497559,
+      "learning_rate": 1.0497237569060773e-07,
+      "loss": 1.6318,
+      "step": 20
+    },
+    {
+      "epoch": 0.0016624182644353318,
+      "grad_norm": 7.397510051727295,
+      "learning_rate": 1.6022099447513812e-07,
+      "loss": 1.6289,
+      "step": 30
+    },
+    {
+      "epoch": 0.002216557685913776,
+      "grad_norm": 6.7981157302856445,
+      "learning_rate": 2.154696132596685e-07,
+      "loss": 1.6163,
+      "step": 40
+    },
+    {
+      "epoch": 0.00277069710739222,
+      "grad_norm": 6.111944675445557,
+      "learning_rate": 2.7071823204419884e-07,
+      "loss": 1.556,
+      "step": 50
+    },
+    {
+      "epoch": 0.0033248365288706637,
+      "grad_norm": 4.9665069580078125,
+      "learning_rate": 3.2596685082872926e-07,
+      "loss": 1.5341,
+      "step": 60
+    },
+    {
+      "epoch": 0.003878975950349108,
+      "grad_norm": 4.7185139656066895,
+      "learning_rate": 3.812154696132597e-07,
+      "loss": 1.4662,
+      "step": 70
+    },
+    {
+      "epoch": 0.004433115371827552,
+      "grad_norm": 3.9775431156158447,
+      "learning_rate": 4.3646408839779006e-07,
+      "loss": 1.4334,
+      "step": 80
+    },
+    {
+      "epoch": 0.0049872547933059955,
+      "grad_norm": 2.934105157852173,
+      "learning_rate": 4.917127071823205e-07,
+      "loss": 1.343,
+      "step": 90
+    },
+    {
+      "epoch": 0.00554139421478444,
+      "grad_norm": 2.5355584621429443,
+      "learning_rate": 5.469613259668509e-07,
+      "loss": 1.2924,
+      "step": 100
+    },
+    {
+      "epoch": 0.00554139421478444,
+      "eval_robovqa_loss": 0.11538717150688171,
+      "step": 100
+    },
+    {
+      "epoch": 0.00554139421478444,
+      "eval_agibot_loss": 0.4453522861003876,
+      "step": 100
+    },
+    {
+      "epoch": 0.00554139421478444,
+      "eval_holoassist_loss": 0.32668429613113403,
+      "step": 100
+    },
+    {
+      "epoch": 0.00554139421478444,
+      "eval_robofail_loss": 0.2020190954208374,
+      "step": 100
+    },
+    {
+      "epoch": 0.006095533636262884,
+      "grad_norm": 2.2609946727752686,
+      "learning_rate": 6.022099447513812e-07,
+      "loss": 1.2432,
+      "step": 110
+    },
+    {
+      "epoch": 0.006649673057741327,
+      "grad_norm": 1.9896363019943237,
+      "learning_rate": 6.574585635359116e-07,
+      "loss": 1.1733,
+      "step": 120
+    },
+    {
+      "epoch": 0.007203812479219772,
+      "grad_norm": 1.9352859258651733,
+      "learning_rate": 7.12707182320442e-07,
+      "loss": 1.1283,
+      "step": 130
+    },
+    {
+      "epoch": 0.007757951900698216,
+      "grad_norm": 2.1029937267303467,
+      "learning_rate": 7.679558011049723e-07,
+      "loss": 1.0797,
+      "step": 140
+    },
+    {
+      "epoch": 0.00831209132217666,
+      "grad_norm": 1.767221212387085,
+      "learning_rate": 8.232044198895027e-07,
+      "loss": 1.048,
+      "step": 150
+    },
+    {
+      "epoch": 0.008866230743655104,
+      "grad_norm": 1.5446223020553589,
+      "learning_rate": 8.784530386740331e-07,
+      "loss": 1.0221,
+      "step": 160
+    },
+    {
+      "epoch": 0.009420370165133548,
+      "grad_norm": 1.7626278400421143,
+      "learning_rate": 9.337016574585634e-07,
+      "loss": 0.973,
+      "step": 170
+    },
+    {
+      "epoch": 0.009974509586611991,
+      "grad_norm": 1.5761367082595825,
+      "learning_rate": 9.889502762430938e-07,
+      "loss": 0.9506,
+      "step": 180
+    },
+    {
+      "epoch": 0.010528649008090436,
+      "grad_norm": 1.9955872297286987,
+      "learning_rate": 9.999995052183149e-07,
+      "loss": 0.9409,
+      "step": 190
+    },
+    {
+      "epoch": 0.01108278842956888,
+      "grad_norm": 1.5445226430892944,
+      "learning_rate": 9.999974951693976e-07,
+      "loss": 0.9204,
+      "step": 200
+    },
+    {
+      "epoch": 0.01108278842956888,
+      "eval_robovqa_loss": 0.11623767763376236,
+      "step": 200
+    },
+    {
+      "epoch": 0.01108278842956888,
+      "eval_agibot_loss": 0.4662611484527588,
+      "step": 200
+    },
+    {
+      "epoch": 0.01108278842956888,
+      "eval_holoassist_loss": 0.3617129921913147,
+      "step": 200
+    },
+    {
+      "epoch": 0.01108278842956888,
+      "eval_robofail_loss": 0.20507261157035828,
+      "step": 200
+    },
+    {
+      "epoch": 0.011636927851047323,
+      "grad_norm": 1.3956602811813354,
+      "learning_rate": 9.999939389356036e-07,
+      "loss": 0.8989,
+      "step": 210
+    },
+    {
+      "epoch": 0.012191067272525767,
+      "grad_norm": 1.6003563404083252,
+      "learning_rate": 9.999888365279303e-07,
+      "loss": 0.8804,
+      "step": 220
+    },
+    {
+      "epoch": 0.012745206694004212,
+      "grad_norm": 1.590865969657898,
+      "learning_rate": 9.999821879621566e-07,
+      "loss": 0.8737,
+      "step": 230
+    },
+    {
+      "epoch": 0.013299346115482655,
+      "grad_norm": 1.7492831945419312,
+      "learning_rate": 9.99973993258842e-07,
+      "loss": 0.8607,
+      "step": 240
+    },
+    {
+      "epoch": 0.0138534855369611,
+      "grad_norm": 1.696846842765808,
+      "learning_rate": 9.999642524433278e-07,
+      "loss": 0.8442,
+      "step": 250
+    },
+    {
+      "epoch": 0.014407624958439544,
+      "grad_norm": 1.539231538772583,
+      "learning_rate": 9.999529655457362e-07,
+      "loss": 0.846,
+      "step": 260
+    },
+    {
+      "epoch": 0.014961764379917987,
+      "grad_norm": 1.487686276435852,
+      "learning_rate": 9.999401326009712e-07,
+      "loss": 0.8375,
+      "step": 270
+    },
+    {
+      "epoch": 0.015515903801396431,
+      "grad_norm": 1.5779697895050049,
+      "learning_rate": 9.999257536487164e-07,
+      "loss": 0.8355,
+      "step": 280
+    },
+    {
+      "epoch": 0.016070043222874874,
+      "grad_norm": 1.5976744890213013,
+      "learning_rate": 9.999098287334378e-07,
+      "loss": 0.8224,
+      "step": 290
+    },
+    {
+      "epoch": 0.01662418264435332,
+      "grad_norm": 1.44376802444458,
+      "learning_rate": 9.99892357904381e-07,
+      "loss": 0.8198,
+      "step": 300
+    },
+    {
+      "epoch": 0.01662418264435332,
+      "eval_robovqa_loss": 0.10422000288963318,
+      "step": 300
+    },
+    {
+      "epoch": 0.01662418264435332,
+      "eval_agibot_loss": 0.47240889072418213,
+      "step": 300
+    },
+    {
+      "epoch": 0.01662418264435332,
+      "eval_holoassist_loss": 0.36908626556396484,
+      "step": 300
+    },
+    {
+      "epoch": 0.01662418264435332,
+      "eval_robofail_loss": 0.2040308117866516,
+      "step": 300
+    },
+    {
+      "epoch": 0.017178322065831763,
+      "grad_norm": 1.5491384267807007,
+      "learning_rate": 9.998733412155723e-07,
+      "loss": 0.8214,
+      "step": 310
+    },
+    {
+      "epoch": 0.017732461487310208,
+      "grad_norm": 1.460451364517212,
+      "learning_rate": 9.998527787258193e-07,
+      "loss": 0.8068,
+      "step": 320
+    },
+    {
+      "epoch": 0.018286600908788652,
+      "grad_norm": 1.449819564819336,
+      "learning_rate": 9.998306704987087e-07,
+      "loss": 0.7934,
+      "step": 330
+    },
+    {
+      "epoch": 0.018840740330267097,
+      "grad_norm": 1.5707648992538452,
+      "learning_rate": 9.998070166026076e-07,
+      "loss": 0.7941,
+      "step": 340
+    },
+    {
+      "epoch": 0.019394879751745538,
+      "grad_norm": 1.8264530897140503,
+      "learning_rate": 9.99781817110663e-07,
+      "loss": 0.7892,
+      "step": 350
+    },
+    {
+      "epoch": 0.019949019173223982,
+      "grad_norm": 1.5225766897201538,
+      "learning_rate": 9.997550721008018e-07,
+      "loss": 0.7947,
+      "step": 360
+    },
+    {
+      "epoch": 0.020503158594702427,
+      "grad_norm": 1.531495213508606,
+      "learning_rate": 9.997267816557293e-07,
+      "loss": 0.7907,
+      "step": 370
+    },
+    {
+      "epoch": 0.02105729801618087,
+      "grad_norm": 1.4460949897766113,
+      "learning_rate": 9.996969458629309e-07,
+      "loss": 0.7744,
+      "step": 380
+    },
+    {
+      "epoch": 0.021611437437659316,
+      "grad_norm": 1.6509114503860474,
+      "learning_rate": 9.9966556481467e-07,
+      "loss": 0.7731,
+      "step": 390
+    },
+    {
+      "epoch": 0.02216557685913776,
+      "grad_norm": 1.4678701162338257,
+      "learning_rate": 9.996326386079894e-07,
+      "loss": 0.7815,
+      "step": 400
+    },
+    {
+      "epoch": 0.02216557685913776,
+      "eval_robovqa_loss": 0.10480137914419174,
+      "step": 400
+    },
+    {
+      "epoch": 0.02216557685913776,
+      "eval_agibot_loss": 0.470235139131546,
+      "step": 400
+    },
+    {
+      "epoch": 0.02216557685913776,
+      "eval_holoassist_loss": 0.3592274487018585,
+      "step": 400
+    },
+    {
+      "epoch": 0.02216557685913776,
+      "eval_robofail_loss": 0.20445893704891205,
+      "step": 400
+    },
+    {
+      "epoch": 0.0227197162806162,
+      "grad_norm": 1.4340639114379883,
+      "learning_rate": 9.995981673447091e-07,
+      "loss": 0.782,
+      "step": 410
+    },
+    {
+      "epoch": 0.023273855702094646,
+      "grad_norm": 1.4501503705978394,
+      "learning_rate": 9.99562151131428e-07,
+      "loss": 0.764,
+      "step": 420
+    },
+    {
+      "epoch": 0.02382799512357309,
+      "grad_norm": 1.5401359796524048,
+      "learning_rate": 9.995245900795215e-07,
+      "loss": 0.7563,
+      "step": 430
+    },
+    {
+      "epoch": 0.024382134545051535,
+      "grad_norm": 1.5172621011734009,
+      "learning_rate": 9.994854843051434e-07,
+      "loss": 0.7487,
+      "step": 440
+    },
+    {
+      "epoch": 0.02493627396652998,
+      "grad_norm": 1.4791797399520874,
+      "learning_rate": 9.994448339292233e-07,
+      "loss": 0.7595,
+      "step": 450
+    },
+    {
+      "epoch": 0.025490413388008424,
+      "grad_norm": 1.4658597707748413,
+      "learning_rate": 9.994026390774686e-07,
+      "loss": 0.7491,
+      "step": 460
+    },
+    {
+      "epoch": 0.02604455280948687,
+      "grad_norm": 1.585005760192871,
+      "learning_rate": 9.993588998803614e-07,
+      "loss": 0.7495,
+      "step": 470
+    },
+    {
+      "epoch": 0.02659869223096531,
+      "grad_norm": 1.5044341087341309,
+      "learning_rate": 9.993136164731602e-07,
+      "loss": 0.7558,
+      "step": 480
+    },
+    {
+      "epoch": 0.027152831652443754,
+      "grad_norm": 1.4776275157928467,
+      "learning_rate": 9.992667889958987e-07,
+      "loss": 0.7404,
+      "step": 490
+    },
+    {
+      "epoch": 0.0277069710739222,
+      "grad_norm": 1.4741315841674805,
+      "learning_rate": 9.99218417593386e-07,
+      "loss": 0.748,
+      "step": 500
+    },
+    {
+      "epoch": 0.0277069710739222,
+      "eval_robovqa_loss": 0.1006377562880516,
+      "step": 500
+    },
+    {
+      "epoch": 0.0277069710739222,
+      "eval_agibot_loss": 0.46475985646247864,
+      "step": 500
+    },
+    {
+      "epoch": 0.0277069710739222,
+      "eval_holoassist_loss": 0.34592315554618835,
+      "step": 500
+    },
+    {
+      "epoch": 0.0277069710739222,
+      "eval_robofail_loss": 0.20546354353427887,
+      "step": 500
+    },
+    {
+      "epoch": 0.028261110495400643,
+      "grad_norm": 1.5555678606033325,
+      "learning_rate": 9.991685024152045e-07,
+      "loss": 0.7346,
+      "step": 510
+    },
+    {
+      "epoch": 0.028815249916879088,
+      "grad_norm": 1.5374836921691895,
+      "learning_rate": 9.991170436157114e-07,
+      "loss": 0.7275,
+      "step": 520
+    },
+    {
+      "epoch": 0.029369389338357532,
+      "grad_norm": 1.3541239500045776,
+      "learning_rate": 9.990640413540375e-07,
+      "loss": 0.7382,
+      "step": 530
+    },
+    {
+      "epoch": 0.029923528759835973,
+      "grad_norm": 1.5267432928085327,
+      "learning_rate": 9.990094957940855e-07,
+      "loss": 0.7293,
+      "step": 540
+    },
+    {
+      "epoch": 0.030477668181314418,
+      "grad_norm": 1.9004600048065186,
+      "learning_rate": 9.989534071045319e-07,
+      "loss": 0.7313,
+      "step": 550
+    },
+    {
+      "epoch": 0.031031807602792862,
+      "grad_norm": 2.152632474899292,
+      "learning_rate": 9.98895775458824e-07,
+      "loss": 0.7337,
+      "step": 560
+    },
+    {
+      "epoch": 0.0315859470242713,
+      "grad_norm": 1.4284332990646362,
+      "learning_rate": 9.988366010351817e-07,
+      "loss": 0.7316,
+      "step": 570
+    },
+    {
+      "epoch": 0.03214008644574975,
+      "grad_norm": 1.4656455516815186,
+      "learning_rate": 9.987758840165947e-07,
+      "loss": 0.7295,
+      "step": 580
+    },
+    {
+      "epoch": 0.03269422586722819,
+      "grad_norm": 1.5224812030792236,
+      "learning_rate": 9.987136245908232e-07,
+      "loss": 0.7164,
+      "step": 590
+    },
+    {
+      "epoch": 0.03324836528870664,
+      "grad_norm": 1.6718392372131348,
+      "learning_rate": 9.986498229503978e-07,
+      "loss": 0.7142,
+      "step": 600
+    },
+    {
+      "epoch": 0.03324836528870664,
+      "eval_robovqa_loss": 0.10026440769433975,
+      "step": 600
+    },
+    {
+      "epoch": 0.03324836528870664,
+      "eval_agibot_loss": 0.45991361141204834,
+      "step": 600
+    },
+    {
+      "epoch": 0.03324836528870664,
+      "eval_holoassist_loss": 0.3549599349498749,
+      "step": 600
+    },
+    {
+      "epoch": 0.03324836528870664,
+      "eval_robofail_loss": 0.20515353977680206,
+      "step": 600
+    },
+    {
+      "epoch": 0.03380250471018508,
+      "grad_norm": 1.45632803440094,
+      "learning_rate": 9.985844792926177e-07,
+      "loss": 0.7272,
+      "step": 610
+    },
+    {
+      "epoch": 0.034356644131663526,
+      "grad_norm": 1.462807059288025,
+      "learning_rate": 9.985175938195504e-07,
+      "loss": 0.7224,
+      "step": 620
+    },
+    {
+      "epoch": 0.03491078355314197,
+      "grad_norm": 1.3945891857147217,
+      "learning_rate": 9.984491667380315e-07,
+      "loss": 0.7114,
+      "step": 630
+    },
+    {
+      "epoch": 0.035464922974620415,
+      "grad_norm": 1.4921393394470215,
+      "learning_rate": 9.98379198259664e-07,
+      "loss": 0.7165,
+      "step": 640
+    },
+    {
+      "epoch": 0.03601906239609886,
+      "grad_norm": 1.4626291990280151,
+      "learning_rate": 9.983076886008178e-07,
+      "loss": 0.7015,
+      "step": 650
+    },
+    {
+      "epoch": 0.036573201817577304,
+      "grad_norm": 1.4525871276855469,
+      "learning_rate": 9.982346379826277e-07,
+      "loss": 0.7022,
+      "step": 660
+    },
+    {
+      "epoch": 0.03712734123905575,
+      "grad_norm": 1.4188086986541748,
+      "learning_rate": 9.981600466309946e-07,
+      "loss": 0.7089,
+      "step": 670
+    },
+    {
+      "epoch": 0.03768148066053419,
+      "grad_norm": 1.3511556386947632,
+      "learning_rate": 9.980839147765839e-07,
+      "loss": 0.7092,
+      "step": 680
+    },
+    {
+      "epoch": 0.03823562008201264,
+      "grad_norm": 1.8048819303512573,
+      "learning_rate": 9.980062426548242e-07,
+      "loss": 0.6955,
+      "step": 690
+    },
+    {
+      "epoch": 0.038789759503491075,
+      "grad_norm": 1.3200774192810059,
+      "learning_rate": 9.97927030505908e-07,
+      "loss": 0.7027,
+      "step": 700
+    },
+    {
+      "epoch": 0.038789759503491075,
+      "eval_robovqa_loss": 0.09996449947357178,
+      "step": 700
+    },
+    {
+      "epoch": 0.038789759503491075,
+      "eval_agibot_loss": 0.44685062766075134,
+      "step": 700
+    },
+    {
+      "epoch": 0.038789759503491075,
+      "eval_holoassist_loss": 0.34074699878692627,
+      "step": 700
+    },
+    {
+      "epoch": 0.038789759503491075,
+      "eval_robofail_loss": 0.20509308576583862,
+      "step": 700
+    },
+    {
+      "epoch": 0.03934389892496952,
+      "grad_norm": 1.4323095083236694,
+      "learning_rate": 9.978462785747897e-07,
+      "loss": 0.7028,
+      "step": 710
+    },
+    {
+      "epoch": 0.039898038346447964,
+      "grad_norm": 1.4774683713912964,
+      "learning_rate": 9.97763987111185e-07,
+      "loss": 0.7005,
+      "step": 720
+    },
+    {
+      "epoch": 0.04045217776792641,
+      "grad_norm": 1.4247633218765259,
+      "learning_rate": 9.976801563695715e-07,
+      "loss": 0.7016,
+      "step": 730
+    },
+    {
+      "epoch": 0.04100631718940485,
+      "grad_norm": 1.5552877187728882,
+      "learning_rate": 9.975947866091856e-07,
+      "loss": 0.702,
+      "step": 740
+    },
+    {
+      "epoch": 0.0415604566108833,
+      "grad_norm": 1.4591457843780518,
+      "learning_rate": 9.975078780940238e-07,
+      "loss": 0.7002,
+      "step": 750
+    },
+    {
+      "epoch": 0.04211459603236174,
+      "grad_norm": 1.4166990518569946,
+      "learning_rate": 9.974194310928404e-07,
+      "loss": 0.6959,
+      "step": 760
+    },
+    {
+      "epoch": 0.04266873545384019,
+      "grad_norm": 1.3892229795455933,
+      "learning_rate": 9.97329445879148e-07,
+      "loss": 0.6871,
+      "step": 770
+    },
+    {
+      "epoch": 0.04322287487531863,
+      "grad_norm": 1.4441155195236206,
+      "learning_rate": 9.97237922731215e-07,
+      "loss": 0.6926,
+      "step": 780
+    },
+    {
+      "epoch": 0.043777014296797076,
+      "grad_norm": 1.453193187713623,
+      "learning_rate": 9.971448619320667e-07,
+      "loss": 0.6897,
+      "step": 790
+    },
+    {
+      "epoch": 0.04433115371827552,
+      "grad_norm": 1.4027305841445923,
+      "learning_rate": 9.97050263769483e-07,
+      "loss": 0.6943,
+      "step": 800
+    },
+    {
+      "epoch": 0.04433115371827552,
+      "eval_robovqa_loss": 0.09529726207256317,
+      "step": 800
+    },
+    {
+      "epoch": 0.04433115371827552,
+      "eval_agibot_loss": 0.45686590671539307,
+      "step": 800
+    },
+    {
+      "epoch": 0.04433115371827552,
+      "eval_holoassist_loss": 0.33440056443214417,
+      "step": 800
+    },
+    {
+      "epoch": 0.04433115371827552,
+      "eval_robofail_loss": 0.20612764358520508,
+      "step": 800
+    },
+    {
+      "epoch": 0.044885293139753965,
+      "grad_norm": 1.3442143201828003,
+      "learning_rate": 9.969541285359973e-07,
+      "loss": 0.6971,
+      "step": 810
+    },
+    {
+      "epoch": 0.0454394325612324,
+      "grad_norm": 1.3541123867034912,
+      "learning_rate": 9.968564565288975e-07,
+      "loss": 0.689,
+      "step": 820
+    },
+    {
+      "epoch": 0.04599357198271085,
+      "grad_norm": 1.3940285444259644,
+      "learning_rate": 9.967572480502223e-07,
+      "loss": 0.6864,
+      "step": 830
+    },
+    {
+      "epoch": 0.04654771140418929,
+      "grad_norm": 1.5295149087905884,
+      "learning_rate": 9.966565034067634e-07,
+      "loss": 0.6879,
+      "step": 840
+    },
+    {
+      "epoch": 0.047101850825667736,
+      "grad_norm": 1.3263767957687378,
+      "learning_rate": 9.965542229100612e-07,
+      "loss": 0.684,
+      "step": 850
+    },
+    {
+      "epoch": 0.04765599024714618,
+      "grad_norm": 1.440314769744873,
+      "learning_rate": 9.96450406876407e-07,
+      "loss": 0.6826,
+      "step": 860
+    },
+    {
+      "epoch": 0.048210129668624625,
+      "grad_norm": 1.4764143228530884,
+      "learning_rate": 9.963450556268397e-07,
+      "loss": 0.6788,
+      "step": 870
+    },
+    {
+      "epoch": 0.04876426909010307,
+      "grad_norm": 1.2885336875915527,
+      "learning_rate": 9.96238169487146e-07,
+      "loss": 0.6846,
+      "step": 880
+    },
+    {
+      "epoch": 0.049318408511581514,
+      "grad_norm": 1.4201648235321045,
+      "learning_rate": 9.961297487878592e-07,
+      "loss": 0.6766,
+      "step": 890
+    },
+    {
+      "epoch": 0.04987254793305996,
+      "grad_norm": 1.342780351638794,
+      "learning_rate": 9.960197938642576e-07,
+      "loss": 0.6727,
+      "step": 900
+    },
+    {
+      "epoch": 0.04987254793305996,
+      "eval_robovqa_loss": 0.09362305700778961,
+      "step": 900
+    },
+    {
+      "epoch": 0.04987254793305996,
+      "eval_agibot_loss": 0.45839279890060425,
+      "step": 900
+    },
+    {
+      "epoch": 0.04987254793305996,
+      "eval_holoassist_loss": 0.32722488045692444,
+      "step": 900
+    },
+    {
+      "epoch": 0.04987254793305996,
+      "eval_robofail_loss": 0.20795565843582153,
+      "step": 900
+    },
+    {
+      "epoch": 0.0504266873545384,
+      "grad_norm": 1.3210734128952026,
+      "learning_rate": 9.959083050563641e-07,
+      "loss": 0.6693,
+      "step": 910
+    },
+    {
+      "epoch": 0.05098082677601685,
+      "grad_norm": 1.5416021347045898,
+      "learning_rate": 9.957952827089455e-07,
+      "loss": 0.6845,
+      "step": 920
+    },
+    {
+      "epoch": 0.05153496619749529,
+      "grad_norm": 1.3415138721466064,
+      "learning_rate": 9.956807271715101e-07,
+      "loss": 0.6731,
+      "step": 930
+    },
+    {
+      "epoch": 0.05208910561897374,
+      "grad_norm": 1.392008662223816,
+      "learning_rate": 9.955646387983077e-07,
+      "loss": 0.6792,
+      "step": 940
+    },
+    {
+      "epoch": 0.052643245040452175,
+      "grad_norm": 1.3660870790481567,
+      "learning_rate": 9.954470179483286e-07,
+      "loss": 0.6784,
+      "step": 950
+    },
+    {
+      "epoch": 0.05319738446193062,
+      "grad_norm": 1.519303560256958,
+      "learning_rate": 9.953278649853014e-07,
+      "loss": 0.6751,
+      "step": 960
+    },
+    {
+      "epoch": 0.053751523883409064,
+      "grad_norm": 1.4362428188323975,
+      "learning_rate": 9.952071802776932e-07,
+      "loss": 0.6734,
+      "step": 970
+    },
+    {
+      "epoch": 0.05430566330488751,
+      "grad_norm": 1.3914905786514282,
+      "learning_rate": 9.950849641987077e-07,
+      "loss": 0.68,
+      "step": 980
+    },
+    {
+      "epoch": 0.05485980272636595,
+      "grad_norm": 1.3374041318893433,
+      "learning_rate": 9.949612171262837e-07,
+      "loss": 0.6798,
+      "step": 990
+    },
+    {
+      "epoch": 0.0554139421478444,
+      "grad_norm": 1.376846432685852,
+      "learning_rate": 9.948359394430952e-07,
+      "loss": 0.6694,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0554139421478444,
+      "eval_robovqa_loss": 0.09472145885229111,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0554139421478444,
+      "eval_agibot_loss": 0.45286884903907776,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0554139421478444,
+      "eval_holoassist_loss": 0.32897913455963135,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0554139421478444,
+      "eval_robofail_loss": 0.2054339051246643,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05596808156932284,
+      "grad_norm": 1.4312970638275146,
+      "learning_rate": 9.94709131536549e-07,
+      "loss": 0.6676,
+      "step": 1010
+    },
+    {
+      "epoch": 0.056522220990801286,
+      "grad_norm": 1.2731726169586182,
+      "learning_rate": 9.945807937987837e-07,
+      "loss": 0.6692,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05707636041227973,
+      "grad_norm": 1.3272837400436401,
+      "learning_rate": 9.944509266266695e-07,
+      "loss": 0.6604,
+      "step": 1030
+    },
+    {
+      "epoch": 0.057630499833758175,
+      "grad_norm": 1.4236698150634766,
+      "learning_rate": 9.943195304218055e-07,
+      "loss": 0.6709,
+      "step": 1040
+    },
+    {
+      "epoch": 0.05818463925523662,
+      "grad_norm": 1.4647186994552612,
+      "learning_rate": 9.941866055905193e-07,
+      "loss": 0.659,
+      "step": 1050
+    },
+    {
+      "epoch": 0.058738778676715064,
+      "grad_norm": 1.2580368518829346,
+      "learning_rate": 9.940521525438657e-07,
+      "loss": 0.6605,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05929291809819351,
+      "grad_norm": 1.5111339092254639,
+      "learning_rate": 9.939161716976255e-07,
+      "loss": 0.6658,
+      "step": 1070
+    },
+    {
+      "epoch": 0.059847057519671946,
+      "grad_norm": 1.5813907384872437,
+      "learning_rate": 9.937786634723036e-07,
+      "loss": 0.6688,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06040119694115039,
+      "grad_norm": 1.3665882349014282,
+      "learning_rate": 9.936396282931285e-07,
+      "loss": 0.667,
+      "step": 1090
+    },
+    {
+      "epoch": 0.060955336362628836,
+      "grad_norm": 1.405683159828186,
+      "learning_rate": 9.934990665900507e-07,
+      "loss": 0.6665,
+      "step": 1100
+    },
+    {
+      "epoch": 0.060955336362628836,
+      "eval_robovqa_loss": 0.0921022817492485,
+      "step": 1100
+    },
+    {
+      "epoch": 0.060955336362628836,
+      "eval_agibot_loss": 0.45899081230163574,
+      "step": 1100
+    },
+    {
+      "epoch": 0.060955336362628836,
+      "eval_holoassist_loss": 0.33612215518951416,
+      "step": 1100
+    },
+    {
+      "epoch": 0.060955336362628836,
+      "eval_robofail_loss": 0.20766699314117432,
+      "step": 1100
+    },
+    {
+      "epoch": 0.06150947578410728,
+      "grad_norm": 1.2842451333999634,
+      "learning_rate": 9.93356978797741e-07,
+      "loss": 0.6655,
+      "step": 1110
+    },
+    {
+      "epoch": 0.062063615205585725,
+      "grad_norm": 1.426346778869629,
+      "learning_rate": 9.932133653555895e-07,
+      "loss": 0.6587,
+      "step": 1120
+    },
+    {
+      "epoch": 0.06261775462706418,
+      "grad_norm": 1.3993076086044312,
+      "learning_rate": 9.930682267077045e-07,
+      "loss": 0.6632,
+      "step": 1130
+    },
+    {
+      "epoch": 0.0631718940485426,
+      "grad_norm": 1.2564736604690552,
+      "learning_rate": 9.929215633029104e-07,
+      "loss": 0.6653,
+      "step": 1140
+    },
+    {
+      "epoch": 0.06372603347002105,
+      "grad_norm": 1.3381142616271973,
+      "learning_rate": 9.927733755947473e-07,
+      "loss": 0.664,
+      "step": 1150
+    },
+    {
+      "epoch": 0.0642801728914995,
+      "grad_norm": 1.4237494468688965,
+      "learning_rate": 9.926236640414683e-07,
+      "loss": 0.6559,
+      "step": 1160
+    },
+    {
+      "epoch": 0.06483431231297794,
+      "grad_norm": 1.4401249885559082,
+      "learning_rate": 9.924724291060393e-07,
+      "loss": 0.6502,
+      "step": 1170
+    },
+    {
+      "epoch": 0.06538845173445638,
+      "grad_norm": 1.3701659440994263,
+      "learning_rate": 9.923196712561372e-07,
+      "loss": 0.6579,
+      "step": 1180
+    },
+    {
+      "epoch": 0.06594259115593483,
+      "grad_norm": 1.3799238204956055,
+      "learning_rate": 9.921653909641478e-07,
+      "loss": 0.6532,
+      "step": 1190
+    },
+    {
+      "epoch": 0.06649673057741327,
+      "grad_norm": 1.3588286638259888,
+      "learning_rate": 9.920095887071654e-07,
+      "loss": 0.6587,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06649673057741327,
+      "eval_robovqa_loss": 0.0921277105808258,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06649673057741327,
+      "eval_agibot_loss": 0.46131935715675354,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06649673057741327,
+      "eval_holoassist_loss": 0.33412444591522217,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06649673057741327,
+      "eval_robofail_loss": 0.2046080082654953,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06705086999889172,
+      "grad_norm": 1.3480877876281738,
+      "learning_rate": 9.918522649669904e-07,
+      "loss": 0.6579,
+      "step": 1210
+    },
+    {
+      "epoch": 0.06760500942037016,
+      "grad_norm": 1.3995741605758667,
+      "learning_rate": 9.916934202301287e-07,
+      "loss": 0.6532,
+      "step": 1220
+    },
+    {
+      "epoch": 0.06815914884184861,
+      "grad_norm": 1.3635810613632202,
+      "learning_rate": 9.915330549877893e-07,
+      "loss": 0.6484,
+      "step": 1230
+    },
+    {
+      "epoch": 0.06871328826332705,
+      "grad_norm": 1.3947722911834717,
+      "learning_rate": 9.913711697358832e-07,
+      "loss": 0.6506,
+      "step": 1240
+    },
+    {
+      "epoch": 0.0692674276848055,
+      "grad_norm": 1.3764095306396484,
+      "learning_rate": 9.91207764975022e-07,
+      "loss": 0.6486,
+      "step": 1250
+    },
+    {
+      "epoch": 0.06982156710628394,
+      "grad_norm": 1.3961690664291382,
+      "learning_rate": 9.910428412105165e-07,
+      "loss": 0.6369,
+      "step": 1260
+    },
+    {
+      "epoch": 0.07037570652776239,
+      "grad_norm": 2.1424410343170166,
+      "learning_rate": 9.90876398952374e-07,
+      "loss": 0.6553,
+      "step": 1270
+    },
+    {
+      "epoch": 0.07092984594924083,
+      "grad_norm": 1.2376459836959839,
+      "learning_rate": 9.907084387152984e-07,
+      "loss": 0.6511,
+      "step": 1280
+    },
+    {
+      "epoch": 0.07148398537071927,
+      "grad_norm": 1.2919180393218994,
+      "learning_rate": 9.905389610186873e-07,
+      "loss": 0.6501,
+      "step": 1290
+    },
+    {
+      "epoch": 0.07203812479219772,
+      "grad_norm": 1.2956198453903198,
+      "learning_rate": 9.90367966386631e-07,
+      "loss": 0.6431,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07203812479219772,
+      "eval_robovqa_loss": 0.09373115003108978,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07203812479219772,
+      "eval_agibot_loss": 0.45393794775009155,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07203812479219772,
+      "eval_holoassist_loss": 0.3385179042816162,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07203812479219772,
+      "eval_robofail_loss": 0.20535911619663239,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07259226421367616,
+      "grad_norm": 1.3182469606399536,
+      "learning_rate": 9.901954553479114e-07,
+      "loss": 0.6464,
+      "step": 1310
+    },
+    {
+      "epoch": 0.07314640363515461,
+      "grad_norm": 1.3178037405014038,
+      "learning_rate": 9.90021428435998e-07,
+      "loss": 0.6407,
+      "step": 1320
+    },
+    {
+      "epoch": 0.07370054305663305,
+      "grad_norm": 1.35499906539917,
+      "learning_rate": 9.8984588618905e-07,
+      "loss": 0.6462,
+      "step": 1330
+    },
+    {
+      "epoch": 0.0742546824781115,
+      "grad_norm": 1.3729677200317383,
+      "learning_rate": 9.896688291499112e-07,
+      "loss": 0.6453,
+      "step": 1340
+    },
+    {
+      "epoch": 0.07480882189958994,
+      "grad_norm": 1.4650894403457642,
+      "learning_rate": 9.894902578661102e-07,
+      "loss": 0.6405,
+      "step": 1350
+    },
+    {
+      "epoch": 0.07536296132106839,
+      "grad_norm": 1.3815444707870483,
+      "learning_rate": 9.893101728898581e-07,
+      "loss": 0.6424,
+      "step": 1360
+    },
+    {
+      "epoch": 0.07591710074254683,
+      "grad_norm": 1.4542597532272339,
+      "learning_rate": 9.891285747780475e-07,
+      "loss": 0.6419,
+      "step": 1370
+    },
+    {
+      "epoch": 0.07647124016402528,
+      "grad_norm": 1.4055119752883911,
+      "learning_rate": 9.889454640922492e-07,
+      "loss": 0.644,
+      "step": 1380
+    },
+    {
+      "epoch": 0.0770253795855037,
+      "grad_norm": 1.4442410469055176,
+      "learning_rate": 9.887608413987123e-07,
+      "loss": 0.6377,
+      "step": 1390
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "grad_norm": 1.2267566919326782,
+      "learning_rate": 9.885747072683612e-07,
+      "loss": 0.634,
+      "step": 1400
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_robovqa_loss": 0.08801572024822235,
+      "step": 1400
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_agibot_loss": 0.4637523889541626,
+      "step": 1400
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_holoassist_loss": 0.3380581736564636,
+      "step": 1400
+    },
+    {
+      "epoch": 0.07757951900698215,
+      "eval_robofail_loss": 0.20991504192352295,
+      "step": 1400
+    },
+    {
+      "epoch": 0.0781336584284606,
+      "grad_norm": 1.4134372472763062,
+      "learning_rate": 9.883870622767943e-07,
+      "loss": 0.6353,
+      "step": 1410
+    },
+    {
+      "epoch": 0.07868779784993904,
+      "grad_norm": 1.353721261024475,
+      "learning_rate": 9.88197907004282e-07,
+      "loss": 0.6401,
+      "step": 1420
+    },
+    {
+      "epoch": 0.07924193727141748,
+      "grad_norm": 1.3269398212432861,
+      "learning_rate": 9.880072420357659e-07,
+      "loss": 0.6401,
+      "step": 1430
+    },
+    {
+      "epoch": 0.07979607669289593,
+      "grad_norm": 1.5019785165786743,
+      "learning_rate": 9.878150679608547e-07,
+      "loss": 0.644,
+      "step": 1440
+    },
+    {
+      "epoch": 0.08035021611437437,
+      "grad_norm": 1.3471015691757202,
+      "learning_rate": 9.876213853738255e-07,
+      "loss": 0.6385,
+      "step": 1450
+    },
+    {
+      "epoch": 0.08090435553585282,
+      "grad_norm": 1.3667032718658447,
+      "learning_rate": 9.87426194873619e-07,
+      "loss": 0.6366,
+      "step": 1460
+    },
+    {
+      "epoch": 0.08145849495733126,
+      "grad_norm": 1.3149726390838623,
+      "learning_rate": 9.872294970638394e-07,
+      "loss": 0.6386,
+      "step": 1470
+    },
+    {
+      "epoch": 0.0820126343788097,
+      "grad_norm": 1.5103143453598022,
+      "learning_rate": 9.870312925527521e-07,
+      "loss": 0.6319,
+      "step": 1480
+    },
+    {
+      "epoch": 0.08256677380028815,
+      "grad_norm": 1.5366952419281006,
+      "learning_rate": 9.868315819532821e-07,
+      "loss": 0.6321,
+      "step": 1490
+    },
+    {
+      "epoch": 0.0831209132217666,
+      "grad_norm": 1.2803966999053955,
+      "learning_rate": 9.866303658830115e-07,
+      "loss": 0.6355,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0831209132217666,
+      "eval_robovqa_loss": 0.09418221563100815,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0831209132217666,
+      "eval_agibot_loss": 0.44802021980285645,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0831209132217666,
+      "eval_holoassist_loss": 0.3368082046508789,
+      "step": 1500
+    },
+    {
+      "epoch": 0.0831209132217666,
+      "eval_robofail_loss": 0.20654596388339996,
+      "step": 1500
+    },
+    {
+      "epoch": 0.08367505264324504,
+      "grad_norm": 1.3204963207244873,
+      "learning_rate": 9.864276449641775e-07,
+      "loss": 0.6438,
+      "step": 1510
+    },
+    {
+      "epoch": 0.08422919206472348,
+      "grad_norm": 1.4203009605407715,
+      "learning_rate": 9.862234198236715e-07,
+      "loss": 0.6305,
+      "step": 1520
+    },
+    {
+      "epoch": 0.08478333148620193,
+      "grad_norm": 1.4584381580352783,
+      "learning_rate": 9.860176910930365e-07,
+      "loss": 0.6347,
+      "step": 1530
+    },
+    {
+      "epoch": 0.08533747090768037,
+      "grad_norm": 1.2317267656326294,
+      "learning_rate": 9.858104594084646e-07,
+      "loss": 0.6351,
+      "step": 1540
+    },
+    {
+      "epoch": 0.08589161032915882,
+      "grad_norm": 1.4563932418823242,
+      "learning_rate": 9.856017254107963e-07,
+      "loss": 0.6294,
+      "step": 1550
+    },
+    {
+      "epoch": 0.08644574975063726,
+      "grad_norm": 1.4380682706832886,
+      "learning_rate": 9.853914897455176e-07,
+      "loss": 0.6382,
+      "step": 1560
+    },
+    {
+      "epoch": 0.08699988917211571,
+      "grad_norm": 1.347097635269165,
+      "learning_rate": 9.851797530627578e-07,
+      "loss": 0.636,
+      "step": 1570
+    },
+    {
+      "epoch": 0.08755402859359415,
+      "grad_norm": 1.2984859943389893,
+      "learning_rate": 9.849665160172886e-07,
+      "loss": 0.6256,
+      "step": 1580
+    },
+    {
+      "epoch": 0.0881081680150726,
+      "grad_norm": 1.3869960308074951,
+      "learning_rate": 9.847517792685208e-07,
+      "loss": 0.6235,
+      "step": 1590
+    },
+    {
+      "epoch": 0.08866230743655104,
+      "grad_norm": 1.175739049911499,
+      "learning_rate": 9.84535543480504e-07,
+      "loss": 0.6337,
+      "step": 1600
+    },
+    {
+      "epoch": 0.08866230743655104,
+      "eval_robovqa_loss": 0.08883585780858994,
+      "step": 1600
+    },
+    {
+      "epoch": 0.08866230743655104,
+      "eval_agibot_loss": 0.45170658826828003,
+      "step": 1600
+    },
+    {
+      "epoch": 0.08866230743655104,
+      "eval_holoassist_loss": 0.32099705934524536,
+      "step": 1600
+    },
+    {
+      "epoch": 0.08866230743655104,
+      "eval_robofail_loss": 0.210236057639122,
+      "step": 1600
+    },
+    {
+      "epoch": 0.08921644685802949,
+      "grad_norm": 1.230870246887207,
+      "learning_rate": 9.843178093219214e-07,
+      "loss": 0.6209,
+      "step": 1610
+    },
+    {
+      "epoch": 0.08977058627950793,
+      "grad_norm": 1.228572130203247,
+      "learning_rate": 9.840985774660916e-07,
+      "loss": 0.6312,
+      "step": 1620
+    },
+    {
+      "epoch": 0.09032472570098637,
+      "grad_norm": 1.4271166324615479,
+      "learning_rate": 9.838778485909636e-07,
+      "loss": 0.6284,
+      "step": 1630
+    },
+    {
+      "epoch": 0.0908788651224648,
+      "grad_norm": 1.3865975141525269,
+      "learning_rate": 9.836556233791161e-07,
+      "loss": 0.6255,
+      "step": 1640
+    },
+    {
+      "epoch": 0.09143300454394325,
+      "grad_norm": 1.330839991569519,
+      "learning_rate": 9.834319025177557e-07,
+      "loss": 0.6355,
+      "step": 1650
+    },
+    {
+      "epoch": 0.0919871439654217,
+      "grad_norm": 1.427540898323059,
+      "learning_rate": 9.83206686698713e-07,
+      "loss": 0.6294,
+      "step": 1660
+    },
+    {
+      "epoch": 0.09254128338690014,
+      "grad_norm": 1.3937218189239502,
+      "learning_rate": 9.82979976618442e-07,
+      "loss": 0.6332,
+      "step": 1670
+    },
+    {
+      "epoch": 0.09309542280837858,
+      "grad_norm": 1.3134374618530273,
+      "learning_rate": 9.827517729780178e-07,
+      "loss": 0.6269,
+      "step": 1680
+    },
+    {
+      "epoch": 0.09364956222985703,
+      "grad_norm": 1.330039143562317,
+      "learning_rate": 9.82522076483134e-07,
+      "loss": 0.6323,
+      "step": 1690
+    },
+    {
+      "epoch": 0.09420370165133547,
+      "grad_norm": 1.3190029859542847,
+      "learning_rate": 9.822908878441003e-07,
+      "loss": 0.6181,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09420370165133547,
+      "eval_robovqa_loss": 0.08934276551008224,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09420370165133547,
+      "eval_agibot_loss": 0.4581466317176819,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09420370165133547,
+      "eval_holoassist_loss": 0.333171546459198,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09420370165133547,
+      "eval_robofail_loss": 0.20570963621139526,
+      "step": 1700
+    },
+    {
+      "epoch": 0.09475784107281392,
+      "grad_norm": 1.415811538696289,
+      "learning_rate": 9.820582077758415e-07,
+      "loss": 0.6221,
+      "step": 1710
+    },
+    {
+      "epoch": 0.09531198049429236,
+      "grad_norm": 1.3484638929367065,
+      "learning_rate": 9.818240369978936e-07,
+      "loss": 0.6222,
+      "step": 1720
+    },
+    {
+      "epoch": 0.0958661199157708,
+      "grad_norm": 1.2800657749176025,
+      "learning_rate": 9.815883762344033e-07,
+      "loss": 0.6183,
+      "step": 1730
+    },
+    {
+      "epoch": 0.09642025933724925,
+      "grad_norm": 1.3222918510437012,
+      "learning_rate": 9.813512262141241e-07,
+      "loss": 0.6261,
+      "step": 1740
+    },
+    {
+      "epoch": 0.0969743987587277,
+      "grad_norm": 1.4485194683074951,
+      "learning_rate": 9.811125876704154e-07,
+      "loss": 0.6188,
+      "step": 1750
+    },
+    {
+      "epoch": 0.09752853818020614,
+      "grad_norm": 1.2481143474578857,
+      "learning_rate": 9.808724613412395e-07,
+      "loss": 0.6285,
+      "step": 1760
+    },
+    {
+      "epoch": 0.09808267760168458,
+      "grad_norm": 1.2966569662094116,
+      "learning_rate": 9.806308479691594e-07,
+      "loss": 0.6203,
+      "step": 1770
+    },
+    {
+      "epoch": 0.09863681702316303,
+      "grad_norm": 1.2091906070709229,
+      "learning_rate": 9.803877483013368e-07,
+      "loss": 0.6214,
+      "step": 1780
+    },
+    {
+      "epoch": 0.09919095644464147,
+      "grad_norm": 1.3324321508407593,
+      "learning_rate": 9.801431630895296e-07,
+      "loss": 0.6245,
+      "step": 1790
+    },
+    {
+      "epoch": 0.09974509586611992,
+      "grad_norm": 1.2723485231399536,
+      "learning_rate": 9.798970930900894e-07,
+      "loss": 0.6159,
+      "step": 1800
+    },
+    {
+      "epoch": 0.09974509586611992,
+      "eval_robovqa_loss": 0.09191058576107025,
+      "step": 1800
+    },
+    {
+      "epoch": 0.09974509586611992,
+      "eval_agibot_loss": 0.45253506302833557,
+      "step": 1800
+    },
+    {
+      "epoch": 0.09974509586611992,
+      "eval_holoassist_loss": 0.3382225036621094,
+      "step": 1800
+    },
+    {
+      "epoch": 0.09974509586611992,
+      "eval_robofail_loss": 0.2035152018070221,
+      "step": 1800
+    },
+    {
+      "epoch": 0.10029923528759836,
+      "grad_norm": 1.2016663551330566,
+      "learning_rate": 9.796495390639594e-07,
+      "loss": 0.6227,
+      "step": 1810
+    },
+    {
+      "epoch": 0.1008533747090768,
+      "grad_norm": 1.438330888748169,
+      "learning_rate": 9.794005017766722e-07,
+      "loss": 0.6198,
+      "step": 1820
+    },
+    {
+      "epoch": 0.10140751413055525,
+      "grad_norm": 1.1787601709365845,
+      "learning_rate": 9.79149981998347e-07,
+      "loss": 0.6178,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1019616535520337,
+      "grad_norm": 1.4867693185806274,
+      "learning_rate": 9.788979805036872e-07,
+      "loss": 0.6145,
+      "step": 1840
+    },
+    {
+      "epoch": 0.10251579297351214,
+      "grad_norm": 1.4216487407684326,
+      "learning_rate": 9.786444980719788e-07,
+      "loss": 0.6215,
+      "step": 1850
+    },
+    {
+      "epoch": 0.10306993239499058,
+      "grad_norm": 1.3391119241714478,
+      "learning_rate": 9.78389535487087e-07,
+      "loss": 0.6226,
+      "step": 1860
+    },
+    {
+      "epoch": 0.10362407181646903,
+      "grad_norm": 1.3732019662857056,
+      "learning_rate": 9.781330935374544e-07,
+      "loss": 0.6028,
+      "step": 1870
+    },
+    {
+      "epoch": 0.10417821123794747,
+      "grad_norm": 1.516499400138855,
+      "learning_rate": 9.77875173016098e-07,
+      "loss": 0.6285,
+      "step": 1880
+    },
+    {
+      "epoch": 0.10473235065942592,
+      "grad_norm": 1.4075045585632324,
+      "learning_rate": 9.77615774720608e-07,
+      "loss": 0.6161,
+      "step": 1890
+    },
+    {
+      "epoch": 0.10528649008090435,
+      "grad_norm": 1.3159065246582031,
+      "learning_rate": 9.773548994531433e-07,
+      "loss": 0.6113,
+      "step": 1900
+    },
+    {
+      "epoch": 0.10528649008090435,
+      "eval_robovqa_loss": 0.08934401720762253,
+      "step": 1900
+    },
+    {
+      "epoch": 0.10528649008090435,
+      "eval_agibot_loss": 0.4520489573478699,
+      "step": 1900
+    },
+    {
+      "epoch": 0.10528649008090435,
+      "eval_holoassist_loss": 0.33469337224960327,
+      "step": 1900
+    },
+    {
+      "epoch": 0.10528649008090435,
+      "eval_robofail_loss": 0.20554986596107483,
+      "step": 1900
+    },
+    {
+      "epoch": 0.1058406295023828,
+      "grad_norm": 1.2817649841308594,
+      "learning_rate": 9.77092548020431e-07,
+      "loss": 0.622,
+      "step": 1910
+    },
+    {
+      "epoch": 0.10639476892386124,
+      "grad_norm": 1.4341367483139038,
+      "learning_rate": 9.76828721233763e-07,
+      "loss": 0.6237,
+      "step": 1920
+    },
+    {
+      "epoch": 0.10694890834533968,
+      "grad_norm": 1.2836683988571167,
+      "learning_rate": 9.765634199089928e-07,
+      "loss": 0.6214,
+      "step": 1930
+    },
+    {
+      "epoch": 0.10750304776681813,
+      "grad_norm": 1.438310980796814,
+      "learning_rate": 9.762966448665348e-07,
+      "loss": 0.6103,
+      "step": 1940
+    },
+    {
+      "epoch": 0.10805718718829657,
+      "grad_norm": 1.3244249820709229,
+      "learning_rate": 9.7602839693136e-07,
+      "loss": 0.6205,
+      "step": 1950
+    },
+    {
+      "epoch": 0.10861132660977502,
+      "grad_norm": 1.8849176168441772,
+      "learning_rate": 9.757586769329944e-07,
+      "loss": 0.6092,
+      "step": 1960
+    },
+    {
+      "epoch": 0.10916546603125346,
+      "grad_norm": 1.272394061088562,
+      "learning_rate": 9.754874857055162e-07,
+      "loss": 0.6138,
+      "step": 1970
+    },
+    {
+      "epoch": 0.1097196054527319,
+      "grad_norm": 1.414918303489685,
+      "learning_rate": 9.752148240875531e-07,
+      "loss": 0.6083,
+      "step": 1980
+    },
+    {
+      "epoch": 0.11027374487421035,
+      "grad_norm": 1.2916584014892578,
+      "learning_rate": 9.749406929222797e-07,
+      "loss": 0.614,
+      "step": 1990
+    },
+    {
+      "epoch": 0.1108278842956888,
+      "grad_norm": 1.4024957418441772,
+      "learning_rate": 9.746650930574154e-07,
+      "loss": 0.616,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1108278842956888,
+      "eval_robovqa_loss": 0.0912373960018158,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1108278842956888,
+      "eval_agibot_loss": 0.458303302526474,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1108278842956888,
+      "eval_holoassist_loss": 0.32850882411003113,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1108278842956888,
+      "eval_robofail_loss": 0.2050502747297287,
+      "step": 2000
+    },
+    {
+      "epoch": 0.11138202371716724,
+      "grad_norm": 1.4106394052505493,
+      "learning_rate": 9.74388025345221e-07,
+      "loss": 0.61,
+      "step": 2010
+    },
+    {
+      "epoch": 0.11193616313864568,
+      "grad_norm": 1.2296212911605835,
+      "learning_rate": 9.741094906424968e-07,
+      "loss": 0.6113,
+      "step": 2020
+    },
+    {
+      "epoch": 0.11249030256012413,
+      "grad_norm": 1.2755067348480225,
+      "learning_rate": 9.738294898105792e-07,
+      "loss": 0.6168,
+      "step": 2030
+    },
+    {
+      "epoch": 0.11304444198160257,
+      "grad_norm": 1.4336880445480347,
+      "learning_rate": 9.735480237153387e-07,
+      "loss": 0.617,
+      "step": 2040
+    },
+    {
+      "epoch": 0.11359858140308102,
+      "grad_norm": 1.4011168479919434,
+      "learning_rate": 9.73265093227177e-07,
+      "loss": 0.6139,
+      "step": 2050
+    },
+    {
+      "epoch": 0.11415272082455946,
+      "grad_norm": 1.3651480674743652,
+      "learning_rate": 9.72980699221024e-07,
+      "loss": 0.6067,
+      "step": 2060
+    },
+    {
+      "epoch": 0.1147068602460379,
+      "grad_norm": 1.3417402505874634,
+      "learning_rate": 9.726948425763356e-07,
+      "loss": 0.6133,
+      "step": 2070
+    },
+    {
+      "epoch": 0.11526099966751635,
+      "grad_norm": 1.2735849618911743,
+      "learning_rate": 9.724075241770908e-07,
+      "loss": 0.6097,
+      "step": 2080
+    },
+    {
+      "epoch": 0.1158151390889948,
+      "grad_norm": 1.417496681213379,
+      "learning_rate": 9.721187449117886e-07,
+      "loss": 0.6206,
+      "step": 2090
+    },
+    {
+      "epoch": 0.11636927851047324,
+      "grad_norm": 1.3894388675689697,
+      "learning_rate": 9.71828505673446e-07,
+      "loss": 0.6113,
+      "step": 2100
+    },
+    {
+      "epoch": 0.11636927851047324,
+      "eval_robovqa_loss": 0.08855857700109482,
+      "step": 2100
+    },
+    {
+      "epoch": 0.11636927851047324,
+      "eval_agibot_loss": 0.456767201423645,
+      "step": 2100
+    },
+    {
+      "epoch": 0.11636927851047324,
+      "eval_holoassist_loss": 0.32925525307655334,
+      "step": 2100
+    },
+    {
+      "epoch": 0.11636927851047324,
+      "eval_robofail_loss": 0.20366796851158142,
+      "step": 2100
+    },
+    {
+      "epoch": 0.11692341793195168,
+      "grad_norm": 1.4312773942947388,
+      "learning_rate": 9.71536807359594e-07,
+      "loss": 0.6085,
+      "step": 2110
+    },
+    {
+      "epoch": 0.11747755735343013,
+      "grad_norm": 1.2678648233413696,
+      "learning_rate": 9.712436508722767e-07,
+      "loss": 0.6039,
+      "step": 2120
+    },
+    {
+      "epoch": 0.11803169677490857,
+      "grad_norm": 1.2888058423995972,
+      "learning_rate": 9.709490371180467e-07,
+      "loss": 0.6089,
+      "step": 2130
+    },
+    {
+      "epoch": 0.11858583619638702,
+      "grad_norm": 1.35187566280365,
+      "learning_rate": 9.70652967007963e-07,
+      "loss": 0.6017,
+      "step": 2140
+    },
+    {
+      "epoch": 0.11913997561786545,
+      "grad_norm": 1.338877558708191,
+      "learning_rate": 9.70355441457589e-07,
+      "loss": 0.6081,
+      "step": 2150
+    },
+    {
+      "epoch": 0.11969411503934389,
+      "grad_norm": 1.3263859748840332,
+      "learning_rate": 9.70056461386988e-07,
+      "loss": 0.6026,
+      "step": 2160
+    },
+    {
+      "epoch": 0.12024825446082234,
+      "grad_norm": 1.244577407836914,
+      "learning_rate": 9.697560277207213e-07,
+      "loss": 0.6012,
+      "step": 2170
+    },
+    {
+      "epoch": 0.12080239388230078,
+      "grad_norm": 1.3033713102340698,
+      "learning_rate": 9.694541413878458e-07,
+      "loss": 0.6111,
+      "step": 2180
+    },
+    {
+      "epoch": 0.12135653330377923,
+      "grad_norm": 1.2658636569976807,
+      "learning_rate": 9.691508033219106e-07,
+      "loss": 0.5984,
+      "step": 2190
+    },
+    {
+      "epoch": 0.12191067272525767,
+      "grad_norm": 1.3781987428665161,
+      "learning_rate": 9.688460144609534e-07,
+      "loss": 0.6024,
+      "step": 2200
+    },
+    {
+      "epoch": 0.12191067272525767,
+      "eval_robovqa_loss": 0.08723922073841095,
+      "step": 2200
+    },
+    {
+      "epoch": 0.12191067272525767,
+      "eval_agibot_loss": 0.46067604422569275,
+      "step": 2200
+    },
+    {
+      "epoch": 0.12191067272525767,
+      "eval_holoassist_loss": 0.32930487394332886,
+      "step": 2200
+    },
+    {
+      "epoch": 0.12191067272525767,
+      "eval_robofail_loss": 0.2064329981803894,
+      "step": 2200
+    },
+    {
+      "epoch": 0.12246481214673612,
+      "grad_norm": 1.7003690004348755,
+      "learning_rate": 9.685397757474988e-07,
+      "loss": 0.602,
+      "step": 2210
+    },
+    {
+      "epoch": 0.12301895156821456,
+      "grad_norm": 1.3479636907577515,
+      "learning_rate": 9.682320881285553e-07,
+      "loss": 0.6067,
+      "step": 2220
+    },
+    {
+      "epoch": 0.123573090989693,
+      "grad_norm": 1.292395830154419,
+      "learning_rate": 9.679229525556112e-07,
+      "loss": 0.6012,
+      "step": 2230
+    },
+    {
+      "epoch": 0.12412723041117145,
+      "grad_norm": 1.2233328819274902,
+      "learning_rate": 9.676123699846327e-07,
+      "loss": 0.5996,
+      "step": 2240
+    },
+    {
+      "epoch": 0.1246813698326499,
+      "grad_norm": 1.228542447090149,
+      "learning_rate": 9.673003413760614e-07,
+      "loss": 0.6044,
+      "step": 2250
+    },
+    {
+      "epoch": 0.12523550925412835,
+      "grad_norm": 1.3216352462768555,
+      "learning_rate": 9.669868676948092e-07,
+      "loss": 0.6055,
+      "step": 2260
+    },
+    {
+      "epoch": 0.12578964867560677,
+      "grad_norm": 1.2861998081207275,
+      "learning_rate": 9.666719499102578e-07,
+      "loss": 0.5987,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1263437880970852,
+      "grad_norm": 1.3339306116104126,
+      "learning_rate": 9.663555889962543e-07,
+      "loss": 0.5983,
+      "step": 2280
+    },
+    {
+      "epoch": 0.12689792751856366,
+      "grad_norm": 1.8364509344100952,
+      "learning_rate": 9.660377859311085e-07,
+      "loss": 0.6038,
+      "step": 2290
+    },
+    {
+      "epoch": 0.1274520669400421,
+      "grad_norm": 1.27078115940094,
+      "learning_rate": 9.6571854169759e-07,
+      "loss": 0.6045,
+      "step": 2300
+    },
+    {
+      "epoch": 0.1274520669400421,
+      "eval_robovqa_loss": 0.0924566388130188,
+      "step": 2300
+    },
+    {
+      "epoch": 0.1274520669400421,
+      "eval_agibot_loss": 0.44426700472831726,
+      "step": 2300
+    },
+    {
+      "epoch": 0.1274520669400421,
+      "eval_holoassist_loss": 0.34075304865837097,
+      "step": 2300
+    },
+    {
+      "epoch": 0.1274520669400421,
+      "eval_robofail_loss": 0.20458699762821198,
+      "step": 2300
+    },
+    {
+      "epoch": 0.12800620636152055,
+      "grad_norm": 1.322947382926941,
+      "learning_rate": 9.65397857282925e-07,
+      "loss": 0.6068,
+      "step": 2310
+    },
+    {
+      "epoch": 0.128560345782999,
+      "grad_norm": 1.244891881942749,
+      "learning_rate": 9.65075733678793e-07,
+      "loss": 0.603,
+      "step": 2320
+    },
+    {
+      "epoch": 0.12911448520447744,
+      "grad_norm": 1.3860280513763428,
+      "learning_rate": 9.647521718813243e-07,
+      "loss": 0.6041,
+      "step": 2330
+    },
+    {
+      "epoch": 0.12966862462595588,
+      "grad_norm": 1.2473706007003784,
+      "learning_rate": 9.644271728910968e-07,
+      "loss": 0.6042,
+      "step": 2340
+    },
+    {
+      "epoch": 0.13022276404743433,
+      "grad_norm": 1.2597438097000122,
+      "learning_rate": 9.641007377131327e-07,
+      "loss": 0.6043,
+      "step": 2350
+    },
+    {
+      "epoch": 0.13077690346891277,
+      "grad_norm": 1.3021448850631714,
+      "learning_rate": 9.63772867356895e-07,
+      "loss": 0.5952,
+      "step": 2360
+    },
+    {
+      "epoch": 0.13133104289039121,
+      "grad_norm": 1.302114486694336,
+      "learning_rate": 9.634435628362854e-07,
+      "loss": 0.5963,
+      "step": 2370
+    },
+    {
+      "epoch": 0.13188518231186966,
+      "grad_norm": 1.4521567821502686,
+      "learning_rate": 9.631128251696403e-07,
+      "loss": 0.5997,
+      "step": 2380
+    },
+    {
+      "epoch": 0.1324393217333481,
+      "grad_norm": 1.416270136833191,
+      "learning_rate": 9.627806553797278e-07,
+      "loss": 0.5975,
+      "step": 2390
+    },
+    {
+      "epoch": 0.13299346115482655,
+      "grad_norm": 1.3267556428909302,
+      "learning_rate": 9.624470544937451e-07,
+      "loss": 0.601,
+      "step": 2400
+    },
+    {
+      "epoch": 0.13299346115482655,
+      "eval_robovqa_loss": 0.08818978071212769,
+      "step": 2400
+    },
+    {
+      "epoch": 0.13299346115482655,
+      "eval_agibot_loss": 0.44986793398857117,
+      "step": 2400
+    },
+    {
+      "epoch": 0.13299346115482655,
+      "eval_holoassist_loss": 0.3259875178337097,
+      "step": 2400
+    },
+    {
+      "epoch": 0.13299346115482655,
+      "eval_robofail_loss": 0.20547394454479218,
+      "step": 2400
+    },
+    {
+      "epoch": 0.133547600576305,
+      "grad_norm": 1.9141323566436768,
+      "learning_rate": 9.621120235433144e-07,
+      "loss": 0.5976,
+      "step": 2410
+    },
+    {
+      "epoch": 0.13410173999778344,
+      "grad_norm": 1.317747950553894,
+      "learning_rate": 9.617755635644808e-07,
+      "loss": 0.5965,
+      "step": 2420
+    },
+    {
+      "epoch": 0.13465587941926188,
+      "grad_norm": 1.287738561630249,
+      "learning_rate": 9.61437675597708e-07,
+      "loss": 0.592,
+      "step": 2430
+    },
+    {
+      "epoch": 0.13521001884074033,
+      "grad_norm": 1.2779525518417358,
+      "learning_rate": 9.610983606878754e-07,
+      "loss": 0.5923,
+      "step": 2440
+    },
+    {
+      "epoch": 0.13576415826221877,
+      "grad_norm": 1.3654398918151855,
+      "learning_rate": 9.607576198842762e-07,
+      "loss": 0.5979,
+      "step": 2450
+    },
+    {
+      "epoch": 0.13631829768369721,
+      "grad_norm": 1.2556051015853882,
+      "learning_rate": 9.604154542406117e-07,
+      "loss": 0.6002,
+      "step": 2460
+    },
+    {
+      "epoch": 0.13687243710517566,
+      "grad_norm": 1.3456052541732788,
+      "learning_rate": 9.600718648149898e-07,
+      "loss": 0.6055,
+      "step": 2470
+    },
+    {
+      "epoch": 0.1374265765266541,
+      "grad_norm": 1.3115618228912354,
+      "learning_rate": 9.597268526699218e-07,
+      "loss": 0.6019,
+      "step": 2480
+    },
+    {
+      "epoch": 0.13798071594813255,
+      "grad_norm": 1.2468616962432861,
+      "learning_rate": 9.593804188723177e-07,
+      "loss": 0.5961,
+      "step": 2490
+    },
+    {
+      "epoch": 0.138534855369611,
+      "grad_norm": 1.3139430284500122,
+      "learning_rate": 9.590325644934848e-07,
+      "loss": 0.5804,
+      "step": 2500
+    },
+    {
+      "epoch": 0.138534855369611,
+      "eval_robovqa_loss": 0.08670637011528015,
+      "step": 2500
+    },
+    {
+      "epoch": 0.138534855369611,
+      "eval_agibot_loss": 0.4556445777416229,
+      "step": 2500
+    },
+    {
+      "epoch": 0.138534855369611,
+      "eval_holoassist_loss": 0.3279184103012085,
+      "step": 2500
+    },
+    {
+      "epoch": 0.138534855369611,
+      "eval_robofail_loss": 0.20952369272708893,
+      "step": 2500
+    },
+    {
+      "epoch": 0.13908899479108944,
+      "grad_norm": 1.3960744142532349,
+      "learning_rate": 9.586832906091225e-07,
+      "loss": 0.5969,
+      "step": 2510
+    },
+    {
+      "epoch": 0.13964313421256788,
+      "grad_norm": 1.2717556953430176,
+      "learning_rate": 9.5833259829932e-07,
+      "loss": 0.5967,
+      "step": 2520
+    },
+    {
+      "epoch": 0.14019727363404633,
+      "grad_norm": 1.3038357496261597,
+      "learning_rate": 9.579804886485534e-07,
+      "loss": 0.6006,
+      "step": 2530
+    },
+    {
+      "epoch": 0.14075141305552477,
+      "grad_norm": 1.746801495552063,
+      "learning_rate": 9.576269627456811e-07,
+      "loss": 0.5881,
+      "step": 2540
+    },
+    {
+      "epoch": 0.14130555247700322,
+      "grad_norm": 1.325109839439392,
+      "learning_rate": 9.572720216839418e-07,
+      "loss": 0.5968,
+      "step": 2550
+    },
+    {
+      "epoch": 0.14185969189848166,
+      "grad_norm": 1.332337498664856,
+      "learning_rate": 9.569156665609497e-07,
+      "loss": 0.5884,
+      "step": 2560
+    },
+    {
+      "epoch": 0.1424138313199601,
+      "grad_norm": 1.4073364734649658,
+      "learning_rate": 9.565578984786922e-07,
+      "loss": 0.5986,
+      "step": 2570
+    },
+    {
+      "epoch": 0.14296797074143855,
+      "grad_norm": 1.3388519287109375,
+      "learning_rate": 9.561987185435259e-07,
+      "loss": 0.5987,
+      "step": 2580
+    },
+    {
+      "epoch": 0.143522110162917,
+      "grad_norm": 1.3040497303009033,
+      "learning_rate": 9.558381278661737e-07,
+      "loss": 0.5911,
+      "step": 2590
+    },
+    {
+      "epoch": 0.14407624958439544,
+      "grad_norm": 1.3956358432769775,
+      "learning_rate": 9.554761275617208e-07,
+      "loss": 0.5976,
+      "step": 2600
+    },
+    {
+      "epoch": 0.14407624958439544,
+      "eval_robovqa_loss": 0.08845953643321991,
+      "step": 2600
+    },
+    {
+      "epoch": 0.14407624958439544,
+      "eval_agibot_loss": 0.4540943205356598,
+      "step": 2600
+    },
+    {
+      "epoch": 0.14407624958439544,
+      "eval_holoassist_loss": 0.3187803030014038,
+      "step": 2600
+    },
+    {
+      "epoch": 0.14407624958439544,
+      "eval_robofail_loss": 0.20738860964775085,
+      "step": 2600
+    },
+    {
+      "epoch": 0.14463038900587388,
+      "grad_norm": 1.3461800813674927,
+      "learning_rate": 9.551127187496118e-07,
+      "loss": 0.6002,
+      "step": 2610
+    },
+    {
+      "epoch": 0.14518452842735233,
+      "grad_norm": 1.2954096794128418,
+      "learning_rate": 9.547479025536464e-07,
+      "loss": 0.5981,
+      "step": 2620
+    },
+    {
+      "epoch": 0.14573866784883077,
+      "grad_norm": 1.1929103136062622,
+      "learning_rate": 9.54381680101977e-07,
+      "loss": 0.5903,
+      "step": 2630
+    },
+    {
+      "epoch": 0.14629280727030922,
+      "grad_norm": 1.445796251296997,
+      "learning_rate": 9.540140525271046e-07,
+      "loss": 0.5911,
+      "step": 2640
+    },
+    {
+      "epoch": 0.14684694669178766,
+      "grad_norm": 2.032557964324951,
+      "learning_rate": 9.536450209658751e-07,
+      "loss": 0.5912,
+      "step": 2650
+    },
+    {
+      "epoch": 0.1474010861132661,
+      "grad_norm": 1.4054884910583496,
+      "learning_rate": 9.532745865594765e-07,
+      "loss": 0.587,
+      "step": 2660
+    },
+    {
+      "epoch": 0.14795522553474455,
+      "grad_norm": 1.4399564266204834,
+      "learning_rate": 9.529027504534344e-07,
+      "loss": 0.5891,
+      "step": 2670
+    },
+    {
+      "epoch": 0.148509364956223,
+      "grad_norm": 1.3753281831741333,
+      "learning_rate": 9.525295137976095e-07,
+      "loss": 0.5907,
+      "step": 2680
+    },
+    {
+      "epoch": 0.14906350437770144,
+      "grad_norm": 1.2290199995040894,
+      "learning_rate": 9.521548777461932e-07,
+      "loss": 0.5809,
+      "step": 2690
+    },
+    {
+      "epoch": 0.14961764379917988,
+      "grad_norm": 1.2867846488952637,
+      "learning_rate": 9.517788434577047e-07,
+      "loss": 0.5905,
+      "step": 2700
+    },
+    {
+      "epoch": 0.14961764379917988,
+      "eval_robovqa_loss": 0.08541330695152283,
+      "step": 2700
+    },
+    {
+      "epoch": 0.14961764379917988,
+      "eval_agibot_loss": 0.46621212363243103,
+      "step": 2700
+    },
+    {
+      "epoch": 0.14961764379917988,
+      "eval_holoassist_loss": 0.3303990960121155,
+      "step": 2700
+    },
+    {
+      "epoch": 0.14961764379917988,
+      "eval_robofail_loss": 0.20784221589565277,
+      "step": 2700
+    },
+    {
+      "epoch": 0.15017178322065833,
+      "grad_norm": 1.3958162069320679,
+      "learning_rate": 9.514014120949867e-07,
+      "loss": 0.5916,
+      "step": 2710
+    },
+    {
+      "epoch": 0.15072592264213677,
+      "grad_norm": 1.309119701385498,
+      "learning_rate": 9.510225848252022e-07,
+      "loss": 0.5941,
+      "step": 2720
+    },
+    {
+      "epoch": 0.15128006206361522,
+      "grad_norm": 1.2587674856185913,
+      "learning_rate": 9.506423628198315e-07,
+      "loss": 0.594,
+      "step": 2730
+    },
+    {
+      "epoch": 0.15183420148509366,
+      "grad_norm": 1.3694630861282349,
+      "learning_rate": 9.502607472546673e-07,
+      "loss": 0.5902,
+      "step": 2740
+    },
+    {
+      "epoch": 0.1523883409065721,
+      "grad_norm": 1.4067084789276123,
+      "learning_rate": 9.49877739309812e-07,
+      "loss": 0.588,
+      "step": 2750
+    },
+    {
+      "epoch": 0.15294248032805055,
+      "grad_norm": 1.2654592990875244,
+      "learning_rate": 9.494933401696736e-07,
+      "loss": 0.5927,
+      "step": 2760
+    },
+    {
+      "epoch": 0.153496619749529,
+      "grad_norm": 1.265487790107727,
+      "learning_rate": 9.491075510229622e-07,
+      "loss": 0.593,
+      "step": 2770
+    },
+    {
+      "epoch": 0.1540507591710074,
+      "grad_norm": 1.2581431865692139,
+      "learning_rate": 9.487203730626869e-07,
+      "loss": 0.5934,
+      "step": 2780
+    },
+    {
+      "epoch": 0.15460489859248586,
+      "grad_norm": 1.429948329925537,
+      "learning_rate": 9.483318074861508e-07,
+      "loss": 0.5888,
+      "step": 2790
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "grad_norm": 1.3362661600112915,
+      "learning_rate": 9.479418554949483e-07,
+      "loss": 0.5874,
+      "step": 2800
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_robovqa_loss": 0.0865456685423851,
+      "step": 2800
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_agibot_loss": 0.4451712667942047,
+      "step": 2800
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_holoassist_loss": 0.32482436299324036,
+      "step": 2800
+    },
+    {
+      "epoch": 0.1551590380139643,
+      "eval_robofail_loss": 0.207450732588768,
+      "step": 2800
+    },
+    {
+      "epoch": 0.15571317743544275,
+      "grad_norm": 1.4643681049346924,
+      "learning_rate": 9.475505182949614e-07,
+      "loss": 0.5927,
+      "step": 2810
+    },
+    {
+      "epoch": 0.1562673168569212,
+      "grad_norm": 1.2370710372924805,
+      "learning_rate": 9.471577970963555e-07,
+      "loss": 0.59,
+      "step": 2820
+    },
+    {
+      "epoch": 0.15682145627839963,
+      "grad_norm": 1.27611243724823,
+      "learning_rate": 9.467636931135757e-07,
+      "loss": 0.5906,
+      "step": 2830
+    },
+    {
+      "epoch": 0.15737559569987808,
+      "grad_norm": 1.4108917713165283,
+      "learning_rate": 9.463682075653432e-07,
+      "loss": 0.5875,
+      "step": 2840
+    },
+    {
+      "epoch": 0.15792973512135652,
+      "grad_norm": 1.3463376760482788,
+      "learning_rate": 9.459713416746519e-07,
+      "loss": 0.5906,
+      "step": 2850
+    },
+    {
+      "epoch": 0.15848387454283497,
+      "grad_norm": 1.347978949546814,
+      "learning_rate": 9.455730966687641e-07,
+      "loss": 0.5806,
+      "step": 2860
+    },
+    {
+      "epoch": 0.1590380139643134,
+      "grad_norm": 1.2450371980667114,
+      "learning_rate": 9.451734737792066e-07,
+      "loss": 0.5844,
+      "step": 2870
+    },
+    {
+      "epoch": 0.15959215338579186,
+      "grad_norm": 1.3256865739822388,
+      "learning_rate": 9.447724742417673e-07,
+      "loss": 0.5833,
+      "step": 2880
+    },
+    {
+      "epoch": 0.1601462928072703,
+      "grad_norm": 1.4263298511505127,
+      "learning_rate": 9.443700992964914e-07,
+      "loss": 0.5856,
+      "step": 2890
+    },
+    {
+      "epoch": 0.16070043222874875,
+      "grad_norm": 1.3590703010559082,
+      "learning_rate": 9.439663501876771e-07,
+      "loss": 0.5912,
+      "step": 2900
+    },
+    {
+      "epoch": 0.16070043222874875,
+      "eval_robovqa_loss": 0.08468560129404068,
+      "step": 2900
+    },
+    {
+      "epoch": 0.16070043222874875,
+      "eval_agibot_loss": 0.4590352177619934,
+      "step": 2900
+    },
+    {
+      "epoch": 0.16070043222874875,
+      "eval_holoassist_loss": 0.322124719619751,
+      "step": 2900
+    },
+    {
+      "epoch": 0.16070043222874875,
+      "eval_robofail_loss": 0.20593470335006714,
+      "step": 2900
+    },
+    {
+      "epoch": 0.1612545716502272,
+      "grad_norm": 1.2781622409820557,
+      "learning_rate": 9.435612281638723e-07,
+      "loss": 0.5872,
+      "step": 2910
+    },
+    {
+      "epoch": 0.16180871107170564,
+      "grad_norm": 1.3716912269592285,
+      "learning_rate": 9.431547344778703e-07,
+      "loss": 0.5908,
+      "step": 2920
+    },
+    {
+      "epoch": 0.16236285049318408,
+      "grad_norm": 1.3691059350967407,
+      "learning_rate": 9.427468703867062e-07,
+      "loss": 0.5875,
+      "step": 2930
+    },
+    {
+      "epoch": 0.16291698991466252,
+      "grad_norm": 1.3187572956085205,
+      "learning_rate": 9.423376371516529e-07,
+      "loss": 0.5841,
+      "step": 2940
+    },
+    {
+      "epoch": 0.16347112933614097,
+      "grad_norm": 1.3411449193954468,
+      "learning_rate": 9.419270360382172e-07,
+      "loss": 0.5883,
+      "step": 2950
+    },
+    {
+      "epoch": 0.1640252687576194,
+      "grad_norm": 1.3425034284591675,
+      "learning_rate": 9.415150683161361e-07,
+      "loss": 0.5805,
+      "step": 2960
+    },
+    {
+      "epoch": 0.16457940817909786,
+      "grad_norm": 1.3590586185455322,
+      "learning_rate": 9.411017352593722e-07,
+      "loss": 0.5783,
+      "step": 2970
+    },
+    {
+      "epoch": 0.1651335476005763,
+      "grad_norm": 1.2968947887420654,
+      "learning_rate": 9.406870381461107e-07,
+      "loss": 0.5817,
+      "step": 2980
+    },
+    {
+      "epoch": 0.16568768702205475,
+      "grad_norm": 1.28688383102417,
+      "learning_rate": 9.402709782587549e-07,
+      "loss": 0.5851,
+      "step": 2990
+    },
+    {
+      "epoch": 0.1662418264435332,
+      "grad_norm": 1.362794280052185,
+      "learning_rate": 9.39853556883922e-07,
+      "loss": 0.58,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1662418264435332,
+      "eval_robovqa_loss": 0.08658679574728012,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1662418264435332,
+      "eval_agibot_loss": 0.44846847653388977,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1662418264435332,
+      "eval_holoassist_loss": 0.34456971287727356,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1662418264435332,
+      "eval_robofail_loss": 0.20725350081920624,
+      "step": 3000
+    },
+    {
+      "epoch": 0.16679596586501164,
+      "grad_norm": 1.3605213165283203,
+      "learning_rate": 9.394347753124402e-07,
+      "loss": 0.5766,
+      "step": 3010
+    },
+    {
+      "epoch": 0.16735010528649008,
+      "grad_norm": 1.1409894227981567,
+      "learning_rate": 9.390146348393429e-07,
+      "loss": 0.5844,
+      "step": 3020
+    },
+    {
+      "epoch": 0.16790424470796853,
+      "grad_norm": 1.2455331087112427,
+      "learning_rate": 9.385931367638667e-07,
+      "loss": 0.5821,
+      "step": 3030
+    },
+    {
+      "epoch": 0.16845838412944697,
+      "grad_norm": 1.2701823711395264,
+      "learning_rate": 9.381702823894457e-07,
+      "loss": 0.5839,
+      "step": 3040
+    },
+    {
+      "epoch": 0.16901252355092541,
+      "grad_norm": 1.4699534177780151,
+      "learning_rate": 9.377460730237088e-07,
+      "loss": 0.5736,
+      "step": 3050
+    },
+    {
+      "epoch": 0.16956666297240386,
+      "grad_norm": 1.2889773845672607,
+      "learning_rate": 9.373205099784746e-07,
+      "loss": 0.58,
+      "step": 3060
+    },
+    {
+      "epoch": 0.1701208023938823,
+      "grad_norm": 1.3061161041259766,
+      "learning_rate": 9.368935945697483e-07,
+      "loss": 0.5833,
+      "step": 3070
+    },
+    {
+      "epoch": 0.17067494181536075,
+      "grad_norm": 1.3538870811462402,
+      "learning_rate": 9.364653281177164e-07,
+      "loss": 0.5887,
+      "step": 3080
+    },
+    {
+      "epoch": 0.1712290812368392,
+      "grad_norm": 1.3636571168899536,
+      "learning_rate": 9.360357119467439e-07,
+      "loss": 0.5807,
+      "step": 3090
+    },
+    {
+      "epoch": 0.17178322065831764,
+      "grad_norm": 1.3242095708847046,
+      "learning_rate": 9.356047473853694e-07,
+      "loss": 0.5771,
+      "step": 3100
+    },
+    {
+      "epoch": 0.17178322065831764,
+      "eval_robovqa_loss": 0.08484404534101486,
+      "step": 3100
+    },
+    {
+      "epoch": 0.17178322065831764,
+      "eval_agibot_loss": 0.4494175314903259,
+      "step": 3100
+    },
+    {
+      "epoch": 0.17178322065831764,
+      "eval_holoassist_loss": 0.32661810517311096,
+      "step": 3100
+    },
+    {
+      "epoch": 0.17178322065831764,
+      "eval_robofail_loss": 0.20312950015068054,
+      "step": 3100
+    },
+    {
+      "epoch": 0.17233736007979608,
+      "grad_norm": 1.2890716791152954,
+      "learning_rate": 9.351724357663018e-07,
+      "loss": 0.5814,
+      "step": 3110
+    },
+    {
+      "epoch": 0.17289149950127453,
+      "grad_norm": 1.3634285926818848,
+      "learning_rate": 9.347387784264147e-07,
+      "loss": 0.5791,
+      "step": 3120
+    },
+    {
+      "epoch": 0.17344563892275297,
+      "grad_norm": 1.360221028327942,
+      "learning_rate": 9.343037767067438e-07,
+      "loss": 0.5782,
+      "step": 3130
+    },
+    {
+      "epoch": 0.17399977834423141,
+      "grad_norm": 1.3228232860565186,
+      "learning_rate": 9.33867431952482e-07,
+      "loss": 0.5753,
+      "step": 3140
+    },
+    {
+      "epoch": 0.17455391776570986,
+      "grad_norm": 1.5344030857086182,
+      "learning_rate": 9.334297455129754e-07,
+      "loss": 0.5722,
+      "step": 3150
+    },
+    {
+      "epoch": 0.1751080571871883,
+      "grad_norm": 1.3420181274414062,
+      "learning_rate": 9.329907187417192e-07,
+      "loss": 0.577,
+      "step": 3160
+    },
+    {
+      "epoch": 0.17566219660866675,
+      "grad_norm": 1.2308703660964966,
+      "learning_rate": 9.325503529963531e-07,
+      "loss": 0.5751,
+      "step": 3170
+    },
+    {
+      "epoch": 0.1762163360301452,
+      "grad_norm": 1.24155855178833,
+      "learning_rate": 9.321086496386577e-07,
+      "loss": 0.5821,
+      "step": 3180
+    },
+    {
+      "epoch": 0.17677047545162364,
+      "grad_norm": 1.36679208278656,
+      "learning_rate": 9.316656100345501e-07,
+      "loss": 0.5767,
+      "step": 3190
+    },
+    {
+      "epoch": 0.17732461487310208,
+      "grad_norm": 1.3114545345306396,
+      "learning_rate": 9.312212355540791e-07,
+      "loss": 0.581,
+      "step": 3200
+    },
+    {
+      "epoch": 0.17732461487310208,
+      "eval_robovqa_loss": 0.08254455775022507,
+      "step": 3200
+    },
+    {
+      "epoch": 0.17732461487310208,
+      "eval_agibot_loss": 0.4544808566570282,
+      "step": 3200
+    },
+    {
+      "epoch": 0.17732461487310208,
+      "eval_holoassist_loss": 0.325687974691391,
+      "step": 3200
+    },
+    {
+      "epoch": 0.17732461487310208,
+      "eval_robofail_loss": 0.20689500868320465,
+      "step": 3200
+    },
+    {
+      "epoch": 0.17787875429458053,
+      "grad_norm": 1.1906070709228516,
+      "learning_rate": 9.307755275714223e-07,
+      "loss": 0.5714,
+      "step": 3210
+    },
+    {
+      "epoch": 0.17843289371605897,
+      "grad_norm": 1.2607958316802979,
+      "learning_rate": 9.303284874648801e-07,
+      "loss": 0.5759,
+      "step": 3220
+    },
+    {
+      "epoch": 0.17898703313753742,
+      "grad_norm": 1.282989501953125,
+      "learning_rate": 9.298801166168726e-07,
+      "loss": 0.5723,
+      "step": 3230
+    },
+    {
+      "epoch": 0.17954117255901586,
+      "grad_norm": 1.4264072179794312,
+      "learning_rate": 9.294304164139355e-07,
+      "loss": 0.5764,
+      "step": 3240
+    },
+    {
+      "epoch": 0.1800953119804943,
+      "grad_norm": 1.227779746055603,
+      "learning_rate": 9.289793882467152e-07,
+      "loss": 0.5871,
+      "step": 3250
+    },
+    {
+      "epoch": 0.18064945140197275,
+      "grad_norm": 1.3435378074645996,
+      "learning_rate": 9.28527033509964e-07,
+      "loss": 0.5753,
+      "step": 3260
+    },
+    {
+      "epoch": 0.1812035908234512,
+      "grad_norm": 1.2623144388198853,
+      "learning_rate": 9.280733536025376e-07,
+      "loss": 0.5728,
+      "step": 3270
+    },
+    {
+      "epoch": 0.1817577302449296,
+      "grad_norm": 1.3590728044509888,
+      "learning_rate": 9.276183499273886e-07,
+      "loss": 0.5837,
+      "step": 3280
+    },
+    {
+      "epoch": 0.18231186966640806,
+      "grad_norm": 1.3204982280731201,
+      "learning_rate": 9.271620238915641e-07,
+      "loss": 0.5723,
+      "step": 3290
+    },
+    {
+      "epoch": 0.1828660090878865,
+      "grad_norm": 1.354812741279602,
+      "learning_rate": 9.267043769061995e-07,
+      "loss": 0.5748,
+      "step": 3300
+    },
+    {
+      "epoch": 0.1828660090878865,
+      "eval_robovqa_loss": 0.08465243130922318,
+      "step": 3300
+    },
+    {
+      "epoch": 0.1828660090878865,
+      "eval_agibot_loss": 0.4504750072956085,
+      "step": 3300
+    },
+    {
+      "epoch": 0.1828660090878865,
+      "eval_holoassist_loss": 0.32552826404571533,
+      "step": 3300
+    },
+    {
+      "epoch": 0.1828660090878865,
+      "eval_robofail_loss": 0.20609597861766815,
+      "step": 3300
+    },
+    {
+      "epoch": 0.18342014850936494,
+      "grad_norm": 1.3790740966796875,
+      "learning_rate": 9.262454103865158e-07,
+      "loss": 0.572,
+      "step": 3310
+    },
+    {
+      "epoch": 0.1839742879308434,
+      "grad_norm": 1.2772265672683716,
+      "learning_rate": 9.257851257518145e-07,
+      "loss": 0.5761,
+      "step": 3320
+    },
+    {
+      "epoch": 0.18452842735232183,
+      "grad_norm": 1.3541902303695679,
+      "learning_rate": 9.253235244254728e-07,
+      "loss": 0.5802,
+      "step": 3330
+    },
+    {
+      "epoch": 0.18508256677380028,
+      "grad_norm": 1.2396548986434937,
+      "learning_rate": 9.248606078349399e-07,
+      "loss": 0.5748,
+      "step": 3340
+    },
+    {
+      "epoch": 0.18563670619527872,
+      "grad_norm": 1.2832047939300537,
+      "learning_rate": 9.24396377411732e-07,
+      "loss": 0.5776,
+      "step": 3350
+    },
+    {
+      "epoch": 0.18619084561675717,
+      "grad_norm": 1.3062899112701416,
+      "learning_rate": 9.239308345914289e-07,
+      "loss": 0.5783,
+      "step": 3360
+    },
+    {
+      "epoch": 0.1867449850382356,
+      "grad_norm": 1.224029541015625,
+      "learning_rate": 9.234639808136679e-07,
+      "loss": 0.573,
+      "step": 3370
+    },
+    {
+      "epoch": 0.18729912445971406,
+      "grad_norm": 1.316295862197876,
+      "learning_rate": 9.229958175221409e-07,
+      "loss": 0.5734,
+      "step": 3380
+    },
+    {
+      "epoch": 0.1878532638811925,
+      "grad_norm": 1.357491135597229,
+      "learning_rate": 9.225263461645891e-07,
+      "loss": 0.5683,
+      "step": 3390
+    },
+    {
+      "epoch": 0.18840740330267094,
+      "grad_norm": 1.2591180801391602,
+      "learning_rate": 9.220555681927991e-07,
+      "loss": 0.5714,
+      "step": 3400
+    },
+    {
+      "epoch": 0.18840740330267094,
+      "eval_robovqa_loss": 0.08414895087480545,
+      "step": 3400
+    },
+    {
+      "epoch": 0.18840740330267094,
+      "eval_agibot_loss": 0.4573032855987549,
+      "step": 3400
+    },
+    {
+      "epoch": 0.18840740330267094,
+      "eval_holoassist_loss": 0.31745555996894836,
+      "step": 3400
+    },
+    {
+      "epoch": 0.18840740330267094,
+      "eval_robofail_loss": 0.20637179911136627,
+      "step": 3400
+    },
+    {
+      "epoch": 0.1889615427241494,
+      "grad_norm": 1.288508653640747,
+      "learning_rate": 9.215834850625972e-07,
+      "loss": 0.576,
+      "step": 3410
+    },
+    {
+      "epoch": 0.18951568214562783,
+      "grad_norm": 1.3303554058074951,
+      "learning_rate": 9.211100982338468e-07,
+      "loss": 0.5728,
+      "step": 3420
+    },
+    {
+      "epoch": 0.19006982156710628,
+      "grad_norm": 1.2700514793395996,
+      "learning_rate": 9.20635409170442e-07,
+      "loss": 0.572,
+      "step": 3430
+    },
+    {
+      "epoch": 0.19062396098858472,
+      "grad_norm": 1.2379971742630005,
+      "learning_rate": 9.201594193403047e-07,
+      "loss": 0.5658,
+      "step": 3440
+    },
+    {
+      "epoch": 0.19117810041006317,
+      "grad_norm": 1.271378755569458,
+      "learning_rate": 9.196821302153782e-07,
+      "loss": 0.5707,
+      "step": 3450
+    },
+    {
+      "epoch": 0.1917322398315416,
+      "grad_norm": 1.2408870458602905,
+      "learning_rate": 9.192035432716249e-07,
+      "loss": 0.5708,
+      "step": 3460
+    },
+    {
+      "epoch": 0.19228637925302006,
+      "grad_norm": 1.3406062126159668,
+      "learning_rate": 9.187236599890195e-07,
+      "loss": 0.573,
+      "step": 3470
+    },
+    {
+      "epoch": 0.1928405186744985,
+      "grad_norm": 1.343049168586731,
+      "learning_rate": 9.182424818515467e-07,
+      "loss": 0.5669,
+      "step": 3480
+    },
+    {
+      "epoch": 0.19339465809597695,
+      "grad_norm": 1.332615613937378,
+      "learning_rate": 9.17760010347194e-07,
+      "loss": 0.5706,
+      "step": 3490
+    },
+    {
+      "epoch": 0.1939487975174554,
+      "grad_norm": 1.3356369733810425,
+      "learning_rate": 9.172762469679493e-07,
+      "loss": 0.5729,
+      "step": 3500
+    },
+    {
+      "epoch": 0.1939487975174554,
+      "eval_robovqa_loss": 0.09070686995983124,
+      "step": 3500
+    },
+    {
+      "epoch": 0.1939487975174554,
+      "eval_agibot_loss": 0.4457252621650696,
+      "step": 3500
+    },
+    {
+      "epoch": 0.1939487975174554,
+      "eval_holoassist_loss": 0.34745416045188904,
+      "step": 3500
+    },
+    {
+      "epoch": 0.1939487975174554,
+      "eval_robofail_loss": 0.20256978273391724,
+      "step": 3500
+    },
+    {
+      "epoch": 0.19450293693893383,
+      "grad_norm": 1.2404258251190186,
+      "learning_rate": 9.167911932097956e-07,
+      "loss": 0.5753,
+      "step": 3510
+    },
+    {
+      "epoch": 0.19505707636041228,
+      "grad_norm": 1.216766357421875,
+      "learning_rate": 9.163048505727058e-07,
+      "loss": 0.5708,
+      "step": 3520
+    },
+    {
+      "epoch": 0.19561121578189072,
+      "grad_norm": 1.3193471431732178,
+      "learning_rate": 9.158172205606387e-07,
+      "loss": 0.5726,
+      "step": 3530
+    },
+    {
+      "epoch": 0.19616535520336917,
+      "grad_norm": 1.3537144660949707,
+      "learning_rate": 9.153283046815342e-07,
+      "loss": 0.5721,
+      "step": 3540
+    },
+    {
+      "epoch": 0.1967194946248476,
+      "grad_norm": 1.2511773109436035,
+      "learning_rate": 9.148381044473085e-07,
+      "loss": 0.5778,
+      "step": 3550
+    },
+    {
+      "epoch": 0.19727363404632606,
+      "grad_norm": 1.3676213026046753,
+      "learning_rate": 9.143466213738496e-07,
+      "loss": 0.5726,
+      "step": 3560
+    },
+    {
+      "epoch": 0.1978277734678045,
+      "grad_norm": 1.356564998626709,
+      "learning_rate": 9.138538569810125e-07,
+      "loss": 0.5718,
+      "step": 3570
+    },
+    {
+      "epoch": 0.19838191288928295,
+      "grad_norm": 1.237135887145996,
+      "learning_rate": 9.133598127926145e-07,
+      "loss": 0.5706,
+      "step": 3580
+    },
+    {
+      "epoch": 0.1989360523107614,
+      "grad_norm": 1.248610019683838,
+      "learning_rate": 9.128644903364306e-07,
+      "loss": 0.5762,
+      "step": 3590
+    },
+    {
+      "epoch": 0.19949019173223984,
+      "grad_norm": 1.5295252799987793,
+      "learning_rate": 9.123678911441888e-07,
+      "loss": 0.5703,
+      "step": 3600
+    },
+    {
+      "epoch": 0.19949019173223984,
+      "eval_robovqa_loss": 0.0857328474521637,
+      "step": 3600
+    },
+    {
+      "epoch": 0.19949019173223984,
+      "eval_agibot_loss": 0.44913172721862793,
+      "step": 3600
+    },
+    {
+      "epoch": 0.19949019173223984,
+      "eval_holoassist_loss": 0.3329644501209259,
+      "step": 3600
+    },
+    {
+      "epoch": 0.19949019173223984,
+      "eval_robofail_loss": 0.20459067821502686,
+      "step": 3600
+    },
+    {
+      "epoch": 0.20004433115371828,
+      "grad_norm": 1.3413406610488892,
+      "learning_rate": 9.118700167515648e-07,
+      "loss": 0.5728,
+      "step": 3610
+    },
+    {
+      "epoch": 0.20059847057519672,
+      "grad_norm": 1.2162930965423584,
+      "learning_rate": 9.113708686981783e-07,
+      "loss": 0.5817,
+      "step": 3620
+    },
+    {
+      "epoch": 0.20115260999667517,
+      "grad_norm": 1.3999488353729248,
+      "learning_rate": 9.108704485275872e-07,
+      "loss": 0.5627,
+      "step": 3630
+    },
+    {
+      "epoch": 0.2017067494181536,
+      "grad_norm": 1.311826229095459,
+      "learning_rate": 9.103687577872835e-07,
+      "loss": 0.5708,
+      "step": 3640
+    },
+    {
+      "epoch": 0.20226088883963206,
+      "grad_norm": 1.2561227083206177,
+      "learning_rate": 9.098657980286884e-07,
+      "loss": 0.5818,
+      "step": 3650
+    },
+    {
+      "epoch": 0.2028150282611105,
+      "grad_norm": 1.3165614604949951,
+      "learning_rate": 9.09361570807147e-07,
+      "loss": 0.5726,
+      "step": 3660
+    },
+    {
+      "epoch": 0.20336916768258895,
+      "grad_norm": 1.3516511917114258,
+      "learning_rate": 9.088560776819241e-07,
+      "loss": 0.5724,
+      "step": 3670
+    },
+    {
+      "epoch": 0.2039233071040674,
+      "grad_norm": 1.2606111764907837,
+      "learning_rate": 9.083493202161994e-07,
+      "loss": 0.5699,
+      "step": 3680
+    },
+    {
+      "epoch": 0.20447744652554584,
+      "grad_norm": 1.2438963651657104,
+      "learning_rate": 9.078412999770621e-07,
+      "loss": 0.5733,
+      "step": 3690
+    },
+    {
+      "epoch": 0.20503158594702428,
+      "grad_norm": 1.3237754106521606,
+      "learning_rate": 9.073320185355066e-07,
+      "loss": 0.5729,
+      "step": 3700
+    },
+    {
+      "epoch": 0.20503158594702428,
+      "eval_robovqa_loss": 0.0849624052643776,
+      "step": 3700
+    },
+    {
+      "epoch": 0.20503158594702428,
+      "eval_agibot_loss": 0.4511083960533142,
+      "step": 3700
+    },
+    {
+      "epoch": 0.20503158594702428,
+      "eval_holoassist_loss": 0.3347799777984619,
+      "step": 3700
+    },
+    {
+      "epoch": 0.20503158594702428,
+      "eval_robofail_loss": 0.2062092274427414,
+      "step": 3700
+    },
+    {
+      "epoch": 0.20558572536850273,
+      "grad_norm": 1.308685541152954,
+      "learning_rate": 9.068214774664271e-07,
+      "loss": 0.5755,
+      "step": 3710
+    },
+    {
+      "epoch": 0.20613986478998117,
+      "grad_norm": 1.2457493543624878,
+      "learning_rate": 9.063096783486134e-07,
+      "loss": 0.567,
+      "step": 3720
+    },
+    {
+      "epoch": 0.20669400421145961,
+      "grad_norm": 1.377298355102539,
+      "learning_rate": 9.057966227647456e-07,
+      "loss": 0.5654,
+      "step": 3730
+    },
+    {
+      "epoch": 0.20724814363293806,
+      "grad_norm": 1.2591472864151,
+      "learning_rate": 9.05282312301389e-07,
+      "loss": 0.5701,
+      "step": 3740
+    },
+    {
+      "epoch": 0.2078022830544165,
+      "grad_norm": 1.3945839405059814,
+      "learning_rate": 9.047667485489898e-07,
+      "loss": 0.5726,
+      "step": 3750
+    },
+    {
+      "epoch": 0.20835642247589495,
+      "grad_norm": 1.2624473571777344,
+      "learning_rate": 9.0424993310187e-07,
+      "loss": 0.5644,
+      "step": 3760
+    },
+    {
+      "epoch": 0.2089105618973734,
+      "grad_norm": 1.2835627794265747,
+      "learning_rate": 9.037318675582217e-07,
+      "loss": 0.5689,
+      "step": 3770
+    },
+    {
+      "epoch": 0.20946470131885184,
+      "grad_norm": 1.2446424961090088,
+      "learning_rate": 9.032125535201033e-07,
+      "loss": 0.5716,
+      "step": 3780
+    },
+    {
+      "epoch": 0.21001884074033025,
+      "grad_norm": 1.2060092687606812,
+      "learning_rate": 9.026919925934337e-07,
+      "loss": 0.5686,
+      "step": 3790
+    },
+    {
+      "epoch": 0.2105729801618087,
+      "grad_norm": 1.381155252456665,
+      "learning_rate": 9.021701863879883e-07,
+      "loss": 0.5708,
+      "step": 3800
+    },
+    {
+      "epoch": 0.2105729801618087,
+      "eval_robovqa_loss": 0.08439735323190689,
+      "step": 3800
+    },
+    {
+      "epoch": 0.2105729801618087,
+      "eval_agibot_loss": 0.45931491255760193,
+      "step": 3800
+    },
+    {
+      "epoch": 0.2105729801618087,
+      "eval_holoassist_loss": 0.3276296854019165,
+      "step": 3800
+    },
+    {
+      "epoch": 0.2105729801618087,
+      "eval_robofail_loss": 0.20795230567455292,
+      "step": 3800
+    },
+    {
+      "epoch": 0.21112711958328714,
+      "grad_norm": 1.2369918823242188,
+      "learning_rate": 9.016471365173924e-07,
+      "loss": 0.5643,
+      "step": 3810
+    },
+    {
+      "epoch": 0.2116812590047656,
+      "grad_norm": 1.3141838312149048,
+      "learning_rate": 9.011228445991178e-07,
+      "loss": 0.5671,
+      "step": 3820
+    },
+    {
+      "epoch": 0.21223539842624403,
+      "grad_norm": 1.1560665369033813,
+      "learning_rate": 9.005973122544772e-07,
+      "loss": 0.5635,
+      "step": 3830
+    },
+    {
+      "epoch": 0.21278953784772248,
+      "grad_norm": 1.2514171600341797,
+      "learning_rate": 9.000705411086191e-07,
+      "loss": 0.5697,
+      "step": 3840
+    },
+    {
+      "epoch": 0.21334367726920092,
+      "grad_norm": 1.2600502967834473,
+      "learning_rate": 8.995425327905229e-07,
+      "loss": 0.5617,
+      "step": 3850
+    },
+    {
+      "epoch": 0.21389781669067937,
+      "grad_norm": 1.2127740383148193,
+      "learning_rate": 8.990132889329935e-07,
+      "loss": 0.567,
+      "step": 3860
+    },
+    {
+      "epoch": 0.2144519561121578,
+      "grad_norm": 1.2815090417861938,
+      "learning_rate": 8.984828111726568e-07,
+      "loss": 0.5599,
+      "step": 3870
+    },
+    {
+      "epoch": 0.21500609553363625,
+      "grad_norm": 1.3086752891540527,
+      "learning_rate": 8.979511011499546e-07,
+      "loss": 0.5632,
+      "step": 3880
+    },
+    {
+      "epoch": 0.2155602349551147,
+      "grad_norm": 1.320613980293274,
+      "learning_rate": 8.97418160509139e-07,
+      "loss": 0.5592,
+      "step": 3890
+    },
+    {
+      "epoch": 0.21611437437659314,
+      "grad_norm": 1.369688868522644,
+      "learning_rate": 8.968839908982677e-07,
+      "loss": 0.5677,
+      "step": 3900
+    },
+    {
+      "epoch": 0.21611437437659314,
+      "eval_robovqa_loss": 0.08452339470386505,
+      "step": 3900
+    },
+    {
+      "epoch": 0.21611437437659314,
+      "eval_agibot_loss": 0.4541947841644287,
+      "step": 3900
+    },
+    {
+      "epoch": 0.21611437437659314,
+      "eval_holoassist_loss": 0.32758042216300964,
+      "step": 3900
+    },
+    {
+      "epoch": 0.21611437437659314,
+      "eval_robofail_loss": 0.20769891142845154,
+      "step": 3900
+    },
+    {
+      "epoch": 0.2166685137980716,
+      "grad_norm": 1.3737530708312988,
+      "learning_rate": 8.963485939691992e-07,
+      "loss": 0.5643,
+      "step": 3910
+    },
+    {
+      "epoch": 0.21722265321955003,
+      "grad_norm": 1.3360264301300049,
+      "learning_rate": 8.958119713775867e-07,
+      "loss": 0.5603,
+      "step": 3920
+    },
+    {
+      "epoch": 0.21777679264102848,
+      "grad_norm": 1.2412455081939697,
+      "learning_rate": 8.952741247828743e-07,
+      "loss": 0.5675,
+      "step": 3930
+    },
+    {
+      "epoch": 0.21833093206250692,
+      "grad_norm": 1.3303290605545044,
+      "learning_rate": 8.947350558482907e-07,
+      "loss": 0.5599,
+      "step": 3940
+    },
+    {
+      "epoch": 0.21888507148398537,
+      "grad_norm": 1.3875236511230469,
+      "learning_rate": 8.941947662408447e-07,
+      "loss": 0.5708,
+      "step": 3950
+    },
+    {
+      "epoch": 0.2194392109054638,
+      "grad_norm": 1.4196513891220093,
+      "learning_rate": 8.936532576313201e-07,
+      "loss": 0.5664,
+      "step": 3960
+    },
+    {
+      "epoch": 0.21999335032694226,
+      "grad_norm": 1.3006395101547241,
+      "learning_rate": 8.9311053169427e-07,
+      "loss": 0.5653,
+      "step": 3970
+    },
+    {
+      "epoch": 0.2205474897484207,
+      "grad_norm": 1.3017299175262451,
+      "learning_rate": 8.925665901080119e-07,
+      "loss": 0.5668,
+      "step": 3980
+    },
+    {
+      "epoch": 0.22110162916989914,
+      "grad_norm": 1.2898036241531372,
+      "learning_rate": 8.920214345546231e-07,
+      "loss": 0.5647,
+      "step": 3990
+    },
+    {
+      "epoch": 0.2216557685913776,
+      "grad_norm": 1.2814085483551025,
+      "learning_rate": 8.914750667199343e-07,
+      "loss": 0.5641,
+      "step": 4000
+    },
+    {
+      "epoch": 0.2216557685913776,
+      "eval_robovqa_loss": 0.08275222033262253,
+      "step": 4000
+    },
+    {
+      "epoch": 0.2216557685913776,
+      "eval_agibot_loss": 0.46179935336112976,
+      "step": 4000
+    },
+    {
+      "epoch": 0.2216557685913776,
+      "eval_holoassist_loss": 0.33820176124572754,
+      "step": 4000
+    },
+    {
+      "epoch": 0.2216557685913776,
+      "eval_robofail_loss": 0.20564843714237213,
+      "step": 4000
+    },
+    {
+      "epoch": 0.22220990801285603,
+      "grad_norm": 1.3665562868118286,
+      "learning_rate": 8.909274882935255e-07,
+      "loss": 0.5632,
+      "step": 4010
+    },
+    {
+      "epoch": 0.22276404743433448,
+      "grad_norm": 1.3643625974655151,
+      "learning_rate": 8.9037870096872e-07,
+      "loss": 0.5598,
+      "step": 4020
+    },
+    {
+      "epoch": 0.22331818685581292,
+      "grad_norm": 1.236586093902588,
+      "learning_rate": 8.898287064425798e-07,
+      "loss": 0.5624,
+      "step": 4030
+    },
+    {
+      "epoch": 0.22387232627729137,
+      "grad_norm": 1.182546854019165,
+      "learning_rate": 8.892775064158997e-07,
+      "loss": 0.5603,
+      "step": 4040
+    },
+    {
+      "epoch": 0.2244264656987698,
+      "grad_norm": 1.2415153980255127,
+      "learning_rate": 8.887251025932028e-07,
+      "loss": 0.5695,
+      "step": 4050
+    },
+    {
+      "epoch": 0.22498060512024826,
+      "grad_norm": 1.2436978816986084,
+      "learning_rate": 8.88171496682734e-07,
+      "loss": 0.5592,
+      "step": 4060
+    },
+    {
+      "epoch": 0.2255347445417267,
+      "grad_norm": 1.3371087312698364,
+      "learning_rate": 8.876166903964566e-07,
+      "loss": 0.564,
+      "step": 4070
+    },
+    {
+      "epoch": 0.22608888396320515,
+      "grad_norm": 1.2987620830535889,
+      "learning_rate": 8.870606854500451e-07,
+      "loss": 0.5616,
+      "step": 4080
+    },
+    {
+      "epoch": 0.2266430233846836,
+      "grad_norm": 1.2397853136062622,
+      "learning_rate": 8.865034835628811e-07,
+      "loss": 0.5599,
+      "step": 4090
+    },
+    {
+      "epoch": 0.22719716280616203,
+      "grad_norm": 1.3552160263061523,
+      "learning_rate": 8.859450864580473e-07,
+      "loss": 0.5654,
+      "step": 4100
+    },
+    {
+      "epoch": 0.22719716280616203,
+      "eval_robovqa_loss": 0.0810282975435257,
+      "step": 4100
+    },
+    {
+      "epoch": 0.22719716280616203,
+      "eval_agibot_loss": 0.4562808573246002,
+      "step": 4100
+    },
+    {
+      "epoch": 0.22719716280616203,
+      "eval_holoassist_loss": 0.33415690064430237,
+      "step": 4100
+    },
+    {
+      "epoch": 0.22719716280616203,
+      "eval_robofail_loss": 0.2051897794008255,
+      "step": 4100
+    },
+    {
+      "epoch": 0.22775130222764048,
+      "grad_norm": 1.2012348175048828,
+      "learning_rate": 8.853854958623229e-07,
+      "loss": 0.5631,
+      "step": 4110
+    },
+    {
+      "epoch": 0.22830544164911892,
+      "grad_norm": 1.247012972831726,
+      "learning_rate": 8.848247135061775e-07,
+      "loss": 0.5665,
+      "step": 4120
+    },
+    {
+      "epoch": 0.22885958107059737,
+      "grad_norm": 1.2206732034683228,
+      "learning_rate": 8.842627411237662e-07,
+      "loss": 0.5581,
+      "step": 4130
+    },
+    {
+      "epoch": 0.2294137204920758,
+      "grad_norm": 1.2207838296890259,
+      "learning_rate": 8.836995804529242e-07,
+      "loss": 0.5563,
+      "step": 4140
+    },
+    {
+      "epoch": 0.22996785991355426,
+      "grad_norm": 1.295717477798462,
+      "learning_rate": 8.831352332351609e-07,
+      "loss": 0.5601,
+      "step": 4150
+    },
+    {
+      "epoch": 0.2305219993350327,
+      "grad_norm": 1.161300539970398,
+      "learning_rate": 8.825697012156557e-07,
+      "loss": 0.5631,
+      "step": 4160
+    },
+    {
+      "epoch": 0.23107613875651115,
+      "grad_norm": 1.2734352350234985,
+      "learning_rate": 8.820029861432513e-07,
+      "loss": 0.5618,
+      "step": 4170
+    },
+    {
+      "epoch": 0.2316302781779896,
+      "grad_norm": 1.3007913827896118,
+      "learning_rate": 8.814350897704491e-07,
+      "loss": 0.5596,
+      "step": 4180
+    },
+    {
+      "epoch": 0.23218441759946803,
+      "grad_norm": 1.301173210144043,
+      "learning_rate": 8.808660138534033e-07,
+      "loss": 0.5562,
+      "step": 4190
+    },
+    {
+      "epoch": 0.23273855702094648,
+      "grad_norm": 1.2733081579208374,
+      "learning_rate": 8.802957601519161e-07,
+      "loss": 0.555,
+      "step": 4200
+    },
+    {
+      "epoch": 0.23273855702094648,
+      "eval_robovqa_loss": 0.08484137058258057,
+      "step": 4200
+    },
+    {
+      "epoch": 0.23273855702094648,
+      "eval_agibot_loss": 0.4595088064670563,
+      "step": 4200
+    },
+    {
+      "epoch": 0.23273855702094648,
+      "eval_holoassist_loss": 0.3368059992790222,
+      "step": 4200
+    },
+    {
+      "epoch": 0.23273855702094648,
+      "eval_robofail_loss": 0.2042950987815857,
+      "step": 4200
+    },
+    {
+      "epoch": 0.23329269644242492,
+      "grad_norm": 1.2433838844299316,
+      "learning_rate": 8.797243304294314e-07,
+      "loss": 0.557,
+      "step": 4210
+    },
+    {
+      "epoch": 0.23384683586390337,
+      "grad_norm": 1.4673044681549072,
+      "learning_rate": 8.791517264530302e-07,
+      "loss": 0.569,
+      "step": 4220
+    },
+    {
+      "epoch": 0.2344009752853818,
+      "grad_norm": 1.3196502923965454,
+      "learning_rate": 8.785779499934247e-07,
+      "loss": 0.5684,
+      "step": 4230
+    },
+    {
+      "epoch": 0.23495511470686026,
+      "grad_norm": 1.2794909477233887,
+      "learning_rate": 8.780030028249524e-07,
+      "loss": 0.5563,
+      "step": 4240
+    },
+    {
+      "epoch": 0.2355092541283387,
+      "grad_norm": 1.3366374969482422,
+      "learning_rate": 8.774268867255718e-07,
+      "loss": 0.5545,
+      "step": 4250
+    },
+    {
+      "epoch": 0.23606339354981715,
+      "grad_norm": 1.2178242206573486,
+      "learning_rate": 8.768496034768556e-07,
+      "loss": 0.5584,
+      "step": 4260
+    },
+    {
+      "epoch": 0.2366175329712956,
+      "grad_norm": 1.3166344165802002,
+      "learning_rate": 8.762711548639859e-07,
+      "loss": 0.5568,
+      "step": 4270
+    },
+    {
+      "epoch": 0.23717167239277404,
+      "grad_norm": 1.4227031469345093,
+      "learning_rate": 8.756915426757491e-07,
+      "loss": 0.5666,
+      "step": 4280
+    },
+    {
+      "epoch": 0.23772581181425245,
+      "grad_norm": 1.3009203672409058,
+      "learning_rate": 8.75110768704529e-07,
+      "loss": 0.564,
+      "step": 4290
+    },
+    {
+      "epoch": 0.2382799512357309,
+      "grad_norm": 1.248992681503296,
+      "learning_rate": 8.745288347463025e-07,
+      "loss": 0.5561,
+      "step": 4300
+    },
+    {
+      "epoch": 0.2382799512357309,
+      "eval_robovqa_loss": 0.08580106496810913,
+      "step": 4300
+    },
+    {
+      "epoch": 0.2382799512357309,
+      "eval_agibot_loss": 0.4556211531162262,
+      "step": 4300
+    },
+    {
+      "epoch": 0.2382799512357309,
+      "eval_holoassist_loss": 0.3428633511066437,
+      "step": 4300
+    },
+    {
+      "epoch": 0.2382799512357309,
+      "eval_robofail_loss": 0.1999615728855133,
+      "step": 4300
+    },
+    {
+      "epoch": 0.23883409065720934,
+      "grad_norm": 1.2841042280197144,
+      "learning_rate": 8.739457426006338e-07,
+      "loss": 0.5695,
+      "step": 4310
+    },
+    {
+      "epoch": 0.23938823007868779,
+      "grad_norm": 1.2788619995117188,
+      "learning_rate": 8.733614940706681e-07,
+      "loss": 0.5579,
+      "step": 4320
+    },
+    {
+      "epoch": 0.23994236950016623,
+      "grad_norm": 1.342121958732605,
+      "learning_rate": 8.727760909631273e-07,
+      "loss": 0.5575,
+      "step": 4330
+    },
+    {
+      "epoch": 0.24049650892164467,
+      "grad_norm": 1.338057518005371,
+      "learning_rate": 8.721895350883031e-07,
+      "loss": 0.5618,
+      "step": 4340
+    },
+    {
+      "epoch": 0.24105064834312312,
+      "grad_norm": 1.281355857849121,
+      "learning_rate": 8.716018282600524e-07,
+      "loss": 0.5637,
+      "step": 4350
+    },
+    {
+      "epoch": 0.24160478776460156,
+      "grad_norm": 1.2831248044967651,
+      "learning_rate": 8.710129722957908e-07,
+      "loss": 0.5705,
+      "step": 4360
+    },
+    {
+      "epoch": 0.24215892718608,
+      "grad_norm": 1.293892741203308,
+      "learning_rate": 8.704229690164882e-07,
+      "loss": 0.5612,
+      "step": 4370
+    },
+    {
+      "epoch": 0.24271306660755845,
+      "grad_norm": 1.3060622215270996,
+      "learning_rate": 8.698318202466618e-07,
+      "loss": 0.5613,
+      "step": 4380
+    },
+    {
+      "epoch": 0.2432672060290369,
+      "grad_norm": 1.2398964166641235,
+      "learning_rate": 8.692395278143714e-07,
+      "loss": 0.5602,
+      "step": 4390
+    },
+    {
+      "epoch": 0.24382134545051534,
+      "grad_norm": 1.2282345294952393,
+      "learning_rate": 8.686460935512132e-07,
+      "loss": 0.5538,
+      "step": 4400
+    },
+    {
+      "epoch": 0.24382134545051534,
+      "eval_robovqa_loss": 0.08638947457075119,
+      "step": 4400
+    },
+    {
+      "epoch": 0.24382134545051534,
+      "eval_agibot_loss": 0.4546915888786316,
+      "step": 4400
+    },
+    {
+      "epoch": 0.24382134545051534,
+      "eval_holoassist_loss": 0.3296436071395874,
+      "step": 4400
+    },
+    {
+      "epoch": 0.24382134545051534,
+      "eval_robofail_loss": 0.20267310738563538,
+      "step": 4400
+    },
+    {
+      "epoch": 0.2443754848719938,
+      "grad_norm": 1.3721836805343628,
+      "learning_rate": 8.68051519292315e-07,
+      "loss": 0.5475,
+      "step": 4410
+    },
+    {
+      "epoch": 0.24492962429347223,
+      "grad_norm": 1.27195143699646,
+      "learning_rate": 8.674558068763289e-07,
+      "loss": 0.5531,
+      "step": 4420
+    },
+    {
+      "epoch": 0.24548376371495068,
+      "grad_norm": 1.4054158926010132,
+      "learning_rate": 8.668589581454277e-07,
+      "loss": 0.5572,
+      "step": 4430
+    },
+    {
+      "epoch": 0.24603790313642912,
+      "grad_norm": 1.3619515895843506,
+      "learning_rate": 8.662609749452974e-07,
+      "loss": 0.5554,
+      "step": 4440
+    },
+    {
+      "epoch": 0.24659204255790756,
+      "grad_norm": 1.3759868144989014,
+      "learning_rate": 8.656618591251322e-07,
+      "loss": 0.5624,
+      "step": 4450
+    },
+    {
+      "epoch": 0.247146181979386,
+      "grad_norm": 1.2735620737075806,
+      "learning_rate": 8.650616125376294e-07,
+      "loss": 0.558,
+      "step": 4460
+    },
+    {
+      "epoch": 0.24770032140086445,
+      "grad_norm": 1.3547358512878418,
+      "learning_rate": 8.644602370389824e-07,
+      "loss": 0.5592,
+      "step": 4470
+    },
+    {
+      "epoch": 0.2482544608223429,
+      "grad_norm": 1.3268604278564453,
+      "learning_rate": 8.638577344888761e-07,
+      "loss": 0.5598,
+      "step": 4480
+    },
+    {
+      "epoch": 0.24880860024382134,
+      "grad_norm": 1.3912461996078491,
+      "learning_rate": 8.632541067504804e-07,
+      "loss": 0.5696,
+      "step": 4490
+    },
+    {
+      "epoch": 0.2493627396652998,
+      "grad_norm": 1.349402666091919,
+      "learning_rate": 8.626493556904448e-07,
+      "loss": 0.5574,
+      "step": 4500
+    },
+    {
+      "epoch": 0.2493627396652998,
+      "eval_robovqa_loss": 0.08722057938575745,
+      "step": 4500
+    },
+    {
+      "epoch": 0.2493627396652998,
+      "eval_agibot_loss": 0.4514520764350891,
+      "step": 4500
+    },
+    {
+      "epoch": 0.2493627396652998,
+      "eval_holoassist_loss": 0.3292100131511688,
+      "step": 4500
+    },
+    {
+      "epoch": 0.2493627396652998,
+      "eval_robofail_loss": 0.19979816675186157,
+      "step": 4500
+    },
+    {
+      "epoch": 0.24991687908677823,
+      "grad_norm": 1.2915725708007812,
+      "learning_rate": 8.620434831788928e-07,
+      "loss": 0.5486,
+      "step": 4510
+    },
+    {
+      "epoch": 0.2504710185082567,
+      "grad_norm": 1.3198262453079224,
+      "learning_rate": 8.614364910894152e-07,
+      "loss": 0.5609,
+      "step": 4520
+    },
+    {
+      "epoch": 0.2510251579297351,
+      "grad_norm": 1.6893151998519897,
+      "learning_rate": 8.608283812990656e-07,
+      "loss": 0.5511,
+      "step": 4530
+    },
+    {
+      "epoch": 0.25157929735121354,
+      "grad_norm": 1.2921971082687378,
+      "learning_rate": 8.602191556883539e-07,
+      "loss": 0.5575,
+      "step": 4540
+    },
+    {
+      "epoch": 0.252133436772692,
+      "grad_norm": 1.328736662864685,
+      "learning_rate": 8.596088161412401e-07,
+      "loss": 0.564,
+      "step": 4550
+    },
+    {
+      "epoch": 0.2526875761941704,
+      "grad_norm": 1.2558881044387817,
+      "learning_rate": 8.589973645451294e-07,
+      "loss": 0.5576,
+      "step": 4560
+    },
+    {
+      "epoch": 0.2532417156156489,
+      "grad_norm": 1.3206793069839478,
+      "learning_rate": 8.583848027908656e-07,
+      "loss": 0.5563,
+      "step": 4570
+    },
+    {
+      "epoch": 0.2537958550371273,
+      "grad_norm": 1.2105953693389893,
+      "learning_rate": 8.577711327727257e-07,
+      "loss": 0.5514,
+      "step": 4580
+    },
+    {
+      "epoch": 0.2543499944586058,
+      "grad_norm": 1.2548668384552002,
+      "learning_rate": 8.571563563884137e-07,
+      "loss": 0.5484,
+      "step": 4590
+    },
+    {
+      "epoch": 0.2549041338800842,
+      "grad_norm": 1.3284965753555298,
+      "learning_rate": 8.565404755390553e-07,
+      "loss": 0.5561,
+      "step": 4600
+    },
+    {
+      "epoch": 0.2549041338800842,
+      "eval_robovqa_loss": 0.08990210294723511,
+      "step": 4600
+    },
+    {
+      "epoch": 0.2549041338800842,
+      "eval_agibot_loss": 0.4536254107952118,
+      "step": 4600
+    },
+    {
+      "epoch": 0.2549041338800842,
+      "eval_holoassist_loss": 0.331330269575119,
+      "step": 4600
+    },
+    {
+      "epoch": 0.2549041338800842,
+      "eval_robofail_loss": 0.20059898495674133,
+      "step": 4600
+    },
+    {
+      "epoch": 0.2554582733015627,
+      "grad_norm": 1.2760227918624878,
+      "learning_rate": 8.559234921291909e-07,
+      "loss": 0.5583,
+      "step": 4610
+    },
+    {
+      "epoch": 0.2560124127230411,
+      "grad_norm": 1.2619925737380981,
+      "learning_rate": 8.553054080667712e-07,
+      "loss": 0.5514,
+      "step": 4620
+    },
+    {
+      "epoch": 0.25656655214451957,
+      "grad_norm": 1.3726162910461426,
+      "learning_rate": 8.546862252631501e-07,
+      "loss": 0.55,
+      "step": 4630
+    },
+    {
+      "epoch": 0.257120691565998,
+      "grad_norm": 1.333472728729248,
+      "learning_rate": 8.540659456330797e-07,
+      "loss": 0.5461,
+      "step": 4640
+    },
+    {
+      "epoch": 0.25767483098747646,
+      "grad_norm": 1.343008041381836,
+      "learning_rate": 8.53444571094703e-07,
+      "loss": 0.5548,
+      "step": 4650
+    },
+    {
+      "epoch": 0.25822897040895487,
+      "grad_norm": 1.2020753622055054,
+      "learning_rate": 8.5282210356955e-07,
+      "loss": 0.5601,
+      "step": 4660
+    },
+    {
+      "epoch": 0.25878310983043334,
+      "grad_norm": 1.2992331981658936,
+      "learning_rate": 8.521985449825297e-07,
+      "loss": 0.5503,
+      "step": 4670
+    },
+    {
+      "epoch": 0.25933724925191176,
+      "grad_norm": 1.295770287513733,
+      "learning_rate": 8.515738972619258e-07,
+      "loss": 0.5507,
+      "step": 4680
+    },
+    {
+      "epoch": 0.25989138867339023,
+      "grad_norm": 1.2067456245422363,
+      "learning_rate": 8.509481623393893e-07,
+      "loss": 0.5542,
+      "step": 4690
+    },
+    {
+      "epoch": 0.26044552809486865,
+      "grad_norm": 1.30301034450531,
+      "learning_rate": 8.50321342149934e-07,
+      "loss": 0.5465,
+      "step": 4700
+    },
+    {
+      "epoch": 0.26044552809486865,
+      "eval_robovqa_loss": 0.08503945171833038,
+      "step": 4700
+    },
+    {
+      "epoch": 0.26044552809486865,
+      "eval_agibot_loss": 0.4502505958080292,
+      "step": 4700
+    },
+    {
+      "epoch": 0.26044552809486865,
+      "eval_holoassist_loss": 0.3247644305229187,
+      "step": 4700
+    },
+    {
+      "epoch": 0.26044552809486865,
+      "eval_robofail_loss": 0.202493816614151,
+      "step": 4700
+    },
+    {
+      "epoch": 0.2609996675163471,
+      "grad_norm": 1.2314953804016113,
+      "learning_rate": 8.496934386319293e-07,
+      "loss": 0.5558,
+      "step": 4710
+    },
+    {
+      "epoch": 0.26155380693782554,
+      "grad_norm": 1.2774019241333008,
+      "learning_rate": 8.490644537270945e-07,
+      "loss": 0.5615,
+      "step": 4720
+    },
+    {
+      "epoch": 0.262107946359304,
+      "grad_norm": 1.2643412351608276,
+      "learning_rate": 8.484343893804937e-07,
+      "loss": 0.5581,
+      "step": 4730
+    },
+    {
+      "epoch": 0.26266208578078243,
+      "grad_norm": 1.2253189086914062,
+      "learning_rate": 8.478032475405283e-07,
+      "loss": 0.5524,
+      "step": 4740
+    },
+    {
+      "epoch": 0.2632162252022609,
+      "grad_norm": 1.2311854362487793,
+      "learning_rate": 8.471710301589322e-07,
+      "loss": 0.5556,
+      "step": 4750
+    },
+    {
+      "epoch": 0.2637703646237393,
+      "grad_norm": 1.2708320617675781,
+      "learning_rate": 8.465377391907649e-07,
+      "loss": 0.5468,
+      "step": 4760
+    },
+    {
+      "epoch": 0.2643245040452178,
+      "grad_norm": 1.2230855226516724,
+      "learning_rate": 8.459033765944061e-07,
+      "loss": 0.5565,
+      "step": 4770
+    },
+    {
+      "epoch": 0.2648786434666962,
+      "grad_norm": 1.2978566884994507,
+      "learning_rate": 8.452679443315496e-07,
+      "loss": 0.5516,
+      "step": 4780
+    },
+    {
+      "epoch": 0.2654327828881747,
+      "grad_norm": 1.2754299640655518,
+      "learning_rate": 8.446314443671963e-07,
+      "loss": 0.5576,
+      "step": 4790
+    },
+    {
+      "epoch": 0.2659869223096531,
+      "grad_norm": 1.2901555299758911,
+      "learning_rate": 8.439938786696495e-07,
+      "loss": 0.5438,
+      "step": 4800
+    },
+    {
+      "epoch": 0.2659869223096531,
+      "eval_robovqa_loss": 0.08473614603281021,
+      "step": 4800
+    },
+    {
+      "epoch": 0.2659869223096531,
+      "eval_agibot_loss": 0.46001389622688293,
+      "step": 4800
+    },
+    {
+      "epoch": 0.2659869223096531,
+      "eval_holoassist_loss": 0.3467250168323517,
+      "step": 4800
+    },
+    {
+      "epoch": 0.2659869223096531,
+      "eval_robofail_loss": 0.2073190063238144,
+      "step": 4800
+    },
+    {
+      "epoch": 0.26654106173113157,
+      "grad_norm": 1.4104182720184326,
+      "learning_rate": 8.43355249210508e-07,
+      "loss": 0.555,
+      "step": 4810
+    },
+    {
+      "epoch": 0.26709520115261,
+      "grad_norm": 1.2581645250320435,
+      "learning_rate": 8.427155579646602e-07,
+      "loss": 0.5573,
+      "step": 4820
+    },
+    {
+      "epoch": 0.26764934057408846,
+      "grad_norm": 1.313294529914856,
+      "learning_rate": 8.420748069102777e-07,
+      "loss": 0.5558,
+      "step": 4830
+    },
+    {
+      "epoch": 0.2682034799955669,
+      "grad_norm": 1.2341513633728027,
+      "learning_rate": 8.414329980288098e-07,
+      "loss": 0.5442,
+      "step": 4840
+    },
+    {
+      "epoch": 0.26875761941704535,
+      "grad_norm": 1.381020426750183,
+      "learning_rate": 8.407901333049764e-07,
+      "loss": 0.5513,
+      "step": 4850
+    },
+    {
+      "epoch": 0.26931175883852376,
+      "grad_norm": 1.2394689321517944,
+      "learning_rate": 8.401462147267633e-07,
+      "loss": 0.5504,
+      "step": 4860
+    },
+    {
+      "epoch": 0.26986589826000223,
+      "grad_norm": 1.2511621713638306,
+      "learning_rate": 8.395012442854147e-07,
+      "loss": 0.5528,
+      "step": 4870
+    },
+    {
+      "epoch": 0.27042003768148065,
+      "grad_norm": 1.4880037307739258,
+      "learning_rate": 8.388552239754276e-07,
+      "loss": 0.559,
+      "step": 4880
+    },
+    {
+      "epoch": 0.2709741771029591,
+      "grad_norm": 1.3510971069335938,
+      "learning_rate": 8.382081557945457e-07,
+      "loss": 0.556,
+      "step": 4890
+    },
+    {
+      "epoch": 0.27152831652443754,
+      "grad_norm": 1.2136772871017456,
+      "learning_rate": 8.375600417437533e-07,
+      "loss": 0.5533,
+      "step": 4900
+    },
+    {
+      "epoch": 0.27152831652443754,
+      "eval_robovqa_loss": 0.08745811879634857,
+      "step": 4900
+    },
+    {
+      "epoch": 0.27152831652443754,
+      "eval_agibot_loss": 0.4579383432865143,
+      "step": 4900
+    },
+    {
+      "epoch": 0.27152831652443754,
+      "eval_holoassist_loss": 0.33530914783477783,
+      "step": 4900
+    },
+    {
+      "epoch": 0.27152831652443754,
+      "eval_robofail_loss": 0.20376922190189362,
+      "step": 4900
+    },
+    {
+      "epoch": 0.272082455945916,
+      "grad_norm": 1.2051427364349365,
+      "learning_rate": 8.369108838272685e-07,
+      "loss": 0.559,
+      "step": 4910
+    },
+    {
+      "epoch": 0.27263659536739443,
+      "grad_norm": 1.2788158655166626,
+      "learning_rate": 8.362606840525377e-07,
+      "loss": 0.5505,
+      "step": 4920
+    },
+    {
+      "epoch": 0.2731907347888729,
+      "grad_norm": 1.392308235168457,
+      "learning_rate": 8.35609444430229e-07,
+      "loss": 0.5456,
+      "step": 4930
+    },
+    {
+      "epoch": 0.2737448742103513,
+      "grad_norm": 1.8300435543060303,
+      "learning_rate": 8.349571669742264e-07,
+      "loss": 0.5478,
+      "step": 4940
+    },
+    {
+      "epoch": 0.2742990136318298,
+      "grad_norm": 1.2759344577789307,
+      "learning_rate": 8.343038537016229e-07,
+      "loss": 0.5552,
+      "step": 4950
+    },
+    {
+      "epoch": 0.2748531530533082,
+      "grad_norm": 1.2899155616760254,
+      "learning_rate": 8.336495066327149e-07,
+      "loss": 0.5586,
+      "step": 4960
+    },
+    {
+      "epoch": 0.2754072924747867,
+      "grad_norm": 1.2377238273620605,
+      "learning_rate": 8.329941277909958e-07,
+      "loss": 0.5531,
+      "step": 4970
+    },
+    {
+      "epoch": 0.2759614318962651,
+      "grad_norm": 1.3091094493865967,
+      "learning_rate": 8.323377192031491e-07,
+      "loss": 0.5552,
+      "step": 4980
+    },
+    {
+      "epoch": 0.27651557131774357,
+      "grad_norm": 1.2853509187698364,
+      "learning_rate": 8.316802828990434e-07,
+      "loss": 0.5518,
+      "step": 4990
+    },
+    {
+      "epoch": 0.277069710739222,
+      "grad_norm": 1.3562005758285522,
+      "learning_rate": 8.310218209117247e-07,
+      "loss": 0.545,
+      "step": 5000
+    },
+    {
+      "epoch": 0.277069710739222,
+      "eval_robovqa_loss": 0.08314009010791779,
+      "step": 5000
+    },
+    {
+      "epoch": 0.277069710739222,
+      "eval_agibot_loss": 0.4647045433521271,
+      "step": 5000
+    },
+    {
+      "epoch": 0.277069710739222,
+      "eval_holoassist_loss": 0.3442862629890442,
+      "step": 5000
+    },
+    {
+      "epoch": 0.277069710739222,
+      "eval_robofail_loss": 0.20672057569026947,
+      "step": 5000
+    },
+    {
+      "epoch": 0.27762385016070046,
+      "grad_norm": 1.2919642925262451,
+      "learning_rate": 8.303623352774114e-07,
+      "loss": 0.5522,
+      "step": 5010
+    },
+    {
+      "epoch": 0.2781779895821789,
+      "grad_norm": 1.2641850709915161,
+      "learning_rate": 8.297018280354868e-07,
+      "loss": 0.5497,
+      "step": 5020
+    },
+    {
+      "epoch": 0.27873212900365735,
+      "grad_norm": 1.1237480640411377,
+      "learning_rate": 8.290403012284942e-07,
+      "loss": 0.554,
+      "step": 5030
+    },
+    {
+      "epoch": 0.27928626842513576,
+      "grad_norm": 1.2922521829605103,
+      "learning_rate": 8.283777569021291e-07,
+      "loss": 0.5525,
+      "step": 5040
+    },
+    {
+      "epoch": 0.2798404078466142,
+      "grad_norm": 1.2149744033813477,
+      "learning_rate": 8.277141971052338e-07,
+      "loss": 0.5549,
+      "step": 5050
+    },
+    {
+      "epoch": 0.28039454726809265,
+      "grad_norm": 1.3019400835037231,
+      "learning_rate": 8.270496238897906e-07,
+      "loss": 0.55,
+      "step": 5060
+    },
+    {
+      "epoch": 0.28094868668957107,
+      "grad_norm": 1.1062990427017212,
+      "learning_rate": 8.263840393109165e-07,
+      "loss": 0.5548,
+      "step": 5070
+    },
+    {
+      "epoch": 0.28150282611104954,
+      "grad_norm": 1.3648393154144287,
+      "learning_rate": 8.257174454268548e-07,
+      "loss": 0.5457,
+      "step": 5080
+    },
+    {
+      "epoch": 0.28205696553252796,
+      "grad_norm": 1.2847150564193726,
+      "learning_rate": 8.250498442989711e-07,
+      "loss": 0.5502,
+      "step": 5090
+    },
+    {
+      "epoch": 0.28261110495400643,
+      "grad_norm": 1.1485751867294312,
+      "learning_rate": 8.243812379917447e-07,
+      "loss": 0.5478,
+      "step": 5100
+    },
+    {
+      "epoch": 0.28261110495400643,
+      "eval_robovqa_loss": 0.08122823387384415,
+      "step": 5100
+    },
+    {
+      "epoch": 0.28261110495400643,
+      "eval_agibot_loss": 0.4676099419593811,
+      "step": 5100
+    },
+    {
+      "epoch": 0.28261110495400643,
+      "eval_holoassist_loss": 0.3342665731906891,
+      "step": 5100
+    },
+    {
+      "epoch": 0.28261110495400643,
+      "eval_robofail_loss": 0.20942069590091705,
+      "step": 5100
+    },
+    {
+      "epoch": 0.28316524437548485,
+      "grad_norm": 1.184983253479004,
+      "learning_rate": 8.237116285727645e-07,
+      "loss": 0.5409,
+      "step": 5110
+    },
+    {
+      "epoch": 0.2837193837969633,
+      "grad_norm": 1.2263656854629517,
+      "learning_rate": 8.230410181127204e-07,
+      "loss": 0.555,
+      "step": 5120
+    },
+    {
+      "epoch": 0.28427352321844174,
+      "grad_norm": 1.3507472276687622,
+      "learning_rate": 8.223694086853985e-07,
+      "loss": 0.5518,
+      "step": 5130
+    },
+    {
+      "epoch": 0.2848276626399202,
+      "grad_norm": 1.3214354515075684,
+      "learning_rate": 8.216968023676737e-07,
+      "loss": 0.5498,
+      "step": 5140
+    },
+    {
+      "epoch": 0.2853818020613986,
+      "grad_norm": 1.257547378540039,
+      "learning_rate": 8.21023201239504e-07,
+      "loss": 0.5474,
+      "step": 5150
+    },
+    {
+      "epoch": 0.2859359414828771,
+      "grad_norm": 1.1825259923934937,
+      "learning_rate": 8.203486073839234e-07,
+      "loss": 0.5562,
+      "step": 5160
+    },
+    {
+      "epoch": 0.2864900809043555,
+      "grad_norm": 1.2845077514648438,
+      "learning_rate": 8.196730228870362e-07,
+      "loss": 0.5485,
+      "step": 5170
+    },
+    {
+      "epoch": 0.287044220325834,
+      "grad_norm": 1.318871021270752,
+      "learning_rate": 8.189964498380099e-07,
+      "loss": 0.5409,
+      "step": 5180
+    },
+    {
+      "epoch": 0.2875983597473124,
+      "grad_norm": 1.3438853025436401,
+      "learning_rate": 8.183188903290688e-07,
+      "loss": 0.5476,
+      "step": 5190
+    },
+    {
+      "epoch": 0.2881524991687909,
+      "grad_norm": 1.279742956161499,
+      "learning_rate": 8.176403464554882e-07,
+      "loss": 0.5359,
+      "step": 5200
+    },
+    {
+      "epoch": 0.2881524991687909,
+      "eval_robovqa_loss": 0.08487728983163834,
+      "step": 5200
+    },
+    {
+      "epoch": 0.2881524991687909,
+      "eval_agibot_loss": 0.44153258204460144,
+      "step": 5200
+    },
+    {
+      "epoch": 0.2881524991687909,
+      "eval_holoassist_loss": 0.32904037833213806,
+      "step": 5200
+    },
+    {
+      "epoch": 0.2881524991687909,
+      "eval_robofail_loss": 0.20529979467391968,
+      "step": 5200
+    },
+    {
+      "epoch": 0.2887066385902693,
+      "grad_norm": 1.1370896100997925,
+      "learning_rate": 8.16960820315587e-07,
+      "loss": 0.5527,
+      "step": 5210
+    },
+    {
+      "epoch": 0.28926077801174777,
+      "grad_norm": 1.4968993663787842,
+      "learning_rate": 8.162803140107219e-07,
+      "loss": 0.5454,
+      "step": 5220
+    },
+    {
+      "epoch": 0.2898149174332262,
+      "grad_norm": 1.258643627166748,
+      "learning_rate": 8.155988296452805e-07,
+      "loss": 0.5471,
+      "step": 5230
+    },
+    {
+      "epoch": 0.29036905685470465,
+      "grad_norm": 1.2155413627624512,
+      "learning_rate": 8.14916369326675e-07,
+      "loss": 0.554,
+      "step": 5240
+    },
+    {
+      "epoch": 0.29092319627618307,
+      "grad_norm": 1.4874836206436157,
+      "learning_rate": 8.142329351653355e-07,
+      "loss": 0.5466,
+      "step": 5250
+    },
+    {
+      "epoch": 0.29147733569766154,
+      "grad_norm": 1.2821260690689087,
+      "learning_rate": 8.13548529274704e-07,
+      "loss": 0.5529,
+      "step": 5260
+    },
+    {
+      "epoch": 0.29203147511913996,
+      "grad_norm": 1.258056640625,
+      "learning_rate": 8.128631537712269e-07,
+      "loss": 0.5407,
+      "step": 5270
+    },
+    {
+      "epoch": 0.29258561454061843,
+      "grad_norm": 1.8037123680114746,
+      "learning_rate": 8.121768107743495e-07,
+      "loss": 0.552,
+      "step": 5280
+    },
+    {
+      "epoch": 0.29313975396209685,
+      "grad_norm": 1.2856199741363525,
+      "learning_rate": 8.114895024065086e-07,
+      "loss": 0.5496,
+      "step": 5290
+    },
+    {
+      "epoch": 0.2936938933835753,
+      "grad_norm": 1.4162452220916748,
+      "learning_rate": 8.108012307931265e-07,
+      "loss": 0.5478,
+      "step": 5300
+    },
+    {
+      "epoch": 0.2936938933835753,
+      "eval_robovqa_loss": 0.08023351430892944,
+      "step": 5300
+    },
+    {
+      "epoch": 0.2936938933835753,
+      "eval_agibot_loss": 0.45170852541923523,
+      "step": 5300
+    },
+    {
+      "epoch": 0.2936938933835753,
+      "eval_holoassist_loss": 0.318655401468277,
+      "step": 5300
+    },
+    {
+      "epoch": 0.2936938933835753,
+      "eval_robofail_loss": 0.2053615152835846,
+      "step": 5300
+    },
+    {
+      "epoch": 0.29424803280505374,
+      "grad_norm": 1.1651155948638916,
+      "learning_rate": 8.101119980626043e-07,
+      "loss": 0.5503,
+      "step": 5310
+    },
+    {
+      "epoch": 0.2948021722265322,
+      "grad_norm": 1.2835332155227661,
+      "learning_rate": 8.09421806346315e-07,
+      "loss": 0.5424,
+      "step": 5320
+    },
+    {
+      "epoch": 0.2953563116480106,
+      "grad_norm": 1.300445795059204,
+      "learning_rate": 8.087306577785973e-07,
+      "loss": 0.5487,
+      "step": 5330
+    },
+    {
+      "epoch": 0.2959104510694891,
+      "grad_norm": 2.3293375968933105,
+      "learning_rate": 8.08038554496749e-07,
+      "loss": 0.5425,
+      "step": 5340
+    },
+    {
+      "epoch": 0.2964645904909675,
+      "grad_norm": 1.290459394454956,
+      "learning_rate": 8.073454986410197e-07,
+      "loss": 0.5428,
+      "step": 5350
+    },
+    {
+      "epoch": 0.297018729912446,
+      "grad_norm": 1.4786099195480347,
+      "learning_rate": 8.066514923546054e-07,
+      "loss": 0.549,
+      "step": 5360
+    },
+    {
+      "epoch": 0.2975728693339244,
+      "grad_norm": 1.3721383810043335,
+      "learning_rate": 8.059565377836408e-07,
+      "loss": 0.5464,
+      "step": 5370
+    },
+    {
+      "epoch": 0.2981270087554029,
+      "grad_norm": 1.280025839805603,
+      "learning_rate": 8.052606370771933e-07,
+      "loss": 0.5476,
+      "step": 5380
+    },
+    {
+      "epoch": 0.2986811481768813,
+      "grad_norm": 1.3125091791152954,
+      "learning_rate": 8.045637923872557e-07,
+      "loss": 0.5499,
+      "step": 5390
+    },
+    {
+      "epoch": 0.29923528759835977,
+      "grad_norm": 1.3247753381729126,
+      "learning_rate": 8.038660058687406e-07,
+      "loss": 0.5453,
+      "step": 5400
+    },
+    {
+      "epoch": 0.29923528759835977,
+      "eval_robovqa_loss": 0.08001672476530075,
+      "step": 5400
+    },
+    {
+      "epoch": 0.29923528759835977,
+      "eval_agibot_loss": 0.4644435942173004,
+      "step": 5400
+    },
+    {
+      "epoch": 0.29923528759835977,
+      "eval_holoassist_loss": 0.3294561505317688,
+      "step": 5400
+    },
+    {
+      "epoch": 0.29923528759835977,
+      "eval_robofail_loss": 0.20842325687408447,
+      "step": 5400
+    },
+    {
+      "epoch": 0.2997894270198382,
+      "grad_norm": 1.2851347923278809,
+      "learning_rate": 8.031672796794722e-07,
+      "loss": 0.5485,
+      "step": 5410
+    },
+    {
+      "epoch": 0.30034356644131666,
+      "grad_norm": 1.4059886932373047,
+      "learning_rate": 8.024676159801818e-07,
+      "loss": 0.5397,
+      "step": 5420
+    },
+    {
+      "epoch": 0.3008977058627951,
+      "grad_norm": 1.2744922637939453,
+      "learning_rate": 8.017670169344983e-07,
+      "loss": 0.5587,
+      "step": 5430
+    },
+    {
+      "epoch": 0.30145184528427355,
+      "grad_norm": 1.971516728401184,
+      "learning_rate": 8.010654847089444e-07,
+      "loss": 0.5418,
+      "step": 5440
+    },
+    {
+      "epoch": 0.30200598470575196,
+      "grad_norm": 1.231545329093933,
+      "learning_rate": 8.00363021472928e-07,
+      "loss": 0.5523,
+      "step": 5450
+    },
+    {
+      "epoch": 0.30256012412723043,
+      "grad_norm": 1.3640284538269043,
+      "learning_rate": 7.996596293987358e-07,
+      "loss": 0.5394,
+      "step": 5460
+    },
+    {
+      "epoch": 0.30311426354870885,
+      "grad_norm": 1.1997969150543213,
+      "learning_rate": 7.989553106615271e-07,
+      "loss": 0.5494,
+      "step": 5470
+    },
+    {
+      "epoch": 0.3036684029701873,
+      "grad_norm": 1.3161402940750122,
+      "learning_rate": 7.982500674393267e-07,
+      "loss": 0.5398,
+      "step": 5480
+    },
+    {
+      "epoch": 0.30422254239166574,
+      "grad_norm": 1.3121156692504883,
+      "learning_rate": 7.975439019130188e-07,
+      "loss": 0.54,
+      "step": 5490
+    },
+    {
+      "epoch": 0.3047766818131442,
+      "grad_norm": 1.321846842765808,
+      "learning_rate": 7.968368162663388e-07,
+      "loss": 0.5461,
+      "step": 5500
+    },
+    {
+      "epoch": 0.3047766818131442,
+      "eval_robovqa_loss": 0.07996631413698196,
+      "step": 5500
+    },
+    {
+      "epoch": 0.3047766818131442,
+      "eval_agibot_loss": 0.4673475921154022,
+      "step": 5500
+    },
+    {
+      "epoch": 0.3047766818131442,
+      "eval_holoassist_loss": 0.33448463678359985,
+      "step": 5500
+    },
+    {
+      "epoch": 0.3047766818131442,
+      "eval_robofail_loss": 0.2088545262813568,
+      "step": 5500
+    },
+    {
+      "epoch": 0.30533082123462263,
+      "grad_norm": 1.2569937705993652,
+      "learning_rate": 7.96128812685868e-07,
+      "loss": 0.5457,
+      "step": 5510
+    },
+    {
+      "epoch": 0.3058849606561011,
+      "grad_norm": 1.216032862663269,
+      "learning_rate": 7.954198933610261e-07,
+      "loss": 0.5427,
+      "step": 5520
+    },
+    {
+      "epoch": 0.3064391000775795,
+      "grad_norm": 1.2663429975509644,
+      "learning_rate": 7.947100604840651e-07,
+      "loss": 0.5515,
+      "step": 5530
+    },
+    {
+      "epoch": 0.306993239499058,
+      "grad_norm": 1.2328871488571167,
+      "learning_rate": 7.939993162500616e-07,
+      "loss": 0.5471,
+      "step": 5540
+    },
+    {
+      "epoch": 0.3075473789205364,
+      "grad_norm": 1.4481216669082642,
+      "learning_rate": 7.932876628569105e-07,
+      "loss": 0.542,
+      "step": 5550
+    },
+    {
+      "epoch": 0.3081015183420148,
+      "grad_norm": 1.1974244117736816,
+      "learning_rate": 7.925751025053181e-07,
+      "loss": 0.544,
+      "step": 5560
+    },
+    {
+      "epoch": 0.3086556577634933,
+      "grad_norm": 1.2052700519561768,
+      "learning_rate": 7.918616373987959e-07,
+      "loss": 0.5451,
+      "step": 5570
+    },
+    {
+      "epoch": 0.3092097971849717,
+      "grad_norm": 1.336724042892456,
+      "learning_rate": 7.911472697436526e-07,
+      "loss": 0.5407,
+      "step": 5580
+    },
+    {
+      "epoch": 0.3097639366064502,
+      "grad_norm": 1.1822980642318726,
+      "learning_rate": 7.904320017489883e-07,
+      "loss": 0.5437,
+      "step": 5590
+    },
+    {
+      "epoch": 0.3103180760279286,
+      "grad_norm": 1.3034085035324097,
+      "learning_rate": 7.89715835626687e-07,
+      "loss": 0.5519,
+      "step": 5600
+    },
+    {
+      "epoch": 0.3103180760279286,
+      "eval_robovqa_loss": 0.08562729507684708,
+      "step": 5600
+    },
+    {
+      "epoch": 0.3103180760279286,
+      "eval_agibot_loss": 0.4425092935562134,
+      "step": 5600
+    },
+    {
+      "epoch": 0.3103180760279286,
+      "eval_holoassist_loss": 0.3395546078681946,
+      "step": 5600
+    },
+    {
+      "epoch": 0.3103180760279286,
+      "eval_robofail_loss": 0.20592272281646729,
+      "step": 5600
+    },
+    {
+      "epoch": 0.3108722154494071,
+      "grad_norm": 1.6111397743225098,
+      "learning_rate": 7.889987735914106e-07,
+      "loss": 0.5426,
+      "step": 5610
+    },
+    {
+      "epoch": 0.3114263548708855,
+      "grad_norm": 1.2398587465286255,
+      "learning_rate": 7.882808178605909e-07,
+      "loss": 0.5373,
+      "step": 5620
+    },
+    {
+      "epoch": 0.31198049429236396,
+      "grad_norm": 1.2601368427276611,
+      "learning_rate": 7.875619706544236e-07,
+      "loss": 0.5452,
+      "step": 5630
+    },
+    {
+      "epoch": 0.3125346337138424,
+      "grad_norm": 1.241650938987732,
+      "learning_rate": 7.868422341958611e-07,
+      "loss": 0.5356,
+      "step": 5640
+    },
+    {
+      "epoch": 0.31308877313532085,
+      "grad_norm": 1.348976969718933,
+      "learning_rate": 7.861216107106063e-07,
+      "loss": 0.5468,
+      "step": 5650
+    },
+    {
+      "epoch": 0.31364291255679927,
+      "grad_norm": 1.2530159950256348,
+      "learning_rate": 7.854001024271042e-07,
+      "loss": 0.543,
+      "step": 5660
+    },
+    {
+      "epoch": 0.31419705197827774,
+      "grad_norm": 1.262052059173584,
+      "learning_rate": 7.846777115765364e-07,
+      "loss": 0.5333,
+      "step": 5670
+    },
+    {
+      "epoch": 0.31475119139975616,
+      "grad_norm": 1.2714749574661255,
+      "learning_rate": 7.839544403928138e-07,
+      "loss": 0.5474,
+      "step": 5680
+    },
+    {
+      "epoch": 0.31530533082123463,
+      "grad_norm": 1.2592079639434814,
+      "learning_rate": 7.832302911125695e-07,
+      "loss": 0.5547,
+      "step": 5690
+    },
+    {
+      "epoch": 0.31585947024271305,
+      "grad_norm": 1.250287413597107,
+      "learning_rate": 7.825052659751521e-07,
+      "loss": 0.5394,
+      "step": 5700
+    },
+    {
+      "epoch": 0.31585947024271305,
+      "eval_robovqa_loss": 0.08397992700338364,
+      "step": 5700
+    },
+    {
+      "epoch": 0.31585947024271305,
+      "eval_agibot_loss": 0.44672471284866333,
+      "step": 5700
+    },
+    {
+      "epoch": 0.31585947024271305,
+      "eval_holoassist_loss": 0.33210450410842896,
+      "step": 5700
+    },
+    {
+      "epoch": 0.31585947024271305,
+      "eval_robofail_loss": 0.20491094887256622,
+      "step": 5700
+    },
+    {
+      "epoch": 0.3164136096641915,
+      "grad_norm": 1.2695801258087158,
+      "learning_rate": 7.817793672226185e-07,
+      "loss": 0.546,
+      "step": 5710
+    },
+    {
+      "epoch": 0.31696774908566994,
+      "grad_norm": 1.3043222427368164,
+      "learning_rate": 7.810525970997275e-07,
+      "loss": 0.5499,
+      "step": 5720
+    },
+    {
+      "epoch": 0.3175218885071484,
+      "grad_norm": 1.258750319480896,
+      "learning_rate": 7.803249578539321e-07,
+      "loss": 0.5411,
+      "step": 5730
+    },
+    {
+      "epoch": 0.3180760279286268,
+      "grad_norm": 1.2317895889282227,
+      "learning_rate": 7.795964517353733e-07,
+      "loss": 0.5387,
+      "step": 5740
+    },
+    {
+      "epoch": 0.3186301673501053,
+      "grad_norm": 1.3481522798538208,
+      "learning_rate": 7.788670809968727e-07,
+      "loss": 0.5442,
+      "step": 5750
+    },
+    {
+      "epoch": 0.3191843067715837,
+      "grad_norm": 1.2665369510650635,
+      "learning_rate": 7.781368478939256e-07,
+      "loss": 0.5408,
+      "step": 5760
+    },
+    {
+      "epoch": 0.3197384461930622,
+      "grad_norm": 1.2960354089736938,
+      "learning_rate": 7.774057546846939e-07,
+      "loss": 0.547,
+      "step": 5770
+    },
+    {
+      "epoch": 0.3202925856145406,
+      "grad_norm": 1.2627025842666626,
+      "learning_rate": 7.766738036299996e-07,
+      "loss": 0.5363,
+      "step": 5780
+    },
+    {
+      "epoch": 0.3208467250360191,
+      "grad_norm": 1.3968957662582397,
+      "learning_rate": 7.759409969933172e-07,
+      "loss": 0.5381,
+      "step": 5790
+    },
+    {
+      "epoch": 0.3214008644574975,
+      "grad_norm": 1.272607445716858,
+      "learning_rate": 7.752073370407674e-07,
+      "loss": 0.5394,
+      "step": 5800
+    },
+    {
+      "epoch": 0.3214008644574975,
+      "eval_robovqa_loss": 0.08504395186901093,
+      "step": 5800
+    },
+    {
+      "epoch": 0.3214008644574975,
+      "eval_agibot_loss": 0.448690265417099,
+      "step": 5800
+    },
+    {
+      "epoch": 0.3214008644574975,
+      "eval_holoassist_loss": 0.33041247725486755,
+      "step": 5800
+    },
+    {
+      "epoch": 0.3214008644574975,
+      "eval_robofail_loss": 0.2049025148153305,
+      "step": 5800
+    },
+    {
+      "epoch": 0.32195500387897596,
+      "grad_norm": 1.2590553760528564,
+      "learning_rate": 7.744728260411092e-07,
+      "loss": 0.5438,
+      "step": 5810
+    },
+    {
+      "epoch": 0.3225091433004544,
+      "grad_norm": 1.3387494087219238,
+      "learning_rate": 7.737374662657337e-07,
+      "loss": 0.5293,
+      "step": 5820
+    },
+    {
+      "epoch": 0.32306328272193285,
+      "grad_norm": 1.458374261856079,
+      "learning_rate": 7.730012599886564e-07,
+      "loss": 0.5436,
+      "step": 5830
+    },
+    {
+      "epoch": 0.32361742214341127,
+      "grad_norm": 1.1990774869918823,
+      "learning_rate": 7.72264209486511e-07,
+      "loss": 0.532,
+      "step": 5840
+    },
+    {
+      "epoch": 0.32417156156488974,
+      "grad_norm": 1.3720946311950684,
+      "learning_rate": 7.715263170385413e-07,
+      "loss": 0.5417,
+      "step": 5850
+    },
+    {
+      "epoch": 0.32472570098636816,
+      "grad_norm": 1.1835837364196777,
+      "learning_rate": 7.707875849265954e-07,
+      "loss": 0.5429,
+      "step": 5860
+    },
+    {
+      "epoch": 0.32527984040784663,
+      "grad_norm": 1.4111766815185547,
+      "learning_rate": 7.700480154351173e-07,
+      "loss": 0.5429,
+      "step": 5870
+    },
+    {
+      "epoch": 0.32583397982932505,
+      "grad_norm": 1.303207278251648,
+      "learning_rate": 7.693076108511409e-07,
+      "loss": 0.5361,
+      "step": 5880
+    },
+    {
+      "epoch": 0.3263881192508035,
+      "grad_norm": 1.194493293762207,
+      "learning_rate": 7.685663734642823e-07,
+      "loss": 0.5352,
+      "step": 5890
+    },
+    {
+      "epoch": 0.32694225867228194,
+      "grad_norm": 1.309129238128662,
+      "learning_rate": 7.678243055667333e-07,
+      "loss": 0.5399,
+      "step": 5900
+    },
+    {
+      "epoch": 0.32694225867228194,
+      "eval_robovqa_loss": 0.08769973367452621,
+      "step": 5900
+    },
+    {
+      "epoch": 0.32694225867228194,
+      "eval_agibot_loss": 0.4411982297897339,
+      "step": 5900
+    },
+    {
+      "epoch": 0.32694225867228194,
+      "eval_holoassist_loss": 0.3273889124393463,
+      "step": 5900
+    },
+    {
+      "epoch": 0.32694225867228194,
+      "eval_robofail_loss": 0.20246879756450653,
+      "step": 5900
+    },
+    {
+      "epoch": 0.3274963980937604,
+      "grad_norm": 1.3569732904434204,
+      "learning_rate": 7.670814094532534e-07,
+      "loss": 0.5433,
+      "step": 5910
+    },
+    {
+      "epoch": 0.3280505375152388,
+      "grad_norm": 1.3239794969558716,
+      "learning_rate": 7.663376874211636e-07,
+      "loss": 0.5353,
+      "step": 5920
+    },
+    {
+      "epoch": 0.3286046769367173,
+      "grad_norm": 1.3012158870697021,
+      "learning_rate": 7.655931417703393e-07,
+      "loss": 0.5396,
+      "step": 5930
+    },
+    {
+      "epoch": 0.3291588163581957,
+      "grad_norm": 1.3455995321273804,
+      "learning_rate": 7.648477748032021e-07,
+      "loss": 0.5481,
+      "step": 5940
+    },
+    {
+      "epoch": 0.3297129557796742,
+      "grad_norm": 1.237106204032898,
+      "learning_rate": 7.64101588824714e-07,
+      "loss": 0.5319,
+      "step": 5950
+    },
+    {
+      "epoch": 0.3302670952011526,
+      "grad_norm": 1.7620875835418701,
+      "learning_rate": 7.633545861423695e-07,
+      "loss": 0.5361,
+      "step": 5960
+    },
+    {
+      "epoch": 0.3308212346226311,
+      "grad_norm": 1.270606279373169,
+      "learning_rate": 7.626067690661886e-07,
+      "loss": 0.5445,
+      "step": 5970
+    },
+    {
+      "epoch": 0.3313753740441095,
+      "grad_norm": 1.3469855785369873,
+      "learning_rate": 7.618581399087098e-07,
+      "loss": 0.5413,
+      "step": 5980
+    },
+    {
+      "epoch": 0.33192951346558797,
+      "grad_norm": 1.333724856376648,
+      "learning_rate": 7.611087009849831e-07,
+      "loss": 0.5484,
+      "step": 5990
+    },
+    {
+      "epoch": 0.3324836528870664,
+      "grad_norm": 1.3538025617599487,
+      "learning_rate": 7.603584546125621e-07,
+      "loss": 0.5358,
+      "step": 6000
+    },
+    {
+      "epoch": 0.3324836528870664,
+      "eval_robovqa_loss": 0.08478526026010513,
+      "step": 6000
+    },
+    {
+      "epoch": 0.3324836528870664,
+      "eval_agibot_loss": 0.4471578598022461,
+      "step": 6000
+    },
+    {
+      "epoch": 0.3324836528870664,
+      "eval_holoassist_loss": 0.33440643548965454,
+      "step": 6000
+    },
+    {
+      "epoch": 0.3324836528870664,
+      "eval_robofail_loss": 0.2043970823287964,
+      "step": 6000
+    },
+    {
+      "epoch": 0.33303779230854486,
+      "grad_norm": 1.3290870189666748,
+      "learning_rate": 7.596074031114977e-07,
+      "loss": 0.536,
+      "step": 6010
+    },
+    {
+      "epoch": 0.3335919317300233,
+      "grad_norm": 1.2340004444122314,
+      "learning_rate": 7.588555488043302e-07,
+      "loss": 0.5417,
+      "step": 6020
+    },
+    {
+      "epoch": 0.33414607115150174,
+      "grad_norm": 1.351810336112976,
+      "learning_rate": 7.581028940160832e-07,
+      "loss": 0.5396,
+      "step": 6030
+    },
+    {
+      "epoch": 0.33470021057298016,
+      "grad_norm": 1.2620718479156494,
+      "learning_rate": 7.573494410742551e-07,
+      "loss": 0.5383,
+      "step": 6040
+    },
+    {
+      "epoch": 0.3352543499944586,
+      "grad_norm": 1.351129412651062,
+      "learning_rate": 7.565951923088126e-07,
+      "loss": 0.5349,
+      "step": 6050
+    },
+    {
+      "epoch": 0.33580848941593705,
+      "grad_norm": 1.362078070640564,
+      "learning_rate": 7.558401500521834e-07,
+      "loss": 0.5347,
+      "step": 6060
+    },
+    {
+      "epoch": 0.33636262883741547,
+      "grad_norm": 1.2638942003250122,
+      "learning_rate": 7.550843166392492e-07,
+      "loss": 0.542,
+      "step": 6070
+    },
+    {
+      "epoch": 0.33691676825889394,
+      "grad_norm": 1.2298474311828613,
+      "learning_rate": 7.543276944073381e-07,
+      "loss": 0.5367,
+      "step": 6080
+    },
+    {
+      "epoch": 0.33747090768037236,
+      "grad_norm": 1.3017504215240479,
+      "learning_rate": 7.535702856962174e-07,
+      "loss": 0.542,
+      "step": 6090
+    },
+    {
+      "epoch": 0.33802504710185083,
+      "grad_norm": 1.3705782890319824,
+      "learning_rate": 7.528120928480868e-07,
+      "loss": 0.5357,
+      "step": 6100
+    },
+    {
+      "epoch": 0.33802504710185083,
+      "eval_robovqa_loss": 0.08662524074316025,
+      "step": 6100
+    },
+    {
+      "epoch": 0.33802504710185083,
+      "eval_agibot_loss": 0.44642162322998047,
+      "step": 6100
+    },
+    {
+      "epoch": 0.33802504710185083,
+      "eval_holoassist_loss": 0.33696967363357544,
+      "step": 6100
+    },
+    {
+      "epoch": 0.33802504710185083,
+      "eval_robofail_loss": 0.20316119492053986,
+      "step": 6100
+    },
+    {
+      "epoch": 0.33857918652332925,
+      "grad_norm": 1.3317469358444214,
+      "learning_rate": 7.520531182075704e-07,
+      "loss": 0.5394,
+      "step": 6110
+    },
+    {
+      "epoch": 0.3391333259448077,
+      "grad_norm": 1.384121298789978,
+      "learning_rate": 7.512933641217105e-07,
+      "loss": 0.5354,
+      "step": 6120
+    },
+    {
+      "epoch": 0.33968746536628613,
+      "grad_norm": 1.2313035726547241,
+      "learning_rate": 7.505328329399591e-07,
+      "loss": 0.5355,
+      "step": 6130
+    },
+    {
+      "epoch": 0.3402416047877646,
+      "grad_norm": 1.2748585939407349,
+      "learning_rate": 7.497715270141717e-07,
+      "loss": 0.5403,
+      "step": 6140
+    },
+    {
+      "epoch": 0.340795744209243,
+      "grad_norm": 1.223103642463684,
+      "learning_rate": 7.490094486985997e-07,
+      "loss": 0.5427,
+      "step": 6150
+    },
+    {
+      "epoch": 0.3413498836307215,
+      "grad_norm": 1.248349666595459,
+      "learning_rate": 7.482466003498827e-07,
+      "loss": 0.5431,
+      "step": 6160
+    },
+    {
+      "epoch": 0.3419040230521999,
+      "grad_norm": 1.2345030307769775,
+      "learning_rate": 7.474829843270416e-07,
+      "loss": 0.5379,
+      "step": 6170
+    },
+    {
+      "epoch": 0.3424581624736784,
+      "grad_norm": 1.2976412773132324,
+      "learning_rate": 7.467186029914715e-07,
+      "loss": 0.536,
+      "step": 6180
+    },
+    {
+      "epoch": 0.3430123018951568,
+      "grad_norm": 1.1983747482299805,
+      "learning_rate": 7.459534587069338e-07,
+      "loss": 0.5406,
+      "step": 6190
+    },
+    {
+      "epoch": 0.3435664413166353,
+      "grad_norm": 1.4065496921539307,
+      "learning_rate": 7.451875538395496e-07,
+      "loss": 0.5409,
+      "step": 6200
+    },
+    {
+      "epoch": 0.3435664413166353,
+      "eval_robovqa_loss": 0.08880460262298584,
+      "step": 6200
+    },
+    {
+      "epoch": 0.3435664413166353,
+      "eval_agibot_loss": 0.45127546787261963,
+      "step": 6200
+    },
+    {
+      "epoch": 0.3435664413166353,
+      "eval_holoassist_loss": 0.31989723443984985,
+      "step": 6200
+    },
+    {
+      "epoch": 0.3435664413166353,
+      "eval_robofail_loss": 0.20277482271194458,
+      "step": 6200
+    },
+    {
+      "epoch": 0.3441205807381137,
+      "grad_norm": 1.2902637720108032,
+      "learning_rate": 7.444208907577916e-07,
+      "loss": 0.5442,
+      "step": 6210
+    },
+    {
+      "epoch": 0.34467472015959216,
+      "grad_norm": 1.1604535579681396,
+      "learning_rate": 7.436534718324775e-07,
+      "loss": 0.5359,
+      "step": 6220
+    },
+    {
+      "epoch": 0.3452288595810706,
+      "grad_norm": 1.26125168800354,
+      "learning_rate": 7.428852994367621e-07,
+      "loss": 0.5391,
+      "step": 6230
+    },
+    {
+      "epoch": 0.34578299900254905,
+      "grad_norm": 1.2504552602767944,
+      "learning_rate": 7.421163759461307e-07,
+      "loss": 0.5456,
+      "step": 6240
+    },
+    {
+      "epoch": 0.34633713842402747,
+      "grad_norm": 1.2832751274108887,
+      "learning_rate": 7.413467037383907e-07,
+      "loss": 0.5361,
+      "step": 6250
+    },
+    {
+      "epoch": 0.34689127784550594,
+      "grad_norm": 1.2961968183517456,
+      "learning_rate": 7.405762851936652e-07,
+      "loss": 0.5387,
+      "step": 6260
+    },
+    {
+      "epoch": 0.34744541726698436,
+      "grad_norm": 1.2128782272338867,
+      "learning_rate": 7.39805122694385e-07,
+      "loss": 0.5446,
+      "step": 6270
+    },
+    {
+      "epoch": 0.34799955668846283,
+      "grad_norm": 1.2728019952774048,
+      "learning_rate": 7.390332186252819e-07,
+      "loss": 0.5279,
+      "step": 6280
+    },
+    {
+      "epoch": 0.34855369610994125,
+      "grad_norm": 1.161348819732666,
+      "learning_rate": 7.382605753733804e-07,
+      "loss": 0.5348,
+      "step": 6290
+    },
+    {
+      "epoch": 0.3491078355314197,
+      "grad_norm": 1.294723629951477,
+      "learning_rate": 7.374871953279912e-07,
+      "loss": 0.5344,
+      "step": 6300
+    },
+    {
+      "epoch": 0.3491078355314197,
+      "eval_robovqa_loss": 0.08474751561880112,
+      "step": 6300
+    },
+    {
+      "epoch": 0.3491078355314197,
+      "eval_agibot_loss": 0.4608617424964905,
+      "step": 6300
+    },
+    {
+      "epoch": 0.3491078355314197,
+      "eval_holoassist_loss": 0.3383069634437561,
+      "step": 6300
+    },
+    {
+      "epoch": 0.3491078355314197,
+      "eval_robofail_loss": 0.20657028257846832,
+      "step": 6300
+    },
+    {
+      "epoch": 0.34966197495289814,
+      "grad_norm": 1.2939543724060059,
+      "learning_rate": 7.367130808807032e-07,
+      "loss": 0.5391,
+      "step": 6310
+    },
+    {
+      "epoch": 0.3502161143743766,
+      "grad_norm": 1.3433183431625366,
+      "learning_rate": 7.359382344253769e-07,
+      "loss": 0.54,
+      "step": 6320
+    },
+    {
+      "epoch": 0.350770253795855,
+      "grad_norm": 1.1951611042022705,
+      "learning_rate": 7.351626583581355e-07,
+      "loss": 0.537,
+      "step": 6330
+    },
+    {
+      "epoch": 0.3513243932173335,
+      "grad_norm": 1.779075264930725,
+      "learning_rate": 7.343863550773594e-07,
+      "loss": 0.529,
+      "step": 6340
+    },
+    {
+      "epoch": 0.3518785326388119,
+      "grad_norm": 1.3062162399291992,
+      "learning_rate": 7.33609326983677e-07,
+      "loss": 0.5316,
+      "step": 6350
+    },
+    {
+      "epoch": 0.3524326720602904,
+      "grad_norm": 1.226210355758667,
+      "learning_rate": 7.328315764799591e-07,
+      "loss": 0.5305,
+      "step": 6360
+    },
+    {
+      "epoch": 0.3529868114817688,
+      "grad_norm": 1.2681347131729126,
+      "learning_rate": 7.320531059713094e-07,
+      "loss": 0.538,
+      "step": 6370
+    },
+    {
+      "epoch": 0.3535409509032473,
+      "grad_norm": 1.4780999422073364,
+      "learning_rate": 7.312739178650588e-07,
+      "loss": 0.5388,
+      "step": 6380
+    },
+    {
+      "epoch": 0.3540950903247257,
+      "grad_norm": 1.2835758924484253,
+      "learning_rate": 7.30494014570757e-07,
+      "loss": 0.5361,
+      "step": 6390
+    },
+    {
+      "epoch": 0.35464922974620416,
+      "grad_norm": 1.2252618074417114,
+      "learning_rate": 7.297133985001655e-07,
+      "loss": 0.5371,
+      "step": 6400
+    },
+    {
+      "epoch": 0.35464922974620416,
+      "eval_robovqa_loss": 0.08416339755058289,
+      "step": 6400
+    },
+    {
+      "epoch": 0.35464922974620416,
+      "eval_agibot_loss": 0.45472925901412964,
+      "step": 6400
+    },
+    {
+      "epoch": 0.35464922974620416,
+      "eval_holoassist_loss": 0.32759320735931396,
+      "step": 6400
+    },
+    {
+      "epoch": 0.35464922974620416,
+      "eval_robofail_loss": 0.20483629405498505,
+      "step": 6400
+    },
+    {
+      "epoch": 0.3552033691676826,
+      "grad_norm": 1.2739777565002441,
+      "learning_rate": 7.289320720672501e-07,
+      "loss": 0.5343,
+      "step": 6410
+    },
+    {
+      "epoch": 0.35575750858916105,
+      "grad_norm": 1.253502368927002,
+      "learning_rate": 7.281500376881729e-07,
+      "loss": 0.5369,
+      "step": 6420
+    },
+    {
+      "epoch": 0.35631164801063947,
+      "grad_norm": 2.1614837646484375,
+      "learning_rate": 7.273672977812855e-07,
+      "loss": 0.5384,
+      "step": 6430
+    },
+    {
+      "epoch": 0.35686578743211794,
+      "grad_norm": 1.2755719423294067,
+      "learning_rate": 7.265838547671211e-07,
+      "loss": 0.5268,
+      "step": 6440
+    },
+    {
+      "epoch": 0.35741992685359636,
+      "grad_norm": 1.1741158962249756,
+      "learning_rate": 7.257997110683878e-07,
+      "loss": 0.5385,
+      "step": 6450
+    },
+    {
+      "epoch": 0.35797406627507483,
+      "grad_norm": 1.1879491806030273,
+      "learning_rate": 7.250148691099595e-07,
+      "loss": 0.5345,
+      "step": 6460
+    },
+    {
+      "epoch": 0.35852820569655325,
+      "grad_norm": 1.2015740871429443,
+      "learning_rate": 7.242293313188701e-07,
+      "loss": 0.5365,
+      "step": 6470
+    },
+    {
+      "epoch": 0.3590823451180317,
+      "grad_norm": 1.2694122791290283,
+      "learning_rate": 7.23443100124305e-07,
+      "loss": 0.5254,
+      "step": 6480
+    },
+    {
+      "epoch": 0.35963648453951014,
+      "grad_norm": 1.2183427810668945,
+      "learning_rate": 7.226561779575942e-07,
+      "loss": 0.5347,
+      "step": 6490
+    },
+    {
+      "epoch": 0.3601906239609886,
+      "grad_norm": 1.1709555387496948,
+      "learning_rate": 7.218685672522039e-07,
+      "loss": 0.5327,
+      "step": 6500
+    },
+    {
+      "epoch": 0.3601906239609886,
+      "eval_robovqa_loss": 0.08565516024827957,
+      "step": 6500
+    },
+    {
+      "epoch": 0.3601906239609886,
+      "eval_agibot_loss": 0.4596942067146301,
+      "step": 6500
+    },
+    {
+      "epoch": 0.3601906239609886,
+      "eval_holoassist_loss": 0.32646551728248596,
+      "step": 6500
+    },
+    {
+      "epoch": 0.3601906239609886,
+      "eval_robofail_loss": 0.20768232643604279,
+      "step": 6500
+    },
+    {
+      "epoch": 0.360744763382467,
+      "grad_norm": 1.2858402729034424,
+      "learning_rate": 7.2108027044373e-07,
+      "loss": 0.5381,
+      "step": 6510
+    },
+    {
+      "epoch": 0.3612989028039455,
+      "grad_norm": 1.533925175666809,
+      "learning_rate": 7.202912899698897e-07,
+      "loss": 0.5382,
+      "step": 6520
+    },
+    {
+      "epoch": 0.3618530422254239,
+      "grad_norm": 1.2853797674179077,
+      "learning_rate": 7.195016282705148e-07,
+      "loss": 0.539,
+      "step": 6530
+    },
+    {
+      "epoch": 0.3624071816469024,
+      "grad_norm": 1.3063576221466064,
+      "learning_rate": 7.187112877875435e-07,
+      "loss": 0.5308,
+      "step": 6540
+    },
+    {
+      "epoch": 0.3629613210683808,
+      "grad_norm": 1.1954528093338013,
+      "learning_rate": 7.179202709650129e-07,
+      "loss": 0.535,
+      "step": 6550
+    },
+    {
+      "epoch": 0.3635154604898592,
+      "grad_norm": 1.2057158946990967,
+      "learning_rate": 7.171285802490516e-07,
+      "loss": 0.5274,
+      "step": 6560
+    },
+    {
+      "epoch": 0.3640695999113377,
+      "grad_norm": 1.1913872957229614,
+      "learning_rate": 7.163362180878727e-07,
+      "loss": 0.5328,
+      "step": 6570
+    },
+    {
+      "epoch": 0.3646237393328161,
+      "grad_norm": 1.4157410860061646,
+      "learning_rate": 7.155431869317651e-07,
+      "loss": 0.5409,
+      "step": 6580
+    },
+    {
+      "epoch": 0.3651778787542946,
+      "grad_norm": 1.3576669692993164,
+      "learning_rate": 7.147494892330865e-07,
+      "loss": 0.5314,
+      "step": 6590
+    },
+    {
+      "epoch": 0.365732018175773,
+      "grad_norm": 1.2326713800430298,
+      "learning_rate": 7.13955127446256e-07,
+      "loss": 0.5414,
+      "step": 6600
+    },
+    {
+      "epoch": 0.365732018175773,
+      "eval_robovqa_loss": 0.08465208858251572,
+      "step": 6600
+    },
+    {
+      "epoch": 0.365732018175773,
+      "eval_agibot_loss": 0.4610261917114258,
+      "step": 6600
+    },
+    {
+      "epoch": 0.365732018175773,
+      "eval_holoassist_loss": 0.3270190954208374,
+      "step": 6600
+    },
+    {
+      "epoch": 0.365732018175773,
+      "eval_robofail_loss": 0.20155483484268188,
+      "step": 6600
+    },
+    {
+      "epoch": 0.36628615759725147,
+      "grad_norm": 1.2170134782791138,
+      "learning_rate": 7.131601040277464e-07,
+      "loss": 0.5428,
+      "step": 6610
+    },
+    {
+      "epoch": 0.3668402970187299,
+      "grad_norm": 1.2462173700332642,
+      "learning_rate": 7.123644214360763e-07,
+      "loss": 0.5364,
+      "step": 6620
+    },
+    {
+      "epoch": 0.36739443644020836,
+      "grad_norm": 1.2774592638015747,
+      "learning_rate": 7.115680821318029e-07,
+      "loss": 0.5406,
+      "step": 6630
+    },
+    {
+      "epoch": 0.3679485758616868,
+      "grad_norm": 1.2668572664260864,
+      "learning_rate": 7.10771088577514e-07,
+      "loss": 0.5344,
+      "step": 6640
+    },
+    {
+      "epoch": 0.36850271528316525,
+      "grad_norm": 1.310331106185913,
+      "learning_rate": 7.099734432378206e-07,
+      "loss": 0.5347,
+      "step": 6650
+    },
+    {
+      "epoch": 0.36905685470464367,
+      "grad_norm": 1.342057466506958,
+      "learning_rate": 7.091751485793497e-07,
+      "loss": 0.5279,
+      "step": 6660
+    },
+    {
+      "epoch": 0.36961099412612214,
+      "grad_norm": 1.1394859552383423,
+      "learning_rate": 7.083762070707357e-07,
+      "loss": 0.5324,
+      "step": 6670
+    },
+    {
+      "epoch": 0.37016513354760056,
+      "grad_norm": 1.1899021863937378,
+      "learning_rate": 7.075766211826135e-07,
+      "loss": 0.5299,
+      "step": 6680
+    },
+    {
+      "epoch": 0.37071927296907903,
+      "grad_norm": 1.2901475429534912,
+      "learning_rate": 7.067763933876106e-07,
+      "loss": 0.5327,
+      "step": 6690
+    },
+    {
+      "epoch": 0.37127341239055744,
+      "grad_norm": 1.2687288522720337,
+      "learning_rate": 7.059755261603401e-07,
+      "loss": 0.5294,
+      "step": 6700
+    },
+    {
+      "epoch": 0.37127341239055744,
+      "eval_robovqa_loss": 0.08572250604629517,
+      "step": 6700
+    },
+    {
+      "epoch": 0.37127341239055744,
+      "eval_agibot_loss": 0.4469570219516754,
+      "step": 6700
+    },
+    {
+      "epoch": 0.37127341239055744,
+      "eval_holoassist_loss": 0.3284895718097687,
+      "step": 6700
+    },
+    {
+      "epoch": 0.37127341239055744,
+      "eval_robofail_loss": 0.20505672693252563,
+      "step": 6700
+    },
+    {
+      "epoch": 0.3718275518120359,
+      "grad_norm": 1.2539387941360474,
+      "learning_rate": 7.051740219773917e-07,
+      "loss": 0.5323,
+      "step": 6710
+    },
+    {
+      "epoch": 0.37238169123351433,
+      "grad_norm": 1.3213162422180176,
+      "learning_rate": 7.043718833173249e-07,
+      "loss": 0.5357,
+      "step": 6720
+    },
+    {
+      "epoch": 0.3729358306549928,
+      "grad_norm": 1.2832000255584717,
+      "learning_rate": 7.035691126606618e-07,
+      "loss": 0.5351,
+      "step": 6730
+    },
+    {
+      "epoch": 0.3734899700764712,
+      "grad_norm": 1.2155544757843018,
+      "learning_rate": 7.027657124898782e-07,
+      "loss": 0.5307,
+      "step": 6740
+    },
+    {
+      "epoch": 0.3740441094979497,
+      "grad_norm": 1.304326057434082,
+      "learning_rate": 7.019616852893973e-07,
+      "loss": 0.5267,
+      "step": 6750
+    },
+    {
+      "epoch": 0.3745982489194281,
+      "grad_norm": 1.3902068138122559,
+      "learning_rate": 7.011570335455809e-07,
+      "loss": 0.5371,
+      "step": 6760
+    },
+    {
+      "epoch": 0.3751523883409066,
+      "grad_norm": 1.2361536026000977,
+      "learning_rate": 7.003517597467218e-07,
+      "loss": 0.5394,
+      "step": 6770
+    },
+    {
+      "epoch": 0.375706527762385,
+      "grad_norm": 1.1949942111968994,
+      "learning_rate": 6.995458663830373e-07,
+      "loss": 0.5354,
+      "step": 6780
+    },
+    {
+      "epoch": 0.3762606671838635,
+      "grad_norm": 1.2823166847229004,
+      "learning_rate": 6.987393559466598e-07,
+      "loss": 0.5319,
+      "step": 6790
+    },
+    {
+      "epoch": 0.3768148066053419,
+      "grad_norm": 1.2059677839279175,
+      "learning_rate": 6.979322309316303e-07,
+      "loss": 0.5357,
+      "step": 6800
+    },
+    {
+      "epoch": 0.3768148066053419,
+      "eval_robovqa_loss": 0.08371194452047348,
+      "step": 6800
+    },
+    {
+      "epoch": 0.3768148066053419,
+      "eval_agibot_loss": 0.4611804485321045,
+      "step": 6800
+    },
+    {
+      "epoch": 0.3768148066053419,
+      "eval_holoassist_loss": 0.32920703291893005,
+      "step": 6800
+    },
+    {
+      "epoch": 0.3768148066053419,
+      "eval_robofail_loss": 0.20658597350120544,
+      "step": 6800
+    },
+    {
+      "epoch": 0.37736894602682036,
+      "grad_norm": 1.2747524976730347,
+      "learning_rate": 6.971244938338905e-07,
+      "loss": 0.5268,
+      "step": 6810
+    },
+    {
+      "epoch": 0.3779230854482988,
+      "grad_norm": 1.4409936666488647,
+      "learning_rate": 6.963161471512746e-07,
+      "loss": 0.524,
+      "step": 6820
+    },
+    {
+      "epoch": 0.37847722486977725,
+      "grad_norm": 1.2878669500350952,
+      "learning_rate": 6.955071933835018e-07,
+      "loss": 0.5371,
+      "step": 6830
+    },
+    {
+      "epoch": 0.37903136429125567,
+      "grad_norm": 1.1786744594573975,
+      "learning_rate": 6.946976350321691e-07,
+      "loss": 0.5363,
+      "step": 6840
+    },
+    {
+      "epoch": 0.37958550371273414,
+      "grad_norm": 1.3510267734527588,
+      "learning_rate": 6.938874746007425e-07,
+      "loss": 0.5345,
+      "step": 6850
+    },
+    {
+      "epoch": 0.38013964313421256,
+      "grad_norm": 1.3208715915679932,
+      "learning_rate": 6.930767145945503e-07,
+      "loss": 0.5236,
+      "step": 6860
+    },
+    {
+      "epoch": 0.38069378255569103,
+      "grad_norm": 1.210531234741211,
+      "learning_rate": 6.922653575207747e-07,
+      "loss": 0.5336,
+      "step": 6870
+    },
+    {
+      "epoch": 0.38124792197716945,
+      "grad_norm": 1.314744472503662,
+      "learning_rate": 6.914534058884445e-07,
+      "loss": 0.5313,
+      "step": 6880
+    },
+    {
+      "epoch": 0.3818020613986479,
+      "grad_norm": 1.2894172668457031,
+      "learning_rate": 6.906408622084267e-07,
+      "loss": 0.5299,
+      "step": 6890
+    },
+    {
+      "epoch": 0.38235620082012634,
+      "grad_norm": 1.2131211757659912,
+      "learning_rate": 6.898277289934197e-07,
+      "loss": 0.5309,
+      "step": 6900
+    },
+    {
+      "epoch": 0.38235620082012634,
+      "eval_robovqa_loss": 0.08408553153276443,
+      "step": 6900
+    },
+    {
+      "epoch": 0.38235620082012634,
+      "eval_agibot_loss": 0.46406516432762146,
+      "step": 6900
+    },
+    {
+      "epoch": 0.38235620082012634,
+      "eval_holoassist_loss": 0.33228954672813416,
+      "step": 6900
+    },
+    {
+      "epoch": 0.38235620082012634,
+      "eval_robofail_loss": 0.2086234837770462,
+      "step": 6900
+    },
+    {
+      "epoch": 0.3829103402416048,
+      "grad_norm": 1.186152696609497,
+      "learning_rate": 6.890140087579443e-07,
+      "loss": 0.5369,
+      "step": 6910
+    },
+    {
+      "epoch": 0.3834644796630832,
+      "grad_norm": 2.1462924480438232,
+      "learning_rate": 6.881997040183373e-07,
+      "loss": 0.5347,
+      "step": 6920
+    },
+    {
+      "epoch": 0.3840186190845617,
+      "grad_norm": 1.2642779350280762,
+      "learning_rate": 6.873848172927426e-07,
+      "loss": 0.5278,
+      "step": 6930
+    },
+    {
+      "epoch": 0.3845727585060401,
+      "grad_norm": 1.3551560640335083,
+      "learning_rate": 6.865693511011036e-07,
+      "loss": 0.5352,
+      "step": 6940
+    },
+    {
+      "epoch": 0.3851268979275186,
+      "grad_norm": 1.437650203704834,
+      "learning_rate": 6.857533079651562e-07,
+      "loss": 0.5321,
+      "step": 6950
+    },
+    {
+      "epoch": 0.385681037348997,
+      "grad_norm": 1.3121628761291504,
+      "learning_rate": 6.8493669040842e-07,
+      "loss": 0.5284,
+      "step": 6960
+    },
+    {
+      "epoch": 0.3862351767704755,
+      "grad_norm": 1.2171533107757568,
+      "learning_rate": 6.841195009561912e-07,
+      "loss": 0.5316,
+      "step": 6970
+    },
+    {
+      "epoch": 0.3867893161919539,
+      "grad_norm": 1.1728253364562988,
+      "learning_rate": 6.833017421355342e-07,
+      "loss": 0.5308,
+      "step": 6980
+    },
+    {
+      "epoch": 0.38734345561343236,
+      "grad_norm": 1.2442706823349,
+      "learning_rate": 6.824834164752746e-07,
+      "loss": 0.5291,
+      "step": 6990
+    },
+    {
+      "epoch": 0.3878975950349108,
+      "grad_norm": 1.2391456365585327,
+      "learning_rate": 6.816645265059905e-07,
+      "loss": 0.5255,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3878975950349108,
+      "eval_robovqa_loss": 0.08468028903007507,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3878975950349108,
+      "eval_agibot_loss": 0.45892828702926636,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3878975950349108,
+      "eval_holoassist_loss": 0.3201963007450104,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3878975950349108,
+      "eval_robofail_loss": 0.20696014165878296,
+      "step": 7000
+    },
+    {
+      "epoch": 0.38845173445638925,
+      "grad_norm": 1.193182349205017,
+      "learning_rate": 6.808450747600051e-07,
+      "loss": 0.5278,
+      "step": 7010
+    },
+    {
+      "epoch": 0.38900587387786767,
+      "grad_norm": 1.4252454042434692,
+      "learning_rate": 6.800250637713788e-07,
+      "loss": 0.5349,
+      "step": 7020
+    },
+    {
+      "epoch": 0.38956001329934614,
+      "grad_norm": 1.3387396335601807,
+      "learning_rate": 6.792044960759016e-07,
+      "loss": 0.5397,
+      "step": 7030
+    },
+    {
+      "epoch": 0.39011415272082456,
+      "grad_norm": 1.295341968536377,
+      "learning_rate": 6.78383374211085e-07,
+      "loss": 0.5336,
+      "step": 7040
+    },
+    {
+      "epoch": 0.39066829214230303,
+      "grad_norm": 1.1483705043792725,
+      "learning_rate": 6.77561700716154e-07,
+      "loss": 0.5357,
+      "step": 7050
+    },
+    {
+      "epoch": 0.39122243156378145,
+      "grad_norm": 1.272436261177063,
+      "learning_rate": 6.767394781320398e-07,
+      "loss": 0.5275,
+      "step": 7060
+    },
+    {
+      "epoch": 0.39177657098525986,
+      "grad_norm": 1.2445727586746216,
+      "learning_rate": 6.759167090013711e-07,
+      "loss": 0.5294,
+      "step": 7070
+    },
+    {
+      "epoch": 0.39233071040673834,
+      "grad_norm": 1.282623052597046,
+      "learning_rate": 6.750933958684672e-07,
+      "loss": 0.5322,
+      "step": 7080
+    },
+    {
+      "epoch": 0.39288484982821675,
+      "grad_norm": 1.2189747095108032,
+      "learning_rate": 6.742695412793293e-07,
+      "loss": 0.5241,
+      "step": 7090
+    },
+    {
+      "epoch": 0.3934389892496952,
+      "grad_norm": 1.2021690607070923,
+      "learning_rate": 6.734451477816334e-07,
+      "loss": 0.5273,
+      "step": 7100
+    },
+    {
+      "epoch": 0.3934389892496952,
+      "eval_robovqa_loss": 0.08241426199674606,
+      "step": 7100
+    },
+    {
+      "epoch": 0.3934389892496952,
+      "eval_agibot_loss": 0.4529961049556732,
+      "step": 7100
+    },
+    {
+      "epoch": 0.3934389892496952,
+      "eval_holoassist_loss": 0.3262406885623932,
+      "step": 7100
+    },
+    {
+      "epoch": 0.3934389892496952,
+      "eval_robofail_loss": 0.20635876059532166,
+      "step": 7100
+    },
+    {
+      "epoch": 0.39399312867117364,
+      "grad_norm": 1.307944893836975,
+      "learning_rate": 6.726202179247217e-07,
+      "loss": 0.5341,
+      "step": 7110
+    },
+    {
+      "epoch": 0.3945472680926521,
+      "grad_norm": 1.2627551555633545,
+      "learning_rate": 6.717947542595952e-07,
+      "loss": 0.5242,
+      "step": 7120
+    },
+    {
+      "epoch": 0.39510140751413053,
+      "grad_norm": 1.2206027507781982,
+      "learning_rate": 6.709687593389055e-07,
+      "loss": 0.5346,
+      "step": 7130
+    },
+    {
+      "epoch": 0.395655546935609,
+      "grad_norm": 1.2405879497528076,
+      "learning_rate": 6.701422357169471e-07,
+      "loss": 0.532,
+      "step": 7140
+    },
+    {
+      "epoch": 0.3962096863570874,
+      "grad_norm": 1.29000723361969,
+      "learning_rate": 6.693151859496493e-07,
+      "loss": 0.5361,
+      "step": 7150
+    },
+    {
+      "epoch": 0.3967638257785659,
+      "grad_norm": 1.4198318719863892,
+      "learning_rate": 6.684876125945688e-07,
+      "loss": 0.533,
+      "step": 7160
+    },
+    {
+      "epoch": 0.3973179652000443,
+      "grad_norm": 1.3135550022125244,
+      "learning_rate": 6.676595182108813e-07,
+      "loss": 0.5307,
+      "step": 7170
+    },
+    {
+      "epoch": 0.3978721046215228,
+      "grad_norm": 1.2542059421539307,
+      "learning_rate": 6.668309053593733e-07,
+      "loss": 0.535,
+      "step": 7180
+    },
+    {
+      "epoch": 0.3984262440430012,
+      "grad_norm": 1.2441365718841553,
+      "learning_rate": 6.660017766024353e-07,
+      "loss": 0.538,
+      "step": 7190
+    },
+    {
+      "epoch": 0.39898038346447967,
+      "grad_norm": 1.4032853841781616,
+      "learning_rate": 6.651721345040526e-07,
+      "loss": 0.5304,
+      "step": 7200
+    },
+    {
+      "epoch": 0.39898038346447967,
+      "eval_robovqa_loss": 0.08628591895103455,
+      "step": 7200
+    },
+    {
+      "epoch": 0.39898038346447967,
+      "eval_agibot_loss": 0.4522637128829956,
+      "step": 7200
+    },
+    {
+      "epoch": 0.39898038346447967,
+      "eval_holoassist_loss": 0.33070409297943115,
+      "step": 7200
+    },
+    {
+      "epoch": 0.39898038346447967,
+      "eval_robofail_loss": 0.205685555934906,
+      "step": 7200
+    },
+    {
+      "epoch": 0.3995345228859581,
+      "grad_norm": 1.3348402976989746,
+      "learning_rate": 6.643419816297984e-07,
+      "loss": 0.5344,
+      "step": 7210
+    },
+    {
+      "epoch": 0.40008866230743656,
+      "grad_norm": 1.4468635320663452,
+      "learning_rate": 6.635113205468251e-07,
+      "loss": 0.529,
+      "step": 7220
+    },
+    {
+      "epoch": 0.400642801728915,
+      "grad_norm": 1.2653415203094482,
+      "learning_rate": 6.626801538238567e-07,
+      "loss": 0.5317,
+      "step": 7230
+    },
+    {
+      "epoch": 0.40119694115039345,
+      "grad_norm": 1.2316300868988037,
+      "learning_rate": 6.618484840311809e-07,
+      "loss": 0.5302,
+      "step": 7240
+    },
+    {
+      "epoch": 0.40175108057187187,
+      "grad_norm": 1.4281355142593384,
+      "learning_rate": 6.610163137406414e-07,
+      "loss": 0.527,
+      "step": 7250
+    },
+    {
+      "epoch": 0.40230521999335034,
+      "grad_norm": 1.2586145401000977,
+      "learning_rate": 6.60183645525629e-07,
+      "loss": 0.5254,
+      "step": 7260
+    },
+    {
+      "epoch": 0.40285935941482875,
+      "grad_norm": 1.3509401082992554,
+      "learning_rate": 6.593504819610746e-07,
+      "loss": 0.532,
+      "step": 7270
+    },
+    {
+      "epoch": 0.4034134988363072,
+      "grad_norm": 1.2600640058517456,
+      "learning_rate": 6.585168256234412e-07,
+      "loss": 0.5282,
+      "step": 7280
+    },
+    {
+      "epoch": 0.40396763825778564,
+      "grad_norm": 1.6616328954696655,
+      "learning_rate": 6.576826790907151e-07,
+      "loss": 0.5247,
+      "step": 7290
+    },
+    {
+      "epoch": 0.4045217776792641,
+      "grad_norm": 1.181994080543518,
+      "learning_rate": 6.568480449423989e-07,
+      "loss": 0.5243,
+      "step": 7300
+    },
+    {
+      "epoch": 0.4045217776792641,
+      "eval_robovqa_loss": 0.08699452131986618,
+      "step": 7300
+    },
+    {
+      "epoch": 0.4045217776792641,
+      "eval_agibot_loss": 0.44423767924308777,
+      "step": 7300
+    },
+    {
+      "epoch": 0.4045217776792641,
+      "eval_holoassist_loss": 0.33769461512565613,
+      "step": 7300
+    },
+    {
+      "epoch": 0.4045217776792641,
+      "eval_robofail_loss": 0.204865500330925,
+      "step": 7300
+    },
+    {
+      "epoch": 0.40507591710074253,
+      "grad_norm": 1.1357425451278687,
+      "learning_rate": 6.560129257595026e-07,
+      "loss": 0.5326,
+      "step": 7310
+    },
+    {
+      "epoch": 0.405630056522221,
+      "grad_norm": 1.2737231254577637,
+      "learning_rate": 6.551773241245368e-07,
+      "loss": 0.5305,
+      "step": 7320
+    },
+    {
+      "epoch": 0.4061841959436994,
+      "grad_norm": 1.3256807327270508,
+      "learning_rate": 6.543412426215037e-07,
+      "loss": 0.5305,
+      "step": 7330
+    },
+    {
+      "epoch": 0.4067383353651779,
+      "grad_norm": 1.3407604694366455,
+      "learning_rate": 6.535046838358891e-07,
+      "loss": 0.5353,
+      "step": 7340
+    },
+    {
+      "epoch": 0.4072924747866563,
+      "grad_norm": 1.5440951585769653,
+      "learning_rate": 6.526676503546552e-07,
+      "loss": 0.5308,
+      "step": 7350
+    },
+    {
+      "epoch": 0.4078466142081348,
+      "grad_norm": 1.2832542657852173,
+      "learning_rate": 6.51830144766232e-07,
+      "loss": 0.5288,
+      "step": 7360
+    },
+    {
+      "epoch": 0.4084007536296132,
+      "grad_norm": 1.214943528175354,
+      "learning_rate": 6.509921696605093e-07,
+      "loss": 0.5369,
+      "step": 7370
+    },
+    {
+      "epoch": 0.4089548930510917,
+      "grad_norm": 1.3448729515075684,
+      "learning_rate": 6.501537276288291e-07,
+      "loss": 0.5244,
+      "step": 7380
+    },
+    {
+      "epoch": 0.4095090324725701,
+      "grad_norm": 1.2517979145050049,
+      "learning_rate": 6.493148212639772e-07,
+      "loss": 0.5261,
+      "step": 7390
+    },
+    {
+      "epoch": 0.41006317189404856,
+      "grad_norm": 1.728542685508728,
+      "learning_rate": 6.48475453160175e-07,
+      "loss": 0.5233,
+      "step": 7400
+    },
+    {
+      "epoch": 0.41006317189404856,
+      "eval_robovqa_loss": 0.08512812852859497,
+      "step": 7400
+    },
+    {
+      "epoch": 0.41006317189404856,
+      "eval_agibot_loss": 0.44625118374824524,
+      "step": 7400
+    },
+    {
+      "epoch": 0.41006317189404856,
+      "eval_holoassist_loss": 0.322357714176178,
+      "step": 7400
+    },
+    {
+      "epoch": 0.41006317189404856,
+      "eval_robofail_loss": 0.20179899036884308,
+      "step": 7400
+    },
+    {
+      "epoch": 0.410617311315527,
+      "grad_norm": 1.193760633468628,
+      "learning_rate": 6.476356259130723e-07,
+      "loss": 0.5283,
+      "step": 7410
+    },
+    {
+      "epoch": 0.41117145073700545,
+      "grad_norm": 1.2643742561340332,
+      "learning_rate": 6.467953421197385e-07,
+      "loss": 0.5256,
+      "step": 7420
+    },
+    {
+      "epoch": 0.41172559015848387,
+      "grad_norm": 1.2828691005706787,
+      "learning_rate": 6.459546043786547e-07,
+      "loss": 0.5298,
+      "step": 7430
+    },
+    {
+      "epoch": 0.41227972957996234,
+      "grad_norm": 1.2830382585525513,
+      "learning_rate": 6.451134152897057e-07,
+      "loss": 0.5307,
+      "step": 7440
+    },
+    {
+      "epoch": 0.41283386900144076,
+      "grad_norm": 1.2327229976654053,
+      "learning_rate": 6.442717774541726e-07,
+      "loss": 0.5354,
+      "step": 7450
+    },
+    {
+      "epoch": 0.41338800842291923,
+      "grad_norm": 1.2327628135681152,
+      "learning_rate": 6.434296934747236e-07,
+      "loss": 0.533,
+      "step": 7460
+    },
+    {
+      "epoch": 0.41394214784439765,
+      "grad_norm": 1.1679733991622925,
+      "learning_rate": 6.425871659554068e-07,
+      "loss": 0.5339,
+      "step": 7470
+    },
+    {
+      "epoch": 0.4144962872658761,
+      "grad_norm": 1.2368323802947998,
+      "learning_rate": 6.41744197501642e-07,
+      "loss": 0.5352,
+      "step": 7480
+    },
+    {
+      "epoch": 0.41505042668735453,
+      "grad_norm": 1.261178731918335,
+      "learning_rate": 6.409007907202123e-07,
+      "loss": 0.5361,
+      "step": 7490
+    },
+    {
+      "epoch": 0.415604566108833,
+      "grad_norm": 1.2287451028823853,
+      "learning_rate": 6.400569482192564e-07,
+      "loss": 0.5295,
+      "step": 7500
+    },
+    {
+      "epoch": 0.415604566108833,
+      "eval_robovqa_loss": 0.08477333188056946,
+      "step": 7500
+    },
+    {
+      "epoch": 0.415604566108833,
+      "eval_agibot_loss": 0.4514187276363373,
+      "step": 7500
+    },
+    {
+      "epoch": 0.415604566108833,
+      "eval_holoassist_loss": 0.32874757051467896,
+      "step": 7500
+    },
+    {
+      "epoch": 0.415604566108833,
+      "eval_robofail_loss": 0.20196111500263214,
+      "step": 7500
+    },
+    {
+      "epoch": 0.4161587055303114,
+      "grad_norm": 1.2695033550262451,
+      "learning_rate": 6.392126726082603e-07,
+      "loss": 0.5233,
+      "step": 7510
+    },
+    {
+      "epoch": 0.4167128449517899,
+      "grad_norm": 1.1952930688858032,
+      "learning_rate": 6.383679664980494e-07,
+      "loss": 0.5233,
+      "step": 7520
+    },
+    {
+      "epoch": 0.4172669843732683,
+      "grad_norm": 1.1895166635513306,
+      "learning_rate": 6.375228325007807e-07,
+      "loss": 0.528,
+      "step": 7530
+    },
+    {
+      "epoch": 0.4178211237947468,
+      "grad_norm": 1.2965315580368042,
+      "learning_rate": 6.366772732299339e-07,
+      "loss": 0.528,
+      "step": 7540
+    },
+    {
+      "epoch": 0.4183752632162252,
+      "grad_norm": 1.4195008277893066,
+      "learning_rate": 6.358312913003041e-07,
+      "loss": 0.5228,
+      "step": 7550
+    },
+    {
+      "epoch": 0.4189294026377037,
+      "grad_norm": 1.5474796295166016,
+      "learning_rate": 6.349848893279931e-07,
+      "loss": 0.5223,
+      "step": 7560
+    },
+    {
+      "epoch": 0.4194835420591821,
+      "grad_norm": 1.360557198524475,
+      "learning_rate": 6.341380699304023e-07,
+      "loss": 0.517,
+      "step": 7570
+    },
+    {
+      "epoch": 0.4200376814806605,
+      "grad_norm": 1.2077003717422485,
+      "learning_rate": 6.332908357262232e-07,
+      "loss": 0.5333,
+      "step": 7580
+    },
+    {
+      "epoch": 0.420591820902139,
+      "grad_norm": 1.272471308708191,
+      "learning_rate": 6.324431893354305e-07,
+      "loss": 0.5227,
+      "step": 7590
+    },
+    {
+      "epoch": 0.4211459603236174,
+      "grad_norm": 1.231727957725525,
+      "learning_rate": 6.31595133379273e-07,
+      "loss": 0.5227,
+      "step": 7600
+    },
+    {
+      "epoch": 0.4211459603236174,
+      "eval_robovqa_loss": 0.08577543497085571,
+      "step": 7600
+    },
+    {
+      "epoch": 0.4211459603236174,
+      "eval_agibot_loss": 0.4520198106765747,
+      "step": 7600
+    },
+    {
+      "epoch": 0.4211459603236174,
+      "eval_holoassist_loss": 0.33360981941223145,
+      "step": 7600
+    },
+    {
+      "epoch": 0.4211459603236174,
+      "eval_robofail_loss": 0.20208145678043365,
+      "step": 7600
+    },
+    {
+      "epoch": 0.42170009974509587,
+      "grad_norm": 1.2620092630386353,
+      "learning_rate": 6.307466704802671e-07,
+      "loss": 0.5364,
+      "step": 7610
+    },
+    {
+      "epoch": 0.4222542391665743,
+      "grad_norm": 1.461891770362854,
+      "learning_rate": 6.298978032621864e-07,
+      "loss": 0.5184,
+      "step": 7620
+    },
+    {
+      "epoch": 0.42280837858805276,
+      "grad_norm": 1.262750506401062,
+      "learning_rate": 6.290485343500556e-07,
+      "loss": 0.529,
+      "step": 7630
+    },
+    {
+      "epoch": 0.4233625180095312,
+      "grad_norm": 1.188185691833496,
+      "learning_rate": 6.281988663701414e-07,
+      "loss": 0.5287,
+      "step": 7640
+    },
+    {
+      "epoch": 0.42391665743100965,
+      "grad_norm": 1.434852123260498,
+      "learning_rate": 6.273488019499442e-07,
+      "loss": 0.5312,
+      "step": 7650
+    },
+    {
+      "epoch": 0.42447079685248806,
+      "grad_norm": 1.1874892711639404,
+      "learning_rate": 6.264983437181911e-07,
+      "loss": 0.5239,
+      "step": 7660
+    },
+    {
+      "epoch": 0.42502493627396654,
+      "grad_norm": 1.2469977140426636,
+      "learning_rate": 6.256474943048263e-07,
+      "loss": 0.5324,
+      "step": 7670
+    },
+    {
+      "epoch": 0.42557907569544495,
+      "grad_norm": 1.3038097620010376,
+      "learning_rate": 6.24796256341004e-07,
+      "loss": 0.5281,
+      "step": 7680
+    },
+    {
+      "epoch": 0.4261332151169234,
+      "grad_norm": 1.2090708017349243,
+      "learning_rate": 6.239446324590801e-07,
+      "loss": 0.5262,
+      "step": 7690
+    },
+    {
+      "epoch": 0.42668735453840184,
+      "grad_norm": 1.1934832334518433,
+      "learning_rate": 6.230926252926037e-07,
+      "loss": 0.5156,
+      "step": 7700
+    },
+    {
+      "epoch": 0.42668735453840184,
+      "eval_robovqa_loss": 0.08639564365148544,
+      "step": 7700
+    },
+    {
+      "epoch": 0.42668735453840184,
+      "eval_agibot_loss": 0.4509347975254059,
+      "step": 7700
+    },
+    {
+      "epoch": 0.42668735453840184,
+      "eval_holoassist_loss": 0.33126771450042725,
+      "step": 7700
+    },
+    {
+      "epoch": 0.42668735453840184,
+      "eval_robofail_loss": 0.2038162350654602,
+      "step": 7700
+    },
+    {
+      "epoch": 0.4272414939598803,
+      "grad_norm": 1.2437551021575928,
+      "learning_rate": 6.22240237476309e-07,
+      "loss": 0.5235,
+      "step": 7710
+    },
+    {
+      "epoch": 0.42779563338135873,
+      "grad_norm": 1.2918784618377686,
+      "learning_rate": 6.213874716461074e-07,
+      "loss": 0.5256,
+      "step": 7720
+    },
+    {
+      "epoch": 0.4283497728028372,
+      "grad_norm": 2.415844440460205,
+      "learning_rate": 6.205343304390797e-07,
+      "loss": 0.5353,
+      "step": 7730
+    },
+    {
+      "epoch": 0.4289039122243156,
+      "grad_norm": 1.2934114933013916,
+      "learning_rate": 6.19680816493467e-07,
+      "loss": 0.5339,
+      "step": 7740
+    },
+    {
+      "epoch": 0.4294580516457941,
+      "grad_norm": 1.2819117307662964,
+      "learning_rate": 6.188269324486634e-07,
+      "loss": 0.5264,
+      "step": 7750
+    },
+    {
+      "epoch": 0.4300121910672725,
+      "grad_norm": 1.27568781375885,
+      "learning_rate": 6.179726809452072e-07,
+      "loss": 0.5311,
+      "step": 7760
+    },
+    {
+      "epoch": 0.430566330488751,
+      "grad_norm": 1.2043118476867676,
+      "learning_rate": 6.171180646247729e-07,
+      "loss": 0.5271,
+      "step": 7770
+    },
+    {
+      "epoch": 0.4311204699102294,
+      "grad_norm": 1.3034669160842896,
+      "learning_rate": 6.162630861301639e-07,
+      "loss": 0.5264,
+      "step": 7780
+    },
+    {
+      "epoch": 0.43167460933170787,
+      "grad_norm": 1.2729039192199707,
+      "learning_rate": 6.154077481053028e-07,
+      "loss": 0.5266,
+      "step": 7790
+    },
+    {
+      "epoch": 0.4322287487531863,
+      "grad_norm": 1.2235006093978882,
+      "learning_rate": 6.145520531952244e-07,
+      "loss": 0.5296,
+      "step": 7800
+    },
+    {
+      "epoch": 0.4322287487531863,
+      "eval_robovqa_loss": 0.08441752195358276,
+      "step": 7800
+    },
+    {
+      "epoch": 0.4322287487531863,
+      "eval_agibot_loss": 0.4513845443725586,
+      "step": 7800
+    },
+    {
+      "epoch": 0.4322287487531863,
+      "eval_holoassist_loss": 0.3398887515068054,
+      "step": 7800
+    },
+    {
+      "epoch": 0.4322287487531863,
+      "eval_robofail_loss": 0.20558781921863556,
+      "step": 7800
+    },
+    {
+      "epoch": 0.43278288817466476,
+      "grad_norm": 1.346104621887207,
+      "learning_rate": 6.136960040460667e-07,
+      "loss": 0.5243,
+      "step": 7810
+    },
+    {
+      "epoch": 0.4333370275961432,
+      "grad_norm": 1.2971364259719849,
+      "learning_rate": 6.128396033050637e-07,
+      "loss": 0.5217,
+      "step": 7820
+    },
+    {
+      "epoch": 0.43389116701762165,
+      "grad_norm": 1.217315673828125,
+      "learning_rate": 6.119828536205366e-07,
+      "loss": 0.5272,
+      "step": 7830
+    },
+    {
+      "epoch": 0.43444530643910007,
+      "grad_norm": 1.2634752988815308,
+      "learning_rate": 6.111257576418853e-07,
+      "loss": 0.5257,
+      "step": 7840
+    },
+    {
+      "epoch": 0.43499944586057854,
+      "grad_norm": 1.1396344900131226,
+      "learning_rate": 6.102683180195805e-07,
+      "loss": 0.5335,
+      "step": 7850
+    },
+    {
+      "epoch": 0.43555358528205695,
+      "grad_norm": 1.2880114316940308,
+      "learning_rate": 6.094105374051561e-07,
+      "loss": 0.5231,
+      "step": 7860
+    },
+    {
+      "epoch": 0.4361077247035354,
+      "grad_norm": 1.2866604328155518,
+      "learning_rate": 6.085524184512003e-07,
+      "loss": 0.52,
+      "step": 7870
+    },
+    {
+      "epoch": 0.43666186412501384,
+      "grad_norm": 1.2020090818405151,
+      "learning_rate": 6.076939638113474e-07,
+      "loss": 0.5247,
+      "step": 7880
+    },
+    {
+      "epoch": 0.4372160035464923,
+      "grad_norm": 1.2148151397705078,
+      "learning_rate": 6.068351761402699e-07,
+      "loss": 0.5296,
+      "step": 7890
+    },
+    {
+      "epoch": 0.43777014296797073,
+      "grad_norm": 1.2650190591812134,
+      "learning_rate": 6.059760580936701e-07,
+      "loss": 0.5264,
+      "step": 7900
+    },
+    {
+      "epoch": 0.43777014296797073,
+      "eval_robovqa_loss": 0.08556153625249863,
+      "step": 7900
+    },
+    {
+      "epoch": 0.43777014296797073,
+      "eval_agibot_loss": 0.4521845579147339,
+      "step": 7900
+    },
+    {
+      "epoch": 0.43777014296797073,
+      "eval_holoassist_loss": 0.33047154545783997,
+      "step": 7900
+    },
+    {
+      "epoch": 0.43777014296797073,
+      "eval_robofail_loss": 0.20660759508609772,
+      "step": 7900
+    },
+    {
+      "epoch": 0.4383242823894492,
+      "grad_norm": 1.1766793727874756,
+      "learning_rate": 6.051166123282718e-07,
+      "loss": 0.5196,
+      "step": 7910
+    },
+    {
+      "epoch": 0.4388784218109276,
+      "grad_norm": 1.3174407482147217,
+      "learning_rate": 6.042568415018125e-07,
+      "loss": 0.5158,
+      "step": 7920
+    },
+    {
+      "epoch": 0.4394325612324061,
+      "grad_norm": 1.3563159704208374,
+      "learning_rate": 6.033967482730351e-07,
+      "loss": 0.5272,
+      "step": 7930
+    },
+    {
+      "epoch": 0.4399867006538845,
+      "grad_norm": 1.1922650337219238,
+      "learning_rate": 6.025363353016787e-07,
+      "loss": 0.5239,
+      "step": 7940
+    },
+    {
+      "epoch": 0.440540840075363,
+      "grad_norm": 1.2641743421554565,
+      "learning_rate": 6.01675605248472e-07,
+      "loss": 0.5293,
+      "step": 7950
+    },
+    {
+      "epoch": 0.4410949794968414,
+      "grad_norm": 1.3047740459442139,
+      "learning_rate": 6.00814560775124e-07,
+      "loss": 0.5244,
+      "step": 7960
+    },
+    {
+      "epoch": 0.44164911891831987,
+      "grad_norm": 1.2906373739242554,
+      "learning_rate": 5.999532045443157e-07,
+      "loss": 0.5157,
+      "step": 7970
+    },
+    {
+      "epoch": 0.4422032583397983,
+      "grad_norm": 1.1945255994796753,
+      "learning_rate": 5.990915392196924e-07,
+      "loss": 0.5237,
+      "step": 7980
+    },
+    {
+      "epoch": 0.44275739776127676,
+      "grad_norm": 1.317314863204956,
+      "learning_rate": 5.982295674658557e-07,
+      "loss": 0.5282,
+      "step": 7990
+    },
+    {
+      "epoch": 0.4433115371827552,
+      "grad_norm": 1.2961201667785645,
+      "learning_rate": 5.973672919483537e-07,
+      "loss": 0.5237,
+      "step": 8000
+    },
+    {
+      "epoch": 0.4433115371827552,
+      "eval_robovqa_loss": 0.08898232877254486,
+      "step": 8000
+    },
+    {
+      "epoch": 0.4433115371827552,
+      "eval_agibot_loss": 0.4464643895626068,
+      "step": 8000
+    },
+    {
+      "epoch": 0.4433115371827552,
+      "eval_holoassist_loss": 0.3426443934440613,
+      "step": 8000
+    },
+    {
+      "epoch": 0.4433115371827552,
+      "eval_robofail_loss": 0.20700007677078247,
+      "step": 8000
+    },
+    {
+      "epoch": 0.44386567660423365,
+      "grad_norm": 1.3610610961914062,
+      "learning_rate": 5.965047153336747e-07,
+      "loss": 0.5381,
+      "step": 8010
+    },
+    {
+      "epoch": 0.44441981602571207,
+      "grad_norm": 1.2463908195495605,
+      "learning_rate": 5.956418402892379e-07,
+      "loss": 0.5204,
+      "step": 8020
+    },
+    {
+      "epoch": 0.44497395544719054,
+      "grad_norm": 1.3063359260559082,
+      "learning_rate": 5.947786694833855e-07,
+      "loss": 0.5232,
+      "step": 8030
+    },
+    {
+      "epoch": 0.44552809486866896,
+      "grad_norm": 1.2920444011688232,
+      "learning_rate": 5.939152055853741e-07,
+      "loss": 0.5238,
+      "step": 8040
+    },
+    {
+      "epoch": 0.44608223429014743,
+      "grad_norm": 1.461942434310913,
+      "learning_rate": 5.930514512653666e-07,
+      "loss": 0.5344,
+      "step": 8050
+    },
+    {
+      "epoch": 0.44663637371162584,
+      "grad_norm": 1.187874674797058,
+      "learning_rate": 5.921874091944241e-07,
+      "loss": 0.53,
+      "step": 8060
+    },
+    {
+      "epoch": 0.44719051313310426,
+      "grad_norm": 1.2797425985336304,
+      "learning_rate": 5.913230820444979e-07,
+      "loss": 0.5262,
+      "step": 8070
+    },
+    {
+      "epoch": 0.44774465255458273,
+      "grad_norm": 1.3904670476913452,
+      "learning_rate": 5.904584724884201e-07,
+      "loss": 0.5214,
+      "step": 8080
+    },
+    {
+      "epoch": 0.44829879197606115,
+      "grad_norm": 1.3671061992645264,
+      "learning_rate": 5.895935831998968e-07,
+      "loss": 0.5229,
+      "step": 8090
+    },
+    {
+      "epoch": 0.4488529313975396,
+      "grad_norm": 1.1164793968200684,
+      "learning_rate": 5.887284168534986e-07,
+      "loss": 0.5231,
+      "step": 8100
+    },
+    {
+      "epoch": 0.4488529313975396,
+      "eval_robovqa_loss": 0.08495265245437622,
+      "step": 8100
+    },
+    {
+      "epoch": 0.4488529313975396,
+      "eval_agibot_loss": 0.45666274428367615,
+      "step": 8100
+    },
+    {
+      "epoch": 0.4488529313975396,
+      "eval_holoassist_loss": 0.3508586585521698,
+      "step": 8100
+    },
+    {
+      "epoch": 0.4488529313975396,
+      "eval_robofail_loss": 0.20492865145206451,
+      "step": 8100
+    },
+    {
+      "epoch": 0.44940707081901804,
+      "grad_norm": 1.224230408668518,
+      "learning_rate": 5.878629761246536e-07,
+      "loss": 0.5205,
+      "step": 8110
+    },
+    {
+      "epoch": 0.4499612102404965,
+      "grad_norm": 1.3383800983428955,
+      "learning_rate": 5.869972636896375e-07,
+      "loss": 0.5258,
+      "step": 8120
+    },
+    {
+      "epoch": 0.45051534966197493,
+      "grad_norm": 1.2039337158203125,
+      "learning_rate": 5.861312822255667e-07,
+      "loss": 0.5251,
+      "step": 8130
+    },
+    {
+      "epoch": 0.4510694890834534,
+      "grad_norm": 1.3457081317901611,
+      "learning_rate": 5.852650344103897e-07,
+      "loss": 0.5362,
+      "step": 8140
+    },
+    {
+      "epoch": 0.4516236285049318,
+      "grad_norm": 1.3456610441207886,
+      "learning_rate": 5.843985229228784e-07,
+      "loss": 0.5229,
+      "step": 8150
+    },
+    {
+      "epoch": 0.4521777679264103,
+      "grad_norm": 1.2955739498138428,
+      "learning_rate": 5.835317504426201e-07,
+      "loss": 0.5302,
+      "step": 8160
+    },
+    {
+      "epoch": 0.4527319073478887,
+      "grad_norm": 1.1416856050491333,
+      "learning_rate": 5.826647196500095e-07,
+      "loss": 0.5204,
+      "step": 8170
+    },
+    {
+      "epoch": 0.4532860467693672,
+      "grad_norm": 1.3080930709838867,
+      "learning_rate": 5.817974332262393e-07,
+      "loss": 0.5225,
+      "step": 8180
+    },
+    {
+      "epoch": 0.4538401861908456,
+      "grad_norm": 1.9958277940750122,
+      "learning_rate": 5.809298938532935e-07,
+      "loss": 0.5208,
+      "step": 8190
+    },
+    {
+      "epoch": 0.45439432561232407,
+      "grad_norm": 1.3589421510696411,
+      "learning_rate": 5.800621042139382e-07,
+      "loss": 0.5275,
+      "step": 8200
+    },
+    {
+      "epoch": 0.45439432561232407,
+      "eval_robovqa_loss": 0.08120116591453552,
+      "step": 8200
+    },
+    {
+      "epoch": 0.45439432561232407,
+      "eval_agibot_loss": 0.46431756019592285,
+      "step": 8200
+    },
+    {
+      "epoch": 0.45439432561232407,
+      "eval_holoassist_loss": 0.3422242999076843,
+      "step": 8200
+    },
+    {
+      "epoch": 0.45439432561232407,
+      "eval_robofail_loss": 0.20700275897979736,
+      "step": 8200
+    },
+    {
+      "epoch": 0.4549484650338025,
+      "grad_norm": 1.1897040605545044,
+      "learning_rate": 5.791940669917132e-07,
+      "loss": 0.5212,
+      "step": 8210
+    },
+    {
+      "epoch": 0.45550260445528096,
+      "grad_norm": 1.279625415802002,
+      "learning_rate": 5.78325784870924e-07,
+      "loss": 0.5195,
+      "step": 8220
+    },
+    {
+      "epoch": 0.4560567438767594,
+      "grad_norm": 1.3310792446136475,
+      "learning_rate": 5.774572605366331e-07,
+      "loss": 0.5242,
+      "step": 8230
+    },
+    {
+      "epoch": 0.45661088329823785,
+      "grad_norm": 1.2163292169570923,
+      "learning_rate": 5.765884966746528e-07,
+      "loss": 0.5213,
+      "step": 8240
+    },
+    {
+      "epoch": 0.45716502271971626,
+      "grad_norm": 1.4702321290969849,
+      "learning_rate": 5.757194959715355e-07,
+      "loss": 0.5268,
+      "step": 8250
+    },
+    {
+      "epoch": 0.45771916214119474,
+      "grad_norm": 1.3633860349655151,
+      "learning_rate": 5.748502611145661e-07,
+      "loss": 0.5205,
+      "step": 8260
+    },
+    {
+      "epoch": 0.45827330156267315,
+      "grad_norm": 1.305685043334961,
+      "learning_rate": 5.739807947917534e-07,
+      "loss": 0.5306,
+      "step": 8270
+    },
+    {
+      "epoch": 0.4588274409841516,
+      "grad_norm": 1.230681300163269,
+      "learning_rate": 5.731110996918227e-07,
+      "loss": 0.5252,
+      "step": 8280
+    },
+    {
+      "epoch": 0.45938158040563004,
+      "grad_norm": 1.269952416419983,
+      "learning_rate": 5.722411785042061e-07,
+      "loss": 0.5206,
+      "step": 8290
+    },
+    {
+      "epoch": 0.4599357198271085,
+      "grad_norm": 1.1457382440567017,
+      "learning_rate": 5.71371033919035e-07,
+      "loss": 0.522,
+      "step": 8300
+    },
+    {
+      "epoch": 0.4599357198271085,
+      "eval_robovqa_loss": 0.08240503072738647,
+      "step": 8300
+    },
+    {
+      "epoch": 0.4599357198271085,
+      "eval_agibot_loss": 0.4640901982784271,
+      "step": 8300
+    },
+    {
+      "epoch": 0.4599357198271085,
+      "eval_holoassist_loss": 0.34453919529914856,
+      "step": 8300
+    },
+    {
+      "epoch": 0.4599357198271085,
+      "eval_robofail_loss": 0.20523151755332947,
+      "step": 8300
+    },
+    {
+      "epoch": 0.46048985924858693,
+      "grad_norm": 1.278571605682373,
+      "learning_rate": 5.705006686271316e-07,
+      "loss": 0.5214,
+      "step": 8310
+    },
+    {
+      "epoch": 0.4610439986700654,
+      "grad_norm": 1.1903367042541504,
+      "learning_rate": 5.696300853200009e-07,
+      "loss": 0.5158,
+      "step": 8320
+    },
+    {
+      "epoch": 0.4615981380915438,
+      "grad_norm": 1.274410367012024,
+      "learning_rate": 5.687592866898218e-07,
+      "loss": 0.5249,
+      "step": 8330
+    },
+    {
+      "epoch": 0.4621522775130223,
+      "grad_norm": 1.221105933189392,
+      "learning_rate": 5.67888275429439e-07,
+      "loss": 0.513,
+      "step": 8340
+    },
+    {
+      "epoch": 0.4627064169345007,
+      "grad_norm": 1.277332067489624,
+      "learning_rate": 5.670170542323548e-07,
+      "loss": 0.519,
+      "step": 8350
+    },
+    {
+      "epoch": 0.4632605563559792,
+      "grad_norm": 1.3239927291870117,
+      "learning_rate": 5.661456257927211e-07,
+      "loss": 0.5215,
+      "step": 8360
+    },
+    {
+      "epoch": 0.4638146957774576,
+      "grad_norm": 1.310385823249817,
+      "learning_rate": 5.652739928053301e-07,
+      "loss": 0.5257,
+      "step": 8370
+    },
+    {
+      "epoch": 0.46436883519893607,
+      "grad_norm": 1.2557979822158813,
+      "learning_rate": 5.644021579656071e-07,
+      "loss": 0.519,
+      "step": 8380
+    },
+    {
+      "epoch": 0.4649229746204145,
+      "grad_norm": 1.186273455619812,
+      "learning_rate": 5.635301239696009e-07,
+      "loss": 0.5213,
+      "step": 8390
+    },
+    {
+      "epoch": 0.46547711404189296,
+      "grad_norm": 1.2588896751403809,
+      "learning_rate": 5.626578935139767e-07,
+      "loss": 0.5185,
+      "step": 8400
+    },
+    {
+      "epoch": 0.46547711404189296,
+      "eval_robovqa_loss": 0.08149432390928268,
+      "step": 8400
+    },
+    {
+      "epoch": 0.46547711404189296,
+      "eval_agibot_loss": 0.463846892118454,
+      "step": 8400
+    },
+    {
+      "epoch": 0.46547711404189296,
+      "eval_holoassist_loss": 0.3400147259235382,
+      "step": 8400
+    },
+    {
+      "epoch": 0.46547711404189296,
+      "eval_robofail_loss": 0.203947976231575,
+      "step": 8400
+    },
+    {
+      "epoch": 0.4660312534633714,
+      "grad_norm": 1.2429656982421875,
+      "learning_rate": 5.617854692960072e-07,
+      "loss": 0.5152,
+      "step": 8410
+    },
+    {
+      "epoch": 0.46658539288484985,
+      "grad_norm": 1.1782028675079346,
+      "learning_rate": 5.609128540135642e-07,
+      "loss": 0.5247,
+      "step": 8420
+    },
+    {
+      "epoch": 0.46713953230632826,
+      "grad_norm": 1.2485072612762451,
+      "learning_rate": 5.600400503651101e-07,
+      "loss": 0.5243,
+      "step": 8430
+    },
+    {
+      "epoch": 0.46769367172780674,
+      "grad_norm": 1.4017689228057861,
+      "learning_rate": 5.591670610496902e-07,
+      "loss": 0.5185,
+      "step": 8440
+    },
+    {
+      "epoch": 0.46824781114928515,
+      "grad_norm": 1.2907381057739258,
+      "learning_rate": 5.582938887669236e-07,
+      "loss": 0.5234,
+      "step": 8450
+    },
+    {
+      "epoch": 0.4688019505707636,
+      "grad_norm": 1.2340366840362549,
+      "learning_rate": 5.574205362169956e-07,
+      "loss": 0.5237,
+      "step": 8460
+    },
+    {
+      "epoch": 0.46935608999224204,
+      "grad_norm": 1.3373587131500244,
+      "learning_rate": 5.565470061006486e-07,
+      "loss": 0.5253,
+      "step": 8470
+    },
+    {
+      "epoch": 0.4699102294137205,
+      "grad_norm": 1.3041826486587524,
+      "learning_rate": 5.556733011191743e-07,
+      "loss": 0.5321,
+      "step": 8480
+    },
+    {
+      "epoch": 0.47046436883519893,
+      "grad_norm": 1.2078323364257812,
+      "learning_rate": 5.547994239744052e-07,
+      "loss": 0.5191,
+      "step": 8490
+    },
+    {
+      "epoch": 0.4710185082566774,
+      "grad_norm": 1.266006588935852,
+      "learning_rate": 5.539253773687059e-07,
+      "loss": 0.5251,
+      "step": 8500
+    },
+    {
+      "epoch": 0.4710185082566774,
+      "eval_robovqa_loss": 0.08348560333251953,
+      "step": 8500
+    },
+    {
+      "epoch": 0.4710185082566774,
+      "eval_agibot_loss": 0.4687187969684601,
+      "step": 8500
+    },
+    {
+      "epoch": 0.4710185082566774,
+      "eval_holoassist_loss": 0.3436199426651001,
+      "step": 8500
+    },
+    {
+      "epoch": 0.4710185082566774,
+      "eval_robofail_loss": 0.20404013991355896,
+      "step": 8500
+    },
+    {
+      "epoch": 0.4715726476781558,
+      "grad_norm": 1.2648141384124756,
+      "learning_rate": 5.530511640049653e-07,
+      "loss": 0.5243,
+      "step": 8510
+    },
+    {
+      "epoch": 0.4721267870996343,
+      "grad_norm": 1.1458015441894531,
+      "learning_rate": 5.521767865865879e-07,
+      "loss": 0.5262,
+      "step": 8520
+    },
+    {
+      "epoch": 0.4726809265211127,
+      "grad_norm": 1.275078535079956,
+      "learning_rate": 5.513022478174856e-07,
+      "loss": 0.5181,
+      "step": 8530
+    },
+    {
+      "epoch": 0.4732350659425912,
+      "grad_norm": 1.1977602243423462,
+      "learning_rate": 5.50427550402069e-07,
+      "loss": 0.5061,
+      "step": 8540
+    },
+    {
+      "epoch": 0.4737892053640696,
+      "grad_norm": 1.2823349237442017,
+      "learning_rate": 5.495526970452397e-07,
+      "loss": 0.5229,
+      "step": 8550
+    },
+    {
+      "epoch": 0.47434334478554807,
+      "grad_norm": 1.3064125776290894,
+      "learning_rate": 5.486776904523811e-07,
+      "loss": 0.521,
+      "step": 8560
+    },
+    {
+      "epoch": 0.4748974842070265,
+      "grad_norm": 1.248189091682434,
+      "learning_rate": 5.478025333293509e-07,
+      "loss": 0.522,
+      "step": 8570
+    },
+    {
+      "epoch": 0.4754516236285049,
+      "grad_norm": 1.2854843139648438,
+      "learning_rate": 5.46927228382472e-07,
+      "loss": 0.5175,
+      "step": 8580
+    },
+    {
+      "epoch": 0.4760057630499834,
+      "grad_norm": 1.2980250120162964,
+      "learning_rate": 5.460517783185243e-07,
+      "loss": 0.5272,
+      "step": 8590
+    },
+    {
+      "epoch": 0.4765599024714618,
+      "grad_norm": 1.3241406679153442,
+      "learning_rate": 5.45176185844737e-07,
+      "loss": 0.5278,
+      "step": 8600
+    },
+    {
+      "epoch": 0.4765599024714618,
+      "eval_robovqa_loss": 0.08678826689720154,
+      "step": 8600
+    },
+    {
+      "epoch": 0.4765599024714618,
+      "eval_agibot_loss": 0.45207032561302185,
+      "step": 8600
+    },
+    {
+      "epoch": 0.4765599024714618,
+      "eval_holoassist_loss": 0.35347750782966614,
+      "step": 8600
+    },
+    {
+      "epoch": 0.4765599024714618,
+      "eval_robofail_loss": 0.20159921050071716,
+      "step": 8600
+    },
+    {
+      "epoch": 0.47711404189294027,
+      "grad_norm": 1.2569903135299683,
+      "learning_rate": 5.44300453668779e-07,
+      "loss": 0.5201,
+      "step": 8610
+    },
+    {
+      "epoch": 0.4776681813144187,
+      "grad_norm": 1.2563506364822388,
+      "learning_rate": 5.434245844987518e-07,
+      "loss": 0.5131,
+      "step": 8620
+    },
+    {
+      "epoch": 0.47822232073589716,
+      "grad_norm": 1.3135932683944702,
+      "learning_rate": 5.425485810431805e-07,
+      "loss": 0.526,
+      "step": 8630
+    },
+    {
+      "epoch": 0.47877646015737557,
+      "grad_norm": 1.190487027168274,
+      "learning_rate": 5.416724460110047e-07,
+      "loss": 0.5152,
+      "step": 8640
+    },
+    {
+      "epoch": 0.47933059957885404,
+      "grad_norm": 1.264605164527893,
+      "learning_rate": 5.407961821115717e-07,
+      "loss": 0.5279,
+      "step": 8650
+    },
+    {
+      "epoch": 0.47988473900033246,
+      "grad_norm": 1.3025952577590942,
+      "learning_rate": 5.399197920546272e-07,
+      "loss": 0.5208,
+      "step": 8660
+    },
+    {
+      "epoch": 0.48043887842181093,
+      "grad_norm": 1.235070824623108,
+      "learning_rate": 5.390432785503067e-07,
+      "loss": 0.5261,
+      "step": 8670
+    },
+    {
+      "epoch": 0.48099301784328935,
+      "grad_norm": 1.1629137992858887,
+      "learning_rate": 5.381666443091279e-07,
+      "loss": 0.5232,
+      "step": 8680
+    },
+    {
+      "epoch": 0.4815471572647678,
+      "grad_norm": 1.2841929197311401,
+      "learning_rate": 5.372898920419811e-07,
+      "loss": 0.5213,
+      "step": 8690
+    },
+    {
+      "epoch": 0.48210129668624624,
+      "grad_norm": 1.2826733589172363,
+      "learning_rate": 5.364130244601225e-07,
+      "loss": 0.5185,
+      "step": 8700
+    },
+    {
+      "epoch": 0.48210129668624624,
+      "eval_robovqa_loss": 0.08282157778739929,
+      "step": 8700
+    },
+    {
+      "epoch": 0.48210129668624624,
+      "eval_agibot_loss": 0.45119625329971313,
+      "step": 8700
+    },
+    {
+      "epoch": 0.48210129668624624,
+      "eval_holoassist_loss": 0.3349094092845917,
+      "step": 8700
+    },
+    {
+      "epoch": 0.48210129668624624,
+      "eval_robofail_loss": 0.20246322453022003,
+      "step": 8700
+    },
+    {
+      "epoch": 0.4826554361077247,
+      "grad_norm": 1.223839521408081,
+      "learning_rate": 5.355360442751645e-07,
+      "loss": 0.5166,
+      "step": 8710
+    },
+    {
+      "epoch": 0.48320957552920313,
+      "grad_norm": 1.2646616697311401,
+      "learning_rate": 5.346589541990673e-07,
+      "loss": 0.5143,
+      "step": 8720
+    },
+    {
+      "epoch": 0.4837637149506816,
+      "grad_norm": 1.371264100074768,
+      "learning_rate": 5.337817569441316e-07,
+      "loss": 0.5191,
+      "step": 8730
+    },
+    {
+      "epoch": 0.48431785437216,
+      "grad_norm": 1.1942094564437866,
+      "learning_rate": 5.329044552229892e-07,
+      "loss": 0.5221,
+      "step": 8740
+    },
+    {
+      "epoch": 0.4848719937936385,
+      "grad_norm": 1.2959182262420654,
+      "learning_rate": 5.320270517485949e-07,
+      "loss": 0.509,
+      "step": 8750
+    },
+    {
+      "epoch": 0.4854261332151169,
+      "grad_norm": 1.211887001991272,
+      "learning_rate": 5.311495492342185e-07,
+      "loss": 0.5147,
+      "step": 8760
+    },
+    {
+      "epoch": 0.4859802726365954,
+      "grad_norm": 1.232651948928833,
+      "learning_rate": 5.302719503934355e-07,
+      "loss": 0.5257,
+      "step": 8770
+    },
+    {
+      "epoch": 0.4865344120580738,
+      "grad_norm": 1.2840099334716797,
+      "learning_rate": 5.293942579401198e-07,
+      "loss": 0.5187,
+      "step": 8780
+    },
+    {
+      "epoch": 0.48708855147955227,
+      "grad_norm": 1.1625505685806274,
+      "learning_rate": 5.285164745884346e-07,
+      "loss": 0.5208,
+      "step": 8790
+    },
+    {
+      "epoch": 0.4876426909010307,
+      "grad_norm": 1.5289745330810547,
+      "learning_rate": 5.276386030528243e-07,
+      "loss": 0.5115,
+      "step": 8800
+    },
+    {
+      "epoch": 0.4876426909010307,
+      "eval_robovqa_loss": 0.08121342211961746,
+      "step": 8800
+    },
+    {
+      "epoch": 0.4876426909010307,
+      "eval_agibot_loss": 0.4579530954360962,
+      "step": 8800
+    },
+    {
+      "epoch": 0.4876426909010307,
+      "eval_holoassist_loss": 0.33154627680778503,
+      "step": 8800
+    },
+    {
+      "epoch": 0.4876426909010307,
+      "eval_robofail_loss": 0.2030121237039566,
+      "step": 8800
+    },
+    {
+      "epoch": 0.48819683032250916,
+      "grad_norm": 1.2941452264785767,
+      "learning_rate": 5.267606460480054e-07,
+      "loss": 0.5146,
+      "step": 8810
+    },
+    {
+      "epoch": 0.4887509697439876,
+      "grad_norm": 1.2435784339904785,
+      "learning_rate": 5.258826062889596e-07,
+      "loss": 0.5166,
+      "step": 8820
+    },
+    {
+      "epoch": 0.48930510916546605,
+      "grad_norm": 1.4292969703674316,
+      "learning_rate": 5.250044864909242e-07,
+      "loss": 0.5223,
+      "step": 8830
+    },
+    {
+      "epoch": 0.48985924858694446,
+      "grad_norm": 1.4205594062805176,
+      "learning_rate": 5.241262893693836e-07,
+      "loss": 0.5258,
+      "step": 8840
+    },
+    {
+      "epoch": 0.49041338800842293,
+      "grad_norm": 1.1854009628295898,
+      "learning_rate": 5.232480176400619e-07,
+      "loss": 0.5263,
+      "step": 8850
+    },
+    {
+      "epoch": 0.49096752742990135,
+      "grad_norm": 1.2696174383163452,
+      "learning_rate": 5.223696740189133e-07,
+      "loss": 0.5259,
+      "step": 8860
+    },
+    {
+      "epoch": 0.4915216668513798,
+      "grad_norm": 1.1964980363845825,
+      "learning_rate": 5.214912612221149e-07,
+      "loss": 0.523,
+      "step": 8870
+    },
+    {
+      "epoch": 0.49207580627285824,
+      "grad_norm": 1.265303611755371,
+      "learning_rate": 5.206127819660574e-07,
+      "loss": 0.5206,
+      "step": 8880
+    },
+    {
+      "epoch": 0.4926299456943367,
+      "grad_norm": 1.7318284511566162,
+      "learning_rate": 5.19734238967337e-07,
+      "loss": 0.5247,
+      "step": 8890
+    },
+    {
+      "epoch": 0.49318408511581513,
+      "grad_norm": 1.2413564920425415,
+      "learning_rate": 5.188556349427472e-07,
+      "loss": 0.5207,
+      "step": 8900
+    },
+    {
+      "epoch": 0.49318408511581513,
+      "eval_robovqa_loss": 0.08408524841070175,
+      "step": 8900
+    },
+    {
+      "epoch": 0.49318408511581513,
+      "eval_agibot_loss": 0.45730045437812805,
+      "step": 8900
+    },
+    {
+      "epoch": 0.49318408511581513,
+      "eval_holoassist_loss": 0.34044402837753296,
+      "step": 8900
+    },
+    {
+      "epoch": 0.49318408511581513,
+      "eval_robofail_loss": 0.2049257904291153,
+      "step": 8900
+    },
+    {
+      "epoch": 0.4937382245372936,
+      "grad_norm": 1.2619150876998901,
+      "learning_rate": 5.179769726092702e-07,
+      "loss": 0.52,
+      "step": 8910
+    },
+    {
+      "epoch": 0.494292363958772,
+      "grad_norm": 1.2281064987182617,
+      "learning_rate": 5.170982546840684e-07,
+      "loss": 0.5169,
+      "step": 8920
+    },
+    {
+      "epoch": 0.4948465033802505,
+      "grad_norm": 1.2819266319274902,
+      "learning_rate": 5.162194838844761e-07,
+      "loss": 0.5172,
+      "step": 8930
+    },
+    {
+      "epoch": 0.4954006428017289,
+      "grad_norm": 1.219595193862915,
+      "learning_rate": 5.153406629279911e-07,
+      "loss": 0.5197,
+      "step": 8940
+    },
+    {
+      "epoch": 0.4959547822232074,
+      "grad_norm": 1.2062796354293823,
+      "learning_rate": 5.144617945322664e-07,
+      "loss": 0.5173,
+      "step": 8950
+    },
+    {
+      "epoch": 0.4965089216446858,
+      "grad_norm": 1.2416777610778809,
+      "learning_rate": 5.135828814151017e-07,
+      "loss": 0.5165,
+      "step": 8960
+    },
+    {
+      "epoch": 0.49706306106616427,
+      "grad_norm": 1.2328424453735352,
+      "learning_rate": 5.127039262944349e-07,
+      "loss": 0.5146,
+      "step": 8970
+    },
+    {
+      "epoch": 0.4976172004876427,
+      "grad_norm": 1.3636809587478638,
+      "learning_rate": 5.118249318883339e-07,
+      "loss": 0.5145,
+      "step": 8980
+    },
+    {
+      "epoch": 0.49817133990912116,
+      "grad_norm": 1.3735278844833374,
+      "learning_rate": 5.10945900914988e-07,
+      "loss": 0.513,
+      "step": 8990
+    },
+    {
+      "epoch": 0.4987254793305996,
+      "grad_norm": 1.3093363046646118,
+      "learning_rate": 5.100668360926996e-07,
+      "loss": 0.5189,
+      "step": 9000
+    },
+    {
+      "epoch": 0.4987254793305996,
+      "eval_robovqa_loss": 0.08362621068954468,
+      "step": 9000
+    },
+    {
+      "epoch": 0.4987254793305996,
+      "eval_agibot_loss": 0.45920562744140625,
+      "step": 9000
+    },
+    {
+      "epoch": 0.4987254793305996,
+      "eval_holoassist_loss": 0.3466956317424774,
+      "step": 9000
+    },
+    {
+      "epoch": 0.4987254793305996,
+      "eval_robofail_loss": 0.20246131718158722,
+      "step": 9000
+    },
+    {
+      "epoch": 0.49927961875207805,
+      "grad_norm": 1.2721766233444214,
+      "learning_rate": 5.091877401398759e-07,
+      "loss": 0.5166,
+      "step": 9010
+    },
+    {
+      "epoch": 0.49983375817355646,
+      "grad_norm": 1.2730083465576172,
+      "learning_rate": 5.083086157750199e-07,
+      "loss": 0.5201,
+      "step": 9020
+    },
+    {
+      "epoch": 0.5003878975950349,
+      "grad_norm": 1.217538595199585,
+      "learning_rate": 5.074294657167233e-07,
+      "loss": 0.5196,
+      "step": 9030
+    },
+    {
+      "epoch": 0.5009420370165134,
+      "grad_norm": 1.2240993976593018,
+      "learning_rate": 5.065502926836565e-07,
+      "loss": 0.5189,
+      "step": 9040
+    },
+    {
+      "epoch": 0.5014961764379918,
+      "grad_norm": 1.7645894289016724,
+      "learning_rate": 5.056710993945614e-07,
+      "loss": 0.5214,
+      "step": 9050
+    },
+    {
+      "epoch": 0.5020503158594702,
+      "grad_norm": 1.3130478858947754,
+      "learning_rate": 5.047918885682419e-07,
+      "loss": 0.521,
+      "step": 9060
+    },
+    {
+      "epoch": 0.5026044552809487,
+      "grad_norm": 1.220805287361145,
+      "learning_rate": 5.039126629235571e-07,
+      "loss": 0.519,
+      "step": 9070
+    },
+    {
+      "epoch": 0.5031585947024271,
+      "grad_norm": 1.2196544408798218,
+      "learning_rate": 5.030334251794111e-07,
+      "loss": 0.5177,
+      "step": 9080
+    },
+    {
+      "epoch": 0.5037127341239056,
+      "grad_norm": 1.2827987670898438,
+      "learning_rate": 5.02154178054746e-07,
+      "loss": 0.5155,
+      "step": 9090
+    },
+    {
+      "epoch": 0.504266873545384,
+      "grad_norm": 1.1669912338256836,
+      "learning_rate": 5.012749242685323e-07,
+      "loss": 0.5156,
+      "step": 9100
+    },
+    {
+      "epoch": 0.504266873545384,
+      "eval_robovqa_loss": 0.08376478403806686,
+      "step": 9100
+    },
+    {
+      "epoch": 0.504266873545384,
+      "eval_agibot_loss": 0.45474299788475037,
+      "step": 9100
+    },
+    {
+      "epoch": 0.504266873545384,
+      "eval_holoassist_loss": 0.3400272727012634,
+      "step": 9100
+    },
+    {
+      "epoch": 0.504266873545384,
+      "eval_robofail_loss": 0.2031857818365097,
+      "step": 9100
+    },
+    {
+      "epoch": 0.5048210129668624,
+      "grad_norm": 1.2615147829055786,
+      "learning_rate": 5.003956665397617e-07,
+      "loss": 0.5229,
+      "step": 9110
+    },
+    {
+      "epoch": 0.5053751523883409,
+      "grad_norm": 1.463517427444458,
+      "learning_rate": 4.995164075874378e-07,
+      "loss": 0.5209,
+      "step": 9120
+    },
+    {
+      "epoch": 0.5059292918098194,
+      "grad_norm": 1.310459017753601,
+      "learning_rate": 4.986371501305678e-07,
+      "loss": 0.5227,
+      "step": 9130
+    },
+    {
+      "epoch": 0.5064834312312978,
+      "grad_norm": 1.347319483757019,
+      "learning_rate": 4.977578968881546e-07,
+      "loss": 0.52,
+      "step": 9140
+    },
+    {
+      "epoch": 0.5070375706527762,
+      "grad_norm": 1.3283882141113281,
+      "learning_rate": 4.968786505791881e-07,
+      "loss": 0.5173,
+      "step": 9150
+    },
+    {
+      "epoch": 0.5075917100742546,
+      "grad_norm": 1.1877269744873047,
+      "learning_rate": 4.959994139226364e-07,
+      "loss": 0.5236,
+      "step": 9160
+    },
+    {
+      "epoch": 0.5081458494957332,
+      "grad_norm": 1.1490060091018677,
+      "learning_rate": 4.95120189637438e-07,
+      "loss": 0.5246,
+      "step": 9170
+    },
+    {
+      "epoch": 0.5086999889172116,
+      "grad_norm": 1.2903001308441162,
+      "learning_rate": 4.942409804424935e-07,
+      "loss": 0.5126,
+      "step": 9180
+    },
+    {
+      "epoch": 0.50925412833869,
+      "grad_norm": 1.2388837337493896,
+      "learning_rate": 4.933617890566557e-07,
+      "loss": 0.514,
+      "step": 9190
+    },
+    {
+      "epoch": 0.5098082677601684,
+      "grad_norm": 1.268949031829834,
+      "learning_rate": 4.924826181987238e-07,
+      "loss": 0.5181,
+      "step": 9200
+    },
+    {
+      "epoch": 0.5098082677601684,
+      "eval_robovqa_loss": 0.0822962298989296,
+      "step": 9200
+    },
+    {
+      "epoch": 0.5098082677601684,
+      "eval_agibot_loss": 0.4542301297187805,
+      "step": 9200
+    },
+    {
+      "epoch": 0.5098082677601684,
+      "eval_holoassist_loss": 0.34197232127189636,
+      "step": 9200
+    },
+    {
+      "epoch": 0.5098082677601684,
+      "eval_robofail_loss": 0.20059822499752045,
+      "step": 9200
+    },
+    {
+      "epoch": 0.5103624071816469,
+      "grad_norm": 1.4202388525009155,
+      "learning_rate": 4.916034705874321e-07,
+      "loss": 0.5138,
+      "step": 9210
+    },
+    {
+      "epoch": 0.5109165466031254,
+      "grad_norm": 1.2928318977355957,
+      "learning_rate": 4.907243489414442e-07,
+      "loss": 0.5175,
+      "step": 9220
+    },
+    {
+      "epoch": 0.5114706860246038,
+      "grad_norm": 1.5077096223831177,
+      "learning_rate": 4.89845255979343e-07,
+      "loss": 0.5053,
+      "step": 9230
+    },
+    {
+      "epoch": 0.5120248254460822,
+      "grad_norm": 1.3263760805130005,
+      "learning_rate": 4.889661944196222e-07,
+      "loss": 0.514,
+      "step": 9240
+    },
+    {
+      "epoch": 0.5125789648675607,
+      "grad_norm": 1.1672049760818481,
+      "learning_rate": 4.880871669806789e-07,
+      "loss": 0.5201,
+      "step": 9250
+    },
+    {
+      "epoch": 0.5131331042890391,
+      "grad_norm": 1.2749038934707642,
+      "learning_rate": 4.872081763808048e-07,
+      "loss": 0.5156,
+      "step": 9260
+    },
+    {
+      "epoch": 0.5136872437105175,
+      "grad_norm": 1.351254940032959,
+      "learning_rate": 4.863292253381773e-07,
+      "loss": 0.5164,
+      "step": 9270
+    },
+    {
+      "epoch": 0.514241383131996,
+      "grad_norm": 1.2750805616378784,
+      "learning_rate": 4.854503165708516e-07,
+      "loss": 0.5125,
+      "step": 9280
+    },
+    {
+      "epoch": 0.5147955225534745,
+      "grad_norm": 1.2452220916748047,
+      "learning_rate": 4.845714527967523e-07,
+      "loss": 0.5159,
+      "step": 9290
+    },
+    {
+      "epoch": 0.5153496619749529,
+      "grad_norm": 1.2344504594802856,
+      "learning_rate": 4.836926367336645e-07,
+      "loss": 0.516,
+      "step": 9300
+    },
+    {
+      "epoch": 0.5153496619749529,
+      "eval_robovqa_loss": 0.08036039769649506,
+      "step": 9300
+    },
+    {
+      "epoch": 0.5153496619749529,
+      "eval_agibot_loss": 0.4660513401031494,
+      "step": 9300
+    },
+    {
+      "epoch": 0.5153496619749529,
+      "eval_holoassist_loss": 0.34042298793792725,
+      "step": 9300
+    },
+    {
+      "epoch": 0.5153496619749529,
+      "eval_robofail_loss": 0.20337824523448944,
+      "step": 9300
+    },
+    {
+      "epoch": 0.5159038013964313,
+      "grad_norm": 1.3226550817489624,
+      "learning_rate": 4.828138710992266e-07,
+      "loss": 0.5206,
+      "step": 9310
+    },
+    {
+      "epoch": 0.5164579408179097,
+      "grad_norm": 1.1979082822799683,
+      "learning_rate": 4.819351586109198e-07,
+      "loss": 0.5162,
+      "step": 9320
+    },
+    {
+      "epoch": 0.5170120802393883,
+      "grad_norm": 1.265073537826538,
+      "learning_rate": 4.810565019860621e-07,
+      "loss": 0.5158,
+      "step": 9330
+    },
+    {
+      "epoch": 0.5175662196608667,
+      "grad_norm": 1.1975135803222656,
+      "learning_rate": 4.801779039417982e-07,
+      "loss": 0.5268,
+      "step": 9340
+    },
+    {
+      "epoch": 0.5181203590823451,
+      "grad_norm": 1.2884241342544556,
+      "learning_rate": 4.792993671950915e-07,
+      "loss": 0.5089,
+      "step": 9350
+    },
+    {
+      "epoch": 0.5186744985038235,
+      "grad_norm": 1.2720446586608887,
+      "learning_rate": 4.784208944627166e-07,
+      "loss": 0.5187,
+      "step": 9360
+    },
+    {
+      "epoch": 0.519228637925302,
+      "grad_norm": 1.334676742553711,
+      "learning_rate": 4.775424884612491e-07,
+      "loss": 0.5239,
+      "step": 9370
+    },
+    {
+      "epoch": 0.5197827773467805,
+      "grad_norm": 1.2757624387741089,
+      "learning_rate": 4.7666415190705874e-07,
+      "loss": 0.5205,
+      "step": 9380
+    },
+    {
+      "epoch": 0.5203369167682589,
+      "grad_norm": 1.3002177476882935,
+      "learning_rate": 4.757858875163009e-07,
+      "loss": 0.5165,
+      "step": 9390
+    },
+    {
+      "epoch": 0.5208910561897373,
+      "grad_norm": 1.2371611595153809,
+      "learning_rate": 4.7490769800490704e-07,
+      "loss": 0.5207,
+      "step": 9400
+    },
+    {
+      "epoch": 0.5208910561897373,
+      "eval_robovqa_loss": 0.08086331933736801,
+      "step": 9400
+    },
+    {
+      "epoch": 0.5208910561897373,
+      "eval_agibot_loss": 0.4594055116176605,
+      "step": 9400
+    },
+    {
+      "epoch": 0.5208910561897373,
+      "eval_holoassist_loss": 0.3427622318267822,
+      "step": 9400
+    },
+    {
+      "epoch": 0.5208910561897373,
+      "eval_robofail_loss": 0.20379258692264557,
+      "step": 9400
+    },
+    {
+      "epoch": 0.5214451956112158,
+      "grad_norm": 1.271382212638855,
+      "learning_rate": 4.7402958608857783e-07,
+      "loss": 0.5188,
+      "step": 9410
+    },
+    {
+      "epoch": 0.5219993350326942,
+      "grad_norm": 1.1211212873458862,
+      "learning_rate": 4.7315155448277325e-07,
+      "loss": 0.5212,
+      "step": 9420
+    },
+    {
+      "epoch": 0.5225534744541727,
+      "grad_norm": 1.2261103391647339,
+      "learning_rate": 4.722736059027054e-07,
+      "loss": 0.5206,
+      "step": 9430
+    },
+    {
+      "epoch": 0.5231076138756511,
+      "grad_norm": 1.1180691719055176,
+      "learning_rate": 4.7139574306332976e-07,
+      "loss": 0.526,
+      "step": 9440
+    },
+    {
+      "epoch": 0.5236617532971296,
+      "grad_norm": 1.2672573328018188,
+      "learning_rate": 4.7051796867933604e-07,
+      "loss": 0.5161,
+      "step": 9450
+    },
+    {
+      "epoch": 0.524215892718608,
+      "grad_norm": 1.2555286884307861,
+      "learning_rate": 4.69640285465141e-07,
+      "loss": 0.5182,
+      "step": 9460
+    },
+    {
+      "epoch": 0.5247700321400864,
+      "grad_norm": 1.3438464403152466,
+      "learning_rate": 4.687626961348792e-07,
+      "loss": 0.5124,
+      "step": 9470
+    },
+    {
+      "epoch": 0.5253241715615649,
+      "grad_norm": 1.2075613737106323,
+      "learning_rate": 4.678852034023951e-07,
+      "loss": 0.5163,
+      "step": 9480
+    },
+    {
+      "epoch": 0.5258783109830434,
+      "grad_norm": 1.2002960443496704,
+      "learning_rate": 4.6700780998123414e-07,
+      "loss": 0.5291,
+      "step": 9490
+    },
+    {
+      "epoch": 0.5264324504045218,
+      "grad_norm": 1.1737651824951172,
+      "learning_rate": 4.661305185846346e-07,
+      "loss": 0.5197,
+      "step": 9500
+    },
+    {
+      "epoch": 0.5264324504045218,
+      "eval_robovqa_loss": 0.0842987671494484,
+      "step": 9500
+    },
+    {
+      "epoch": 0.5264324504045218,
+      "eval_agibot_loss": 0.4500093162059784,
+      "step": 9500
+    },
+    {
+      "epoch": 0.5264324504045218,
+      "eval_holoassist_loss": 0.3461896479129791,
+      "step": 9500
+    },
+    {
+      "epoch": 0.5264324504045218,
+      "eval_robofail_loss": 0.20294909179210663,
+      "step": 9500
+    },
+    {
+      "epoch": 0.5269865898260002,
+      "grad_norm": 1.2431118488311768,
+      "learning_rate": 4.6525333192551963e-07,
+      "loss": 0.5157,
+      "step": 9510
+    },
+    {
+      "epoch": 0.5275407292474786,
+      "grad_norm": 1.2925978899002075,
+      "learning_rate": 4.6437625271648854e-07,
+      "loss": 0.5112,
+      "step": 9520
+    },
+    {
+      "epoch": 0.5280948686689572,
+      "grad_norm": 1.2391204833984375,
+      "learning_rate": 4.6349928366980775e-07,
+      "loss": 0.5193,
+      "step": 9530
+    },
+    {
+      "epoch": 0.5286490080904356,
+      "grad_norm": 1.2042982578277588,
+      "learning_rate": 4.626224274974035e-07,
+      "loss": 0.5184,
+      "step": 9540
+    },
+    {
+      "epoch": 0.529203147511914,
+      "grad_norm": 1.3827855587005615,
+      "learning_rate": 4.617456869108532e-07,
+      "loss": 0.5195,
+      "step": 9550
+    },
+    {
+      "epoch": 0.5297572869333924,
+      "grad_norm": 1.2540674209594727,
+      "learning_rate": 4.60869064621376e-07,
+      "loss": 0.5166,
+      "step": 9560
+    },
+    {
+      "epoch": 0.5303114263548709,
+      "grad_norm": 1.3960951566696167,
+      "learning_rate": 4.5999256333982624e-07,
+      "loss": 0.5116,
+      "step": 9570
+    },
+    {
+      "epoch": 0.5308655657763494,
+      "grad_norm": 1.33521568775177,
+      "learning_rate": 4.5911618577668313e-07,
+      "loss": 0.5111,
+      "step": 9580
+    },
+    {
+      "epoch": 0.5314197051978278,
+      "grad_norm": 1.2253435850143433,
+      "learning_rate": 4.5823993464204386e-07,
+      "loss": 0.5124,
+      "step": 9590
+    },
+    {
+      "epoch": 0.5319738446193062,
+      "grad_norm": 1.2996859550476074,
+      "learning_rate": 4.573638126456147e-07,
+      "loss": 0.5161,
+      "step": 9600
+    },
+    {
+      "epoch": 0.5319738446193062,
+      "eval_robovqa_loss": 0.08296090364456177,
+      "step": 9600
+    },
+    {
+      "epoch": 0.5319738446193062,
+      "eval_agibot_loss": 0.44863563776016235,
+      "step": 9600
+    },
+    {
+      "epoch": 0.5319738446193062,
+      "eval_holoassist_loss": 0.33110836148262024,
+      "step": 9600
+    },
+    {
+      "epoch": 0.5319738446193062,
+      "eval_robofail_loss": 0.20604386925697327,
+      "step": 9600
+    },
+    {
+      "epoch": 0.5325279840407846,
+      "grad_norm": 1.2988568544387817,
+      "learning_rate": 4.56487822496702e-07,
+      "loss": 0.5146,
+      "step": 9610
+    },
+    {
+      "epoch": 0.5330821234622631,
+      "grad_norm": 1.2217482328414917,
+      "learning_rate": 4.5561196690420516e-07,
+      "loss": 0.5112,
+      "step": 9620
+    },
+    {
+      "epoch": 0.5336362628837416,
+      "grad_norm": 1.3185545206069946,
+      "learning_rate": 4.5473624857660667e-07,
+      "loss": 0.5049,
+      "step": 9630
+    },
+    {
+      "epoch": 0.53419040230522,
+      "grad_norm": 1.262885332107544,
+      "learning_rate": 4.5386067022196526e-07,
+      "loss": 0.5186,
+      "step": 9640
+    },
+    {
+      "epoch": 0.5347445417266984,
+      "grad_norm": 1.2184704542160034,
+      "learning_rate": 4.529852345479067e-07,
+      "loss": 0.5165,
+      "step": 9650
+    },
+    {
+      "epoch": 0.5352986811481769,
+      "grad_norm": 1.1610207557678223,
+      "learning_rate": 4.521099442616149e-07,
+      "loss": 0.5166,
+      "step": 9660
+    },
+    {
+      "epoch": 0.5358528205696553,
+      "grad_norm": 1.478848934173584,
+      "learning_rate": 4.5123480206982495e-07,
+      "loss": 0.5159,
+      "step": 9670
+    },
+    {
+      "epoch": 0.5364069599911337,
+      "grad_norm": 1.2694194316864014,
+      "learning_rate": 4.503598106788135e-07,
+      "loss": 0.5183,
+      "step": 9680
+    },
+    {
+      "epoch": 0.5369610994126122,
+      "grad_norm": 1.4139130115509033,
+      "learning_rate": 4.4948497279439113e-07,
+      "loss": 0.5165,
+      "step": 9690
+    },
+    {
+      "epoch": 0.5375152388340907,
+      "grad_norm": 1.2383179664611816,
+      "learning_rate": 4.486102911218936e-07,
+      "loss": 0.5151,
+      "step": 9700
+    },
+    {
+      "epoch": 0.5375152388340907,
+      "eval_robovqa_loss": 0.08289963006973267,
+      "step": 9700
+    },
+    {
+      "epoch": 0.5375152388340907,
+      "eval_agibot_loss": 0.4551071226596832,
+      "step": 9700
+    },
+    {
+      "epoch": 0.5375152388340907,
+      "eval_holoassist_loss": 0.3371936082839966,
+      "step": 9700
+    },
+    {
+      "epoch": 0.5375152388340907,
+      "eval_robofail_loss": 0.20261794328689575,
+      "step": 9700
+    },
+    {
+      "epoch": 0.5380693782555691,
+      "grad_norm": 1.1938608884811401,
+      "learning_rate": 4.477357683661733e-07,
+      "loss": 0.5121,
+      "step": 9710
+    },
+    {
+      "epoch": 0.5386235176770475,
+      "grad_norm": 1.2525290250778198,
+      "learning_rate": 4.4686140723159177e-07,
+      "loss": 0.5171,
+      "step": 9720
+    },
+    {
+      "epoch": 0.5391776570985259,
+      "grad_norm": 1.274208664894104,
+      "learning_rate": 4.4598721042201064e-07,
+      "loss": 0.5229,
+      "step": 9730
+    },
+    {
+      "epoch": 0.5397317965200045,
+      "grad_norm": 1.3121837377548218,
+      "learning_rate": 4.4511318064078285e-07,
+      "loss": 0.5172,
+      "step": 9740
+    },
+    {
+      "epoch": 0.5402859359414829,
+      "grad_norm": 1.3660664558410645,
+      "learning_rate": 4.4423932059074543e-07,
+      "loss": 0.5125,
+      "step": 9750
+    },
+    {
+      "epoch": 0.5408400753629613,
+      "grad_norm": 1.3352311849594116,
+      "learning_rate": 4.4336563297421035e-07,
+      "loss": 0.5137,
+      "step": 9760
+    },
+    {
+      "epoch": 0.5413942147844397,
+      "grad_norm": 1.2712814807891846,
+      "learning_rate": 4.424921204929561e-07,
+      "loss": 0.5139,
+      "step": 9770
+    },
+    {
+      "epoch": 0.5419483542059182,
+      "grad_norm": 1.3154189586639404,
+      "learning_rate": 4.416187858482201e-07,
+      "loss": 0.5215,
+      "step": 9780
+    },
+    {
+      "epoch": 0.5425024936273967,
+      "grad_norm": 1.3212343454360962,
+      "learning_rate": 4.407456317406892e-07,
+      "loss": 0.5189,
+      "step": 9790
+    },
+    {
+      "epoch": 0.5430566330488751,
+      "grad_norm": 1.2240793704986572,
+      "learning_rate": 4.398726608704923e-07,
+      "loss": 0.5149,
+      "step": 9800
+    },
+    {
+      "epoch": 0.5430566330488751,
+      "eval_robovqa_loss": 0.08315795660018921,
+      "step": 9800
+    },
+    {
+      "epoch": 0.5430566330488751,
+      "eval_agibot_loss": 0.4597175717353821,
+      "step": 9800
+    },
+    {
+      "epoch": 0.5430566330488751,
+      "eval_holoassist_loss": 0.3351593017578125,
+      "step": 9800
+    },
+    {
+      "epoch": 0.5430566330488751,
+      "eval_robofail_loss": 0.2045988142490387,
+      "step": 9800
+    },
+    {
+      "epoch": 0.5436107724703535,
+      "grad_norm": 1.2900890111923218,
+      "learning_rate": 4.3899987593719195e-07,
+      "loss": 0.5204,
+      "step": 9810
+    },
+    {
+      "epoch": 0.544164911891832,
+      "grad_norm": 1.2192814350128174,
+      "learning_rate": 4.38127279639775e-07,
+      "loss": 0.5202,
+      "step": 9820
+    },
+    {
+      "epoch": 0.5447190513133104,
+      "grad_norm": 1.6530661582946777,
+      "learning_rate": 4.3725487467664547e-07,
+      "loss": 0.509,
+      "step": 9830
+    },
+    {
+      "epoch": 0.5452731907347889,
+      "grad_norm": 1.2759709358215332,
+      "learning_rate": 4.3638266374561596e-07,
+      "loss": 0.5213,
+      "step": 9840
+    },
+    {
+      "epoch": 0.5458273301562673,
+      "grad_norm": 1.2029930353164673,
+      "learning_rate": 4.3551064954389816e-07,
+      "loss": 0.5088,
+      "step": 9850
+    },
+    {
+      "epoch": 0.5463814695777458,
+      "grad_norm": 1.260711908340454,
+      "learning_rate": 4.346388347680964e-07,
+      "loss": 0.5139,
+      "step": 9860
+    },
+    {
+      "epoch": 0.5469356089992242,
+      "grad_norm": 1.257436990737915,
+      "learning_rate": 4.337672221141974e-07,
+      "loss": 0.5168,
+      "step": 9870
+    },
+    {
+      "epoch": 0.5474897484207026,
+      "grad_norm": 1.3473694324493408,
+      "learning_rate": 4.3289581427756346e-07,
+      "loss": 0.5033,
+      "step": 9880
+    },
+    {
+      "epoch": 0.548043887842181,
+      "grad_norm": 1.2202413082122803,
+      "learning_rate": 4.3202461395292337e-07,
+      "loss": 0.5156,
+      "step": 9890
+    },
+    {
+      "epoch": 0.5485980272636596,
+      "grad_norm": 1.2304188013076782,
+      "learning_rate": 4.3115362383436415e-07,
+      "loss": 0.5186,
+      "step": 9900
+    },
+    {
+      "epoch": 0.5485980272636596,
+      "eval_robovqa_loss": 0.08176163583993912,
+      "step": 9900
+    },
+    {
+      "epoch": 0.5485980272636596,
+      "eval_agibot_loss": 0.461038202047348,
+      "step": 9900
+    },
+    {
+      "epoch": 0.5485980272636596,
+      "eval_holoassist_loss": 0.3447757065296173,
+      "step": 9900
+    },
+    {
+      "epoch": 0.5485980272636596,
+      "eval_robofail_loss": 0.20287621021270752,
+      "step": 9900
+    },
+    {
+      "epoch": 0.549152166685138,
+      "grad_norm": 1.428905725479126,
+      "learning_rate": 4.302828466153227e-07,
+      "loss": 0.5123,
+      "step": 9910
+    },
+    {
+      "epoch": 0.5497063061066164,
+      "grad_norm": 1.5408859252929688,
+      "learning_rate": 4.294122849885777e-07,
+      "loss": 0.5097,
+      "step": 9920
+    },
+    {
+      "epoch": 0.5502604455280948,
+      "grad_norm": 1.272096872329712,
+      "learning_rate": 4.2854194164624103e-07,
+      "loss": 0.5253,
+      "step": 9930
+    },
+    {
+      "epoch": 0.5508145849495734,
+      "grad_norm": 1.6935001611709595,
+      "learning_rate": 4.276718192797498e-07,
+      "loss": 0.5091,
+      "step": 9940
+    },
+    {
+      "epoch": 0.5513687243710518,
+      "grad_norm": 1.2529058456420898,
+      "learning_rate": 4.268019205798572e-07,
+      "loss": 0.5169,
+      "step": 9950
+    },
+    {
+      "epoch": 0.5519228637925302,
+      "grad_norm": 1.2424004077911377,
+      "learning_rate": 4.2593224823662533e-07,
+      "loss": 0.5097,
+      "step": 9960
+    },
+    {
+      "epoch": 0.5524770032140086,
+      "grad_norm": 1.2132890224456787,
+      "learning_rate": 4.2506280493941637e-07,
+      "loss": 0.5162,
+      "step": 9970
+    },
+    {
+      "epoch": 0.5530311426354871,
+      "grad_norm": 1.5297126770019531,
+      "learning_rate": 4.2419359337688355e-07,
+      "loss": 0.5028,
+      "step": 9980
+    },
+    {
+      "epoch": 0.5535852820569656,
+      "grad_norm": 1.314738392829895,
+      "learning_rate": 4.2332461623696416e-07,
+      "loss": 0.5158,
+      "step": 9990
+    },
+    {
+      "epoch": 0.554139421478444,
+      "grad_norm": 1.302743911743164,
+      "learning_rate": 4.224558762068701e-07,
+      "loss": 0.5155,
+      "step": 10000
+    },
+    {
+      "epoch": 0.554139421478444,
+      "eval_robovqa_loss": 0.07916717231273651,
+      "step": 10000
+    },
+    {
+      "epoch": 0.554139421478444,
+      "eval_agibot_loss": 0.4603254199028015,
+      "step": 10000
+    },
+    {
+      "epoch": 0.554139421478444,
+      "eval_holoassist_loss": 0.33338749408721924,
+      "step": 10000
+    },
+    {
+      "epoch": 0.554139421478444,
+      "eval_robofail_loss": 0.20513920485973358,
+      "step": 10000
+    },
+    {
+      "epoch": 0.5546935608999224,
+      "grad_norm": 1.2900762557983398,
+      "learning_rate": 4.2158737597308014e-07,
+      "loss": 0.5143,
+      "step": 10010
+    },
+    {
+      "epoch": 0.5552477003214009,
+      "grad_norm": 1.2788184881210327,
+      "learning_rate": 4.20719118221332e-07,
+      "loss": 0.5107,
+      "step": 10020
+    },
+    {
+      "epoch": 0.5558018397428793,
+      "grad_norm": 1.1807020902633667,
+      "learning_rate": 4.1985110563661263e-07,
+      "loss": 0.5197,
+      "step": 10030
+    },
+    {
+      "epoch": 0.5563559791643578,
+      "grad_norm": 1.1640006303787231,
+      "learning_rate": 4.1898334090315164e-07,
+      "loss": 0.5149,
+      "step": 10040
+    },
+    {
+      "epoch": 0.5569101185858362,
+      "grad_norm": 1.3342106342315674,
+      "learning_rate": 4.1811582670441196e-07,
+      "loss": 0.5086,
+      "step": 10050
+    },
+    {
+      "epoch": 0.5574642580073147,
+      "grad_norm": 1.245856523513794,
+      "learning_rate": 4.172485657230814e-07,
+      "loss": 0.5108,
+      "step": 10060
+    },
+    {
+      "epoch": 0.5580183974287931,
+      "grad_norm": 1.2804210186004639,
+      "learning_rate": 4.1638156064106537e-07,
+      "loss": 0.5147,
+      "step": 10070
+    },
+    {
+      "epoch": 0.5585725368502715,
+      "grad_norm": 1.3750206232070923,
+      "learning_rate": 4.1551481413947725e-07,
+      "loss": 0.5181,
+      "step": 10080
+    },
+    {
+      "epoch": 0.55912667627175,
+      "grad_norm": 1.441153883934021,
+      "learning_rate": 4.146483288986314e-07,
+      "loss": 0.5124,
+      "step": 10090
+    },
+    {
+      "epoch": 0.5596808156932284,
+      "grad_norm": 1.3643174171447754,
+      "learning_rate": 4.1378210759803386e-07,
+      "loss": 0.5169,
+      "step": 10100
+    },
+    {
+      "epoch": 0.5596808156932284,
+      "eval_robovqa_loss": 0.08305657655000687,
+      "step": 10100
+    },
+    {
+      "epoch": 0.5596808156932284,
+      "eval_agibot_loss": 0.45587924122810364,
+      "step": 10100
+    },
+    {
+      "epoch": 0.5596808156932284,
+      "eval_holoassist_loss": 0.33771294355392456,
+      "step": 10100
+    },
+    {
+      "epoch": 0.5596808156932284,
+      "eval_robofail_loss": 0.20248712599277496,
+      "step": 10100
+    },
+    {
+      "epoch": 0.5602349551147069,
+      "grad_norm": 1.2820388078689575,
+      "learning_rate": 4.129161529163746e-07,
+      "loss": 0.5046,
+      "step": 10110
+    },
+    {
+      "epoch": 0.5607890945361853,
+      "grad_norm": 1.3095673322677612,
+      "learning_rate": 4.120504675315192e-07,
+      "loss": 0.5072,
+      "step": 10120
+    },
+    {
+      "epoch": 0.5613432339576637,
+      "grad_norm": 1.1812983751296997,
+      "learning_rate": 4.1118505412050025e-07,
+      "loss": 0.512,
+      "step": 10130
+    },
+    {
+      "epoch": 0.5618973733791421,
+      "grad_norm": 1.297916054725647,
+      "learning_rate": 4.1031991535950936e-07,
+      "loss": 0.5221,
+      "step": 10140
+    },
+    {
+      "epoch": 0.5624515128006207,
+      "grad_norm": 1.400581955909729,
+      "learning_rate": 4.0945505392388924e-07,
+      "loss": 0.5071,
+      "step": 10150
+    },
+    {
+      "epoch": 0.5630056522220991,
+      "grad_norm": 1.2638837099075317,
+      "learning_rate": 4.0859047248812407e-07,
+      "loss": 0.51,
+      "step": 10160
+    },
+    {
+      "epoch": 0.5635597916435775,
+      "grad_norm": 1.2735354900360107,
+      "learning_rate": 4.0772617372583307e-07,
+      "loss": 0.5141,
+      "step": 10170
+    },
+    {
+      "epoch": 0.5641139310650559,
+      "grad_norm": 1.282848596572876,
+      "learning_rate": 4.068621603097611e-07,
+      "loss": 0.5231,
+      "step": 10180
+    },
+    {
+      "epoch": 0.5646680704865344,
+      "grad_norm": 1.3757786750793457,
+      "learning_rate": 4.059984349117702e-07,
+      "loss": 0.5178,
+      "step": 10190
+    },
+    {
+      "epoch": 0.5652222099080129,
+      "grad_norm": 1.2608166933059692,
+      "learning_rate": 4.0513500020283226e-07,
+      "loss": 0.51,
+      "step": 10200
+    },
+    {
+      "epoch": 0.5652222099080129,
+      "eval_robovqa_loss": 0.08139218389987946,
+      "step": 10200
+    },
+    {
+      "epoch": 0.5652222099080129,
+      "eval_agibot_loss": 0.4587564468383789,
+      "step": 10200
+    },
+    {
+      "epoch": 0.5652222099080129,
+      "eval_holoassist_loss": 0.3467942774295807,
+      "step": 10200
+    },
+    {
+      "epoch": 0.5652222099080129,
+      "eval_robofail_loss": 0.2060483992099762,
+      "step": 10200
+    },
+    {
+      "epoch": 0.5657763493294913,
+      "grad_norm": 1.2833223342895508,
+      "learning_rate": 4.042718588530199e-07,
+      "loss": 0.5119,
+      "step": 10210
+    },
+    {
+      "epoch": 0.5663304887509697,
+      "grad_norm": 1.195594310760498,
+      "learning_rate": 4.034090135314986e-07,
+      "loss": 0.5087,
+      "step": 10220
+    },
+    {
+      "epoch": 0.5668846281724482,
+      "grad_norm": 1.2788141965866089,
+      "learning_rate": 4.0254646690651897e-07,
+      "loss": 0.5089,
+      "step": 10230
+    },
+    {
+      "epoch": 0.5674387675939266,
+      "grad_norm": 1.4006081819534302,
+      "learning_rate": 4.016842216454068e-07,
+      "loss": 0.5108,
+      "step": 10240
+    },
+    {
+      "epoch": 0.5679929070154051,
+      "grad_norm": 1.276726245880127,
+      "learning_rate": 4.00822280414557e-07,
+      "loss": 0.5085,
+      "step": 10250
+    },
+    {
+      "epoch": 0.5685470464368835,
+      "grad_norm": 1.28522527217865,
+      "learning_rate": 3.999606458794241e-07,
+      "loss": 0.5099,
+      "step": 10260
+    },
+    {
+      "epoch": 0.569101185858362,
+      "grad_norm": 1.2189635038375854,
+      "learning_rate": 3.9909932070451355e-07,
+      "loss": 0.5271,
+      "step": 10270
+    },
+    {
+      "epoch": 0.5696553252798404,
+      "grad_norm": 1.2800782918930054,
+      "learning_rate": 3.9823830755337515e-07,
+      "loss": 0.5168,
+      "step": 10280
+    },
+    {
+      "epoch": 0.5702094647013188,
+      "grad_norm": 1.2787199020385742,
+      "learning_rate": 3.9737760908859264e-07,
+      "loss": 0.5274,
+      "step": 10290
+    },
+    {
+      "epoch": 0.5707636041227973,
+      "grad_norm": 1.3957058191299438,
+      "learning_rate": 3.965172279717777e-07,
+      "loss": 0.511,
+      "step": 10300
+    },
+    {
+      "epoch": 0.5707636041227973,
+      "eval_robovqa_loss": 0.08217086642980576,
+      "step": 10300
+    },
+    {
+      "epoch": 0.5707636041227973,
+      "eval_agibot_loss": 0.4622708559036255,
+      "step": 10300
+    },
+    {
+      "epoch": 0.5707636041227973,
+      "eval_holoassist_loss": 0.3426809310913086,
+      "step": 10300
+    },
+    {
+      "epoch": 0.5707636041227973,
+      "eval_robofail_loss": 0.20447364449501038,
+      "step": 10300
+    },
+    {
+      "epoch": 0.5713177435442758,
+      "grad_norm": 1.1915863752365112,
+      "learning_rate": 3.956571668635601e-07,
+      "loss": 0.5101,
+      "step": 10310
+    },
+    {
+      "epoch": 0.5718718829657542,
+      "grad_norm": 1.3483299016952515,
+      "learning_rate": 3.947974284235801e-07,
+      "loss": 0.5069,
+      "step": 10320
+    },
+    {
+      "epoch": 0.5724260223872326,
+      "grad_norm": 1.1761835813522339,
+      "learning_rate": 3.939380153104801e-07,
+      "loss": 0.5074,
+      "step": 10330
+    },
+    {
+      "epoch": 0.572980161808711,
+      "grad_norm": 1.3815217018127441,
+      "learning_rate": 3.9307893018189673e-07,
+      "loss": 0.5111,
+      "step": 10340
+    },
+    {
+      "epoch": 0.5735343012301896,
+      "grad_norm": 1.3156071901321411,
+      "learning_rate": 3.922201756944519e-07,
+      "loss": 0.5067,
+      "step": 10350
+    },
+    {
+      "epoch": 0.574088440651668,
+      "grad_norm": 1.2532910108566284,
+      "learning_rate": 3.9136175450374557e-07,
+      "loss": 0.5059,
+      "step": 10360
+    },
+    {
+      "epoch": 0.5746425800731464,
+      "grad_norm": 1.3710987567901611,
+      "learning_rate": 3.905036692643465e-07,
+      "loss": 0.51,
+      "step": 10370
+    },
+    {
+      "epoch": 0.5751967194946248,
+      "grad_norm": 1.2168991565704346,
+      "learning_rate": 3.8964592262978496e-07,
+      "loss": 0.5103,
+      "step": 10380
+    },
+    {
+      "epoch": 0.5757508589161033,
+      "grad_norm": 1.3010125160217285,
+      "learning_rate": 3.887885172525442e-07,
+      "loss": 0.5097,
+      "step": 10390
+    },
+    {
+      "epoch": 0.5763049983375818,
+      "grad_norm": 1.3021339178085327,
+      "learning_rate": 3.879314557840515e-07,
+      "loss": 0.5091,
+      "step": 10400
+    },
+    {
+      "epoch": 0.5763049983375818,
+      "eval_robovqa_loss": 0.08287476748228073,
+      "step": 10400
+    },
+    {
+      "epoch": 0.5763049983375818,
+      "eval_agibot_loss": 0.4622899889945984,
+      "step": 10400
+    },
+    {
+      "epoch": 0.5763049983375818,
+      "eval_holoassist_loss": 0.33248263597488403,
+      "step": 10400
+    },
+    {
+      "epoch": 0.5763049983375818,
+      "eval_robofail_loss": 0.20454718172550201,
+      "step": 10400
+    },
+    {
+      "epoch": 0.5768591377590602,
+      "grad_norm": 1.2836110591888428,
+      "learning_rate": 3.870747408746716e-07,
+      "loss": 0.5026,
+      "step": 10410
+    },
+    {
+      "epoch": 0.5774132771805386,
+      "grad_norm": 1.1346837282180786,
+      "learning_rate": 3.8621837517369666e-07,
+      "loss": 0.5173,
+      "step": 10420
+    },
+    {
+      "epoch": 0.5779674166020171,
+      "grad_norm": 1.4633063077926636,
+      "learning_rate": 3.8536236132933937e-07,
+      "loss": 0.5165,
+      "step": 10430
+    },
+    {
+      "epoch": 0.5785215560234955,
+      "grad_norm": 1.256270408630371,
+      "learning_rate": 3.8450670198872466e-07,
+      "loss": 0.5144,
+      "step": 10440
+    },
+    {
+      "epoch": 0.579075695444974,
+      "grad_norm": 1.38346529006958,
+      "learning_rate": 3.8365139979788033e-07,
+      "loss": 0.5044,
+      "step": 10450
+    },
+    {
+      "epoch": 0.5796298348664524,
+      "grad_norm": 1.2874271869659424,
+      "learning_rate": 3.8279645740173055e-07,
+      "loss": 0.5108,
+      "step": 10460
+    },
+    {
+      "epoch": 0.5801839742879309,
+      "grad_norm": 1.3414663076400757,
+      "learning_rate": 3.819418774440868e-07,
+      "loss": 0.5087,
+      "step": 10470
+    },
+    {
+      "epoch": 0.5807381137094093,
+      "grad_norm": 1.284894347190857,
+      "learning_rate": 3.8108766256763905e-07,
+      "loss": 0.5147,
+      "step": 10480
+    },
+    {
+      "epoch": 0.5812922531308877,
+      "grad_norm": 1.32367742061615,
+      "learning_rate": 3.8023381541394916e-07,
+      "loss": 0.5157,
+      "step": 10490
+    },
+    {
+      "epoch": 0.5818463925523661,
+      "grad_norm": 1.3269833326339722,
+      "learning_rate": 3.7938033862344114e-07,
+      "loss": 0.5119,
+      "step": 10500
+    },
+    {
+      "epoch": 0.5818463925523661,
+      "eval_robovqa_loss": 0.07955945283174515,
+      "step": 10500
+    },
+    {
+      "epoch": 0.5818463925523661,
+      "eval_agibot_loss": 0.4592565894126892,
+      "step": 10500
+    },
+    {
+      "epoch": 0.5818463925523661,
+      "eval_holoassist_loss": 0.33604174852371216,
+      "step": 10500
+    },
+    {
+      "epoch": 0.5818463925523661,
+      "eval_robofail_loss": 0.20552362501621246,
+      "step": 10500
+    },
+    {
+      "epoch": 0.5824005319738447,
+      "grad_norm": 1.2892416715621948,
+      "learning_rate": 3.7852723483539436e-07,
+      "loss": 0.5075,
+      "step": 10510
+    },
+    {
+      "epoch": 0.5829546713953231,
+      "grad_norm": 1.258042812347412,
+      "learning_rate": 3.7767450668793417e-07,
+      "loss": 0.509,
+      "step": 10520
+    },
+    {
+      "epoch": 0.5835088108168015,
+      "grad_norm": 1.2057397365570068,
+      "learning_rate": 3.768221568180246e-07,
+      "loss": 0.5125,
+      "step": 10530
+    },
+    {
+      "epoch": 0.5840629502382799,
+      "grad_norm": 1.2721567153930664,
+      "learning_rate": 3.759701878614596e-07,
+      "loss": 0.5085,
+      "step": 10540
+    },
+    {
+      "epoch": 0.5846170896597584,
+      "grad_norm": 1.2213821411132812,
+      "learning_rate": 3.7511860245285595e-07,
+      "loss": 0.5107,
+      "step": 10550
+    },
+    {
+      "epoch": 0.5851712290812369,
+      "grad_norm": 1.2931671142578125,
+      "learning_rate": 3.742674032256431e-07,
+      "loss": 0.5097,
+      "step": 10560
+    },
+    {
+      "epoch": 0.5857253685027153,
+      "grad_norm": 1.31634521484375,
+      "learning_rate": 3.7341659281205757e-07,
+      "loss": 0.5141,
+      "step": 10570
+    },
+    {
+      "epoch": 0.5862795079241937,
+      "grad_norm": 1.2987046241760254,
+      "learning_rate": 3.7256617384313246e-07,
+      "loss": 0.5145,
+      "step": 10580
+    },
+    {
+      "epoch": 0.5868336473456721,
+      "grad_norm": 1.2949585914611816,
+      "learning_rate": 3.7171614894869103e-07,
+      "loss": 0.5138,
+      "step": 10590
+    },
+    {
+      "epoch": 0.5873877867671506,
+      "grad_norm": 1.2861192226409912,
+      "learning_rate": 3.7086652075733794e-07,
+      "loss": 0.5037,
+      "step": 10600
+    },
+    {
+      "epoch": 0.5873877867671506,
+      "eval_robovqa_loss": 0.08052704483270645,
+      "step": 10600
+    },
+    {
+      "epoch": 0.5873877867671506,
+      "eval_agibot_loss": 0.4665200114250183,
+      "step": 10600
+    },
+    {
+      "epoch": 0.5873877867671506,
+      "eval_holoassist_loss": 0.34057754278182983,
+      "step": 10600
+    },
+    {
+      "epoch": 0.5873877867671506,
+      "eval_robofail_loss": 0.20628111064434052,
+      "step": 10600
+    },
+    {
+      "epoch": 0.5879419261886291,
+      "grad_norm": 1.2719950675964355,
+      "learning_rate": 3.700172918964505e-07,
+      "loss": 0.5061,
+      "step": 10610
+    },
+    {
+      "epoch": 0.5884960656101075,
+      "grad_norm": 1.358014464378357,
+      "learning_rate": 3.691684649921719e-07,
+      "loss": 0.5149,
+      "step": 10620
+    },
+    {
+      "epoch": 0.5890502050315859,
+      "grad_norm": 1.2660759687423706,
+      "learning_rate": 3.683200426694015e-07,
+      "loss": 0.5085,
+      "step": 10630
+    },
+    {
+      "epoch": 0.5896043444530644,
+      "grad_norm": 1.2590179443359375,
+      "learning_rate": 3.674720275517884e-07,
+      "loss": 0.51,
+      "step": 10640
+    },
+    {
+      "epoch": 0.5901584838745428,
+      "grad_norm": 1.2114074230194092,
+      "learning_rate": 3.666244222617218e-07,
+      "loss": 0.5101,
+      "step": 10650
+    },
+    {
+      "epoch": 0.5907126232960213,
+      "grad_norm": 1.1873806715011597,
+      "learning_rate": 3.657772294203236e-07,
+      "loss": 0.5113,
+      "step": 10660
+    },
+    {
+      "epoch": 0.5912667627174997,
+      "grad_norm": 1.2068397998809814,
+      "learning_rate": 3.6493045164744077e-07,
+      "loss": 0.4996,
+      "step": 10670
+    },
+    {
+      "epoch": 0.5918209021389782,
+      "grad_norm": 1.2771177291870117,
+      "learning_rate": 3.640840915616365e-07,
+      "loss": 0.5101,
+      "step": 10680
+    },
+    {
+      "epoch": 0.5923750415604566,
+      "grad_norm": 1.2414298057556152,
+      "learning_rate": 3.632381517801818e-07,
+      "loss": 0.5029,
+      "step": 10690
+    },
+    {
+      "epoch": 0.592929180981935,
+      "grad_norm": 1.2731387615203857,
+      "learning_rate": 3.6239263491904883e-07,
+      "loss": 0.5141,
+      "step": 10700
+    },
+    {
+      "epoch": 0.592929180981935,
+      "eval_robovqa_loss": 0.07989407330751419,
+      "step": 10700
+    },
+    {
+      "epoch": 0.592929180981935,
+      "eval_agibot_loss": 0.4614154100418091,
+      "step": 10700
+    },
+    {
+      "epoch": 0.592929180981935,
+      "eval_holoassist_loss": 0.3357793092727661,
+      "step": 10700
+    },
+    {
+      "epoch": 0.592929180981935,
+      "eval_robofail_loss": 0.2075977474451065,
+      "step": 10700
+    },
+    {
+      "epoch": 0.5934833204034135,
+      "grad_norm": 1.323078989982605,
+      "learning_rate": 3.615475435929009e-07,
+      "loss": 0.5142,
+      "step": 10710
+    },
+    {
+      "epoch": 0.594037459824892,
+      "grad_norm": 1.2474323511123657,
+      "learning_rate": 3.6070288041508644e-07,
+      "loss": 0.5087,
+      "step": 10720
+    },
+    {
+      "epoch": 0.5945915992463704,
+      "grad_norm": 1.3588931560516357,
+      "learning_rate": 3.5985864799762933e-07,
+      "loss": 0.5046,
+      "step": 10730
+    },
+    {
+      "epoch": 0.5951457386678488,
+      "grad_norm": 1.51687753200531,
+      "learning_rate": 3.590148489512213e-07,
+      "loss": 0.5075,
+      "step": 10740
+    },
+    {
+      "epoch": 0.5956998780893272,
+      "grad_norm": 1.3568520545959473,
+      "learning_rate": 3.5817148588521393e-07,
+      "loss": 0.5106,
+      "step": 10750
+    },
+    {
+      "epoch": 0.5962540175108058,
+      "grad_norm": 1.2613948583602905,
+      "learning_rate": 3.573285614076112e-07,
+      "loss": 0.5102,
+      "step": 10760
+    },
+    {
+      "epoch": 0.5968081569322842,
+      "grad_norm": 1.3599427938461304,
+      "learning_rate": 3.564860781250598e-07,
+      "loss": 0.5175,
+      "step": 10770
+    },
+    {
+      "epoch": 0.5973622963537626,
+      "grad_norm": 1.3462934494018555,
+      "learning_rate": 3.5564403864284307e-07,
+      "loss": 0.5136,
+      "step": 10780
+    },
+    {
+      "epoch": 0.597916435775241,
+      "grad_norm": 1.284169316291809,
+      "learning_rate": 3.5480244556487103e-07,
+      "loss": 0.5044,
+      "step": 10790
+    },
+    {
+      "epoch": 0.5984705751967195,
+      "grad_norm": 1.2594971656799316,
+      "learning_rate": 3.539613014936739e-07,
+      "loss": 0.5141,
+      "step": 10800
+    },
+    {
+      "epoch": 0.5984705751967195,
+      "eval_robovqa_loss": 0.08020659536123276,
+      "step": 10800
+    },
+    {
+      "epoch": 0.5984705751967195,
+      "eval_agibot_loss": 0.45892098546028137,
+      "step": 10800
+    },
+    {
+      "epoch": 0.5984705751967195,
+      "eval_holoassist_loss": 0.33629173040390015,
+      "step": 10800
+    },
+    {
+      "epoch": 0.5984705751967195,
+      "eval_robofail_loss": 0.20860034227371216,
+      "step": 10800
+    },
+    {
+      "epoch": 0.599024714618198,
+      "grad_norm": 1.249420404434204,
+      "learning_rate": 3.531206090303934e-07,
+      "loss": 0.5185,
+      "step": 10810
+    },
+    {
+      "epoch": 0.5995788540396764,
+      "grad_norm": 1.1957871913909912,
+      "learning_rate": 3.5228037077477425e-07,
+      "loss": 0.5138,
+      "step": 10820
+    },
+    {
+      "epoch": 0.6001329934611548,
+      "grad_norm": 1.2496684789657593,
+      "learning_rate": 3.51440589325157e-07,
+      "loss": 0.5198,
+      "step": 10830
+    },
+    {
+      "epoch": 0.6006871328826333,
+      "grad_norm": 1.260445475578308,
+      "learning_rate": 3.5060126727846904e-07,
+      "loss": 0.513,
+      "step": 10840
+    },
+    {
+      "epoch": 0.6012412723041117,
+      "grad_norm": 1.3042587041854858,
+      "learning_rate": 3.4976240723021786e-07,
+      "loss": 0.5124,
+      "step": 10850
+    },
+    {
+      "epoch": 0.6017954117255901,
+      "grad_norm": 1.2447161674499512,
+      "learning_rate": 3.48924011774482e-07,
+      "loss": 0.513,
+      "step": 10860
+    },
+    {
+      "epoch": 0.6023495511470686,
+      "grad_norm": 1.3216168880462646,
+      "learning_rate": 3.4808608350390273e-07,
+      "loss": 0.5082,
+      "step": 10870
+    },
+    {
+      "epoch": 0.6029036905685471,
+      "grad_norm": 1.2410131692886353,
+      "learning_rate": 3.4724862500967733e-07,
+      "loss": 0.5078,
+      "step": 10880
+    },
+    {
+      "epoch": 0.6034578299900255,
+      "grad_norm": 1.2269093990325928,
+      "learning_rate": 3.4641163888155023e-07,
+      "loss": 0.5139,
+      "step": 10890
+    },
+    {
+      "epoch": 0.6040119694115039,
+      "grad_norm": 1.313354253768921,
+      "learning_rate": 3.4557512770780467e-07,
+      "loss": 0.516,
+      "step": 10900
+    },
+    {
+      "epoch": 0.6040119694115039,
+      "eval_robovqa_loss": 0.07769393920898438,
+      "step": 10900
+    },
+    {
+      "epoch": 0.6040119694115039,
+      "eval_agibot_loss": 0.4607686698436737,
+      "step": 10900
+    },
+    {
+      "epoch": 0.6040119694115039,
+      "eval_holoassist_loss": 0.33852553367614746,
+      "step": 10900
+    },
+    {
+      "epoch": 0.6040119694115039,
+      "eval_robofail_loss": 0.20862308144569397,
+      "step": 10900
+    },
+    {
+      "epoch": 0.6045661088329823,
+      "grad_norm": 1.2986255884170532,
+      "learning_rate": 3.447390940752558e-07,
+      "loss": 0.5114,
+      "step": 10910
+    },
+    {
+      "epoch": 0.6051202482544609,
+      "grad_norm": 1.199716329574585,
+      "learning_rate": 3.4390354056924143e-07,
+      "loss": 0.5101,
+      "step": 10920
+    },
+    {
+      "epoch": 0.6056743876759393,
+      "grad_norm": 1.1841987371444702,
+      "learning_rate": 3.4306846977361496e-07,
+      "loss": 0.5185,
+      "step": 10930
+    },
+    {
+      "epoch": 0.6062285270974177,
+      "grad_norm": 1.265993595123291,
+      "learning_rate": 3.4223388427073706e-07,
+      "loss": 0.5123,
+      "step": 10940
+    },
+    {
+      "epoch": 0.6067826665188961,
+      "grad_norm": 1.2016042470932007,
+      "learning_rate": 3.413997866414676e-07,
+      "loss": 0.4977,
+      "step": 10950
+    },
+    {
+      "epoch": 0.6073368059403746,
+      "grad_norm": 1.219091773033142,
+      "learning_rate": 3.4056617946515767e-07,
+      "loss": 0.5106,
+      "step": 10960
+    },
+    {
+      "epoch": 0.6078909453618531,
+      "grad_norm": 1.2360432147979736,
+      "learning_rate": 3.3973306531964206e-07,
+      "loss": 0.5105,
+      "step": 10970
+    },
+    {
+      "epoch": 0.6084450847833315,
+      "grad_norm": 1.2271119356155396,
+      "learning_rate": 3.389004467812303e-07,
+      "loss": 0.5065,
+      "step": 10980
+    },
+    {
+      "epoch": 0.6089992242048099,
+      "grad_norm": 1.2606054544448853,
+      "learning_rate": 3.3806832642470007e-07,
+      "loss": 0.5044,
+      "step": 10990
+    },
+    {
+      "epoch": 0.6095533636262884,
+      "grad_norm": 1.2872308492660522,
+      "learning_rate": 3.372367068232877e-07,
+      "loss": 0.5147,
+      "step": 11000
+    },
+    {
+      "epoch": 0.6095533636262884,
+      "eval_robovqa_loss": 0.07822806388139725,
+      "step": 11000
+    },
+    {
+      "epoch": 0.6095533636262884,
+      "eval_agibot_loss": 0.46361038088798523,
+      "step": 11000
+    },
+    {
+      "epoch": 0.6095533636262884,
+      "eval_holoassist_loss": 0.3325677216053009,
+      "step": 11000
+    },
+    {
+      "epoch": 0.6095533636262884,
+      "eval_robofail_loss": 0.20678497850894928,
+      "step": 11000
+    },
+    {
+      "epoch": 0.6101075030477668,
+      "grad_norm": 1.3499499559402466,
+      "learning_rate": 3.3640559054868155e-07,
+      "loss": 0.514,
+      "step": 11010
+    },
+    {
+      "epoch": 0.6106616424692453,
+      "grad_norm": 1.4686654806137085,
+      "learning_rate": 3.355749801710135e-07,
+      "loss": 0.5121,
+      "step": 11020
+    },
+    {
+      "epoch": 0.6112157818907237,
+      "grad_norm": 1.240614414215088,
+      "learning_rate": 3.347448782588505e-07,
+      "loss": 0.5128,
+      "step": 11030
+    },
+    {
+      "epoch": 0.6117699213122022,
+      "grad_norm": 1.2551908493041992,
+      "learning_rate": 3.339152873791877e-07,
+      "loss": 0.5185,
+      "step": 11040
+    },
+    {
+      "epoch": 0.6123240607336806,
+      "grad_norm": 1.1896616220474243,
+      "learning_rate": 3.330862100974396e-07,
+      "loss": 0.508,
+      "step": 11050
+    },
+    {
+      "epoch": 0.612878200155159,
+      "grad_norm": 1.1615222692489624,
+      "learning_rate": 3.322576489774326e-07,
+      "loss": 0.5016,
+      "step": 11060
+    },
+    {
+      "epoch": 0.6134323395766375,
+      "grad_norm": 1.4366735219955444,
+      "learning_rate": 3.3142960658139685e-07,
+      "loss": 0.5094,
+      "step": 11070
+    },
+    {
+      "epoch": 0.613986478998116,
+      "grad_norm": 1.1435726881027222,
+      "learning_rate": 3.3060208546995834e-07,
+      "loss": 0.5059,
+      "step": 11080
+    },
+    {
+      "epoch": 0.6145406184195944,
+      "grad_norm": 1.2331095933914185,
+      "learning_rate": 3.297750882021312e-07,
+      "loss": 0.5049,
+      "step": 11090
+    },
+    {
+      "epoch": 0.6150947578410728,
+      "grad_norm": 1.2976559400558472,
+      "learning_rate": 3.2894861733530975e-07,
+      "loss": 0.5144,
+      "step": 11100
+    },
+    {
+      "epoch": 0.6150947578410728,
+      "eval_robovqa_loss": 0.0811123177409172,
+      "step": 11100
+    },
+    {
+      "epoch": 0.6150947578410728,
+      "eval_agibot_loss": 0.4637828767299652,
+      "step": 11100
+    },
+    {
+      "epoch": 0.6150947578410728,
+      "eval_holoassist_loss": 0.3379456698894501,
+      "step": 11100
+    },
+    {
+      "epoch": 0.6150947578410728,
+      "eval_robofail_loss": 0.20718984305858612,
+      "step": 11100
+    },
+    {
+      "epoch": 0.6156488972625512,
+      "grad_norm": 1.6445434093475342,
+      "learning_rate": 3.2812267542526006e-07,
+      "loss": 0.5022,
+      "step": 11110
+    },
+    {
+      "epoch": 0.6162030366840296,
+      "grad_norm": 1.2329752445220947,
+      "learning_rate": 3.272972650261129e-07,
+      "loss": 0.5124,
+      "step": 11120
+    },
+    {
+      "epoch": 0.6167571761055082,
+      "grad_norm": 1.2814217805862427,
+      "learning_rate": 3.2647238869035506e-07,
+      "loss": 0.5061,
+      "step": 11130
+    },
+    {
+      "epoch": 0.6173113155269866,
+      "grad_norm": 1.332851529121399,
+      "learning_rate": 3.256480489688219e-07,
+      "loss": 0.5126,
+      "step": 11140
+    },
+    {
+      "epoch": 0.617865454948465,
+      "grad_norm": 1.256279706954956,
+      "learning_rate": 3.2482424841068977e-07,
+      "loss": 0.513,
+      "step": 11150
+    },
+    {
+      "epoch": 0.6184195943699434,
+      "grad_norm": 1.2528589963912964,
+      "learning_rate": 3.24000989563467e-07,
+      "loss": 0.516,
+      "step": 11160
+    },
+    {
+      "epoch": 0.618973733791422,
+      "grad_norm": 1.3208445310592651,
+      "learning_rate": 3.2317827497298717e-07,
+      "loss": 0.5211,
+      "step": 11170
+    },
+    {
+      "epoch": 0.6195278732129004,
+      "grad_norm": 1.318082571029663,
+      "learning_rate": 3.2235610718340095e-07,
+      "loss": 0.515,
+      "step": 11180
+    },
+    {
+      "epoch": 0.6200820126343788,
+      "grad_norm": 1.2063268423080444,
+      "learning_rate": 3.215344887371675e-07,
+      "loss": 0.5132,
+      "step": 11190
+    },
+    {
+      "epoch": 0.6206361520558572,
+      "grad_norm": 1.2269483804702759,
+      "learning_rate": 3.2071342217504803e-07,
+      "loss": 0.5075,
+      "step": 11200
+    },
+    {
+      "epoch": 0.6206361520558572,
+      "eval_robovqa_loss": 0.07939257472753525,
+      "step": 11200
+    },
+    {
+      "epoch": 0.6206361520558572,
+      "eval_agibot_loss": 0.46379852294921875,
+      "step": 11200
+    },
+    {
+      "epoch": 0.6206361520558572,
+      "eval_holoassist_loss": 0.33547863364219666,
+      "step": 11200
+    },
+    {
+      "epoch": 0.6206361520558572,
+      "eval_robofail_loss": 0.2077992856502533,
+      "step": 11200
+    },
+    {
+      "epoch": 0.6211902914773357,
+      "grad_norm": 1.2985550165176392,
+      "learning_rate": 3.198929100360962e-07,
+      "loss": 0.5175,
+      "step": 11210
+    },
+    {
+      "epoch": 0.6217444308988141,
+      "grad_norm": 1.199530839920044,
+      "learning_rate": 3.190729548576519e-07,
+      "loss": 0.5033,
+      "step": 11220
+    },
+    {
+      "epoch": 0.6222985703202926,
+      "grad_norm": 1.2140053510665894,
+      "learning_rate": 3.1825355917533256e-07,
+      "loss": 0.5088,
+      "step": 11230
+    },
+    {
+      "epoch": 0.622852709741771,
+      "grad_norm": 1.634189486503601,
+      "learning_rate": 3.17434725523025e-07,
+      "loss": 0.511,
+      "step": 11240
+    },
+    {
+      "epoch": 0.6234068491632495,
+      "grad_norm": 1.3240587711334229,
+      "learning_rate": 3.166164564328785e-07,
+      "loss": 0.5116,
+      "step": 11250
+    },
+    {
+      "epoch": 0.6239609885847279,
+      "grad_norm": 1.338757872581482,
+      "learning_rate": 3.1579875443529626e-07,
+      "loss": 0.5123,
+      "step": 11260
+    },
+    {
+      "epoch": 0.6245151280062063,
+      "grad_norm": 1.2171815633773804,
+      "learning_rate": 3.1498162205892796e-07,
+      "loss": 0.5146,
+      "step": 11270
+    },
+    {
+      "epoch": 0.6250692674276848,
+      "grad_norm": 1.3075615167617798,
+      "learning_rate": 3.141650618306617e-07,
+      "loss": 0.4989,
+      "step": 11280
+    },
+    {
+      "epoch": 0.6256234068491633,
+      "grad_norm": 1.2877293825149536,
+      "learning_rate": 3.13349076275616e-07,
+      "loss": 0.4972,
+      "step": 11290
+    },
+    {
+      "epoch": 0.6261775462706417,
+      "grad_norm": 1.213877558708191,
+      "learning_rate": 3.1253366791713283e-07,
+      "loss": 0.5087,
+      "step": 11300
+    },
+    {
+      "epoch": 0.6261775462706417,
+      "eval_robovqa_loss": 0.07869003713130951,
+      "step": 11300
+    },
+    {
+      "epoch": 0.6261775462706417,
+      "eval_agibot_loss": 0.4625292122364044,
+      "step": 11300
+    },
+    {
+      "epoch": 0.6261775462706417,
+      "eval_holoassist_loss": 0.336405485868454,
+      "step": 11300
+    },
+    {
+      "epoch": 0.6261775462706417,
+      "eval_robofail_loss": 0.20608943700790405,
+      "step": 11300
+    },
+    {
+      "epoch": 0.6267316856921201,
+      "grad_norm": 1.1909806728363037,
+      "learning_rate": 3.1171883927676903e-07,
+      "loss": 0.513,
+      "step": 11310
+    },
+    {
+      "epoch": 0.6272858251135985,
+      "grad_norm": 1.3025124073028564,
+      "learning_rate": 3.109045928742884e-07,
+      "loss": 0.5046,
+      "step": 11320
+    },
+    {
+      "epoch": 0.6278399645350771,
+      "grad_norm": 1.3179051876068115,
+      "learning_rate": 3.100909312276548e-07,
+      "loss": 0.504,
+      "step": 11330
+    },
+    {
+      "epoch": 0.6283941039565555,
+      "grad_norm": 1.244532585144043,
+      "learning_rate": 3.092778568530231e-07,
+      "loss": 0.5002,
+      "step": 11340
+    },
+    {
+      "epoch": 0.6289482433780339,
+      "grad_norm": 1.2796992063522339,
+      "learning_rate": 3.084653722647328e-07,
+      "loss": 0.5006,
+      "step": 11350
+    },
+    {
+      "epoch": 0.6295023827995123,
+      "grad_norm": 1.281209945678711,
+      "learning_rate": 3.076534799752994e-07,
+      "loss": 0.5068,
+      "step": 11360
+    },
+    {
+      "epoch": 0.6300565222209908,
+      "grad_norm": 1.1903914213180542,
+      "learning_rate": 3.0684218249540626e-07,
+      "loss": 0.5085,
+      "step": 11370
+    },
+    {
+      "epoch": 0.6306106616424693,
+      "grad_norm": 1.2716257572174072,
+      "learning_rate": 3.0603148233389787e-07,
+      "loss": 0.5092,
+      "step": 11380
+    },
+    {
+      "epoch": 0.6311648010639477,
+      "grad_norm": 1.2868614196777344,
+      "learning_rate": 3.052213819977717e-07,
+      "loss": 0.4969,
+      "step": 11390
+    },
+    {
+      "epoch": 0.6317189404854261,
+      "grad_norm": 1.184383749961853,
+      "learning_rate": 3.044118839921696e-07,
+      "loss": 0.5045,
+      "step": 11400
+    },
+    {
+      "epoch": 0.6317189404854261,
+      "eval_robovqa_loss": 0.07939770072698593,
+      "step": 11400
+    },
+    {
+      "epoch": 0.6317189404854261,
+      "eval_agibot_loss": 0.4706380367279053,
+      "step": 11400
+    },
+    {
+      "epoch": 0.6317189404854261,
+      "eval_holoassist_loss": 0.3449689447879791,
+      "step": 11400
+    },
+    {
+      "epoch": 0.6317189404854261,
+      "eval_robofail_loss": 0.20936480164527893,
+      "step": 11400
+    },
+    {
+      "epoch": 0.6322730799069046,
+      "grad_norm": 1.449634075164795,
+      "learning_rate": 3.036029908203718e-07,
+      "loss": 0.506,
+      "step": 11410
+    },
+    {
+      "epoch": 0.632827219328383,
+      "grad_norm": 1.2799584865570068,
+      "learning_rate": 3.02794704983787e-07,
+      "loss": 0.515,
+      "step": 11420
+    },
+    {
+      "epoch": 0.6333813587498615,
+      "grad_norm": 1.220954418182373,
+      "learning_rate": 3.019870289819468e-07,
+      "loss": 0.5105,
+      "step": 11430
+    },
+    {
+      "epoch": 0.6339354981713399,
+      "grad_norm": 1.1510711908340454,
+      "learning_rate": 3.0117996531249664e-07,
+      "loss": 0.5084,
+      "step": 11440
+    },
+    {
+      "epoch": 0.6344896375928184,
+      "grad_norm": 1.4284131526947021,
+      "learning_rate": 3.003735164711879e-07,
+      "loss": 0.5066,
+      "step": 11450
+    },
+    {
+      "epoch": 0.6350437770142968,
+      "grad_norm": 1.3893249034881592,
+      "learning_rate": 2.995676849518715e-07,
+      "loss": 0.5029,
+      "step": 11460
+    },
+    {
+      "epoch": 0.6355979164357752,
+      "grad_norm": 1.2251434326171875,
+      "learning_rate": 2.987624732464887e-07,
+      "loss": 0.5221,
+      "step": 11470
+    },
+    {
+      "epoch": 0.6361520558572537,
+      "grad_norm": 1.2039496898651123,
+      "learning_rate": 2.9795788384506434e-07,
+      "loss": 0.51,
+      "step": 11480
+    },
+    {
+      "epoch": 0.6367061952787322,
+      "grad_norm": 1.39149010181427,
+      "learning_rate": 2.971539192356989e-07,
+      "loss": 0.5139,
+      "step": 11490
+    },
+    {
+      "epoch": 0.6372603347002106,
+      "grad_norm": 1.3003735542297363,
+      "learning_rate": 2.9635058190456043e-07,
+      "loss": 0.5092,
+      "step": 11500
+    },
+    {
+      "epoch": 0.6372603347002106,
+      "eval_robovqa_loss": 0.07975860685110092,
+      "step": 11500
+    },
+    {
+      "epoch": 0.6372603347002106,
+      "eval_agibot_loss": 0.4655715227127075,
+      "step": 11500
+    },
+    {
+      "epoch": 0.6372603347002106,
+      "eval_holoassist_loss": 0.3415605127811432,
+      "step": 11500
+    },
+    {
+      "epoch": 0.6372603347002106,
+      "eval_robofail_loss": 0.20872145891189575,
+      "step": 11500
+    },
+    {
+      "epoch": 0.637814474121689,
+      "grad_norm": 1.3987857103347778,
+      "learning_rate": 2.9554787433587756e-07,
+      "loss": 0.5161,
+      "step": 11510
+    },
+    {
+      "epoch": 0.6383686135431674,
+      "grad_norm": 1.2389594316482544,
+      "learning_rate": 2.9474579901193165e-07,
+      "loss": 0.5094,
+      "step": 11520
+    },
+    {
+      "epoch": 0.638922752964646,
+      "grad_norm": 1.295186161994934,
+      "learning_rate": 2.939443584130481e-07,
+      "loss": 0.5098,
+      "step": 11530
+    },
+    {
+      "epoch": 0.6394768923861244,
+      "grad_norm": 1.3744940757751465,
+      "learning_rate": 2.9314355501759025e-07,
+      "loss": 0.5136,
+      "step": 11540
+    },
+    {
+      "epoch": 0.6400310318076028,
+      "grad_norm": 1.3898969888687134,
+      "learning_rate": 2.923433913019508e-07,
+      "loss": 0.5167,
+      "step": 11550
+    },
+    {
+      "epoch": 0.6405851712290812,
+      "grad_norm": 1.349948525428772,
+      "learning_rate": 2.91543869740544e-07,
+      "loss": 0.5154,
+      "step": 11560
+    },
+    {
+      "epoch": 0.6411393106505597,
+      "grad_norm": 1.2924193143844604,
+      "learning_rate": 2.9074499280579867e-07,
+      "loss": 0.5062,
+      "step": 11570
+    },
+    {
+      "epoch": 0.6416934500720382,
+      "grad_norm": 1.2585166692733765,
+      "learning_rate": 2.8994676296814996e-07,
+      "loss": 0.5098,
+      "step": 11580
+    },
+    {
+      "epoch": 0.6422475894935166,
+      "grad_norm": 1.3129992485046387,
+      "learning_rate": 2.8914918269603187e-07,
+      "loss": 0.5087,
+      "step": 11590
+    },
+    {
+      "epoch": 0.642801728914995,
+      "grad_norm": 1.2629730701446533,
+      "learning_rate": 2.883522544558705e-07,
+      "loss": 0.5155,
+      "step": 11600
+    },
+    {
+      "epoch": 0.642801728914995,
+      "eval_robovqa_loss": 0.07941696047782898,
+      "step": 11600
+    },
+    {
+      "epoch": 0.642801728914995,
+      "eval_agibot_loss": 0.4650159180164337,
+      "step": 11600
+    },
+    {
+      "epoch": 0.642801728914995,
+      "eval_holoassist_loss": 0.32951605319976807,
+      "step": 11600
+    },
+    {
+      "epoch": 0.642801728914995,
+      "eval_robofail_loss": 0.2089124321937561,
+      "step": 11600
+    },
+    {
+      "epoch": 0.6433558683364734,
+      "grad_norm": 1.3020241260528564,
+      "learning_rate": 2.875559807120742e-07,
+      "loss": 0.5023,
+      "step": 11610
+    },
+    {
+      "epoch": 0.6439100077579519,
+      "grad_norm": 1.3346240520477295,
+      "learning_rate": 2.8676036392702886e-07,
+      "loss": 0.5094,
+      "step": 11620
+    },
+    {
+      "epoch": 0.6444641471794303,
+      "grad_norm": 1.2748826742172241,
+      "learning_rate": 2.8596540656108735e-07,
+      "loss": 0.5078,
+      "step": 11630
+    },
+    {
+      "epoch": 0.6450182866009088,
+      "grad_norm": 1.2600610256195068,
+      "learning_rate": 2.851711110725647e-07,
+      "loss": 0.5128,
+      "step": 11640
+    },
+    {
+      "epoch": 0.6455724260223872,
+      "grad_norm": 1.2421635389328003,
+      "learning_rate": 2.843774799177285e-07,
+      "loss": 0.5054,
+      "step": 11650
+    },
+    {
+      "epoch": 0.6461265654438657,
+      "grad_norm": 1.2695767879486084,
+      "learning_rate": 2.8358451555079156e-07,
+      "loss": 0.5016,
+      "step": 11660
+    },
+    {
+      "epoch": 0.6466807048653441,
+      "grad_norm": 1.2956106662750244,
+      "learning_rate": 2.827922204239056e-07,
+      "loss": 0.5118,
+      "step": 11670
+    },
+    {
+      "epoch": 0.6472348442868225,
+      "grad_norm": 1.2208421230316162,
+      "learning_rate": 2.8200059698715226e-07,
+      "loss": 0.5038,
+      "step": 11680
+    },
+    {
+      "epoch": 0.647788983708301,
+      "grad_norm": 1.43162202835083,
+      "learning_rate": 2.8120964768853625e-07,
+      "loss": 0.4971,
+      "step": 11690
+    },
+    {
+      "epoch": 0.6483431231297795,
+      "grad_norm": 1.236934781074524,
+      "learning_rate": 2.804193749739774e-07,
+      "loss": 0.5057,
+      "step": 11700
+    },
+    {
+      "epoch": 0.6483431231297795,
+      "eval_robovqa_loss": 0.07947414368391037,
+      "step": 11700
+    },
+    {
+      "epoch": 0.6483431231297795,
+      "eval_agibot_loss": 0.46406620740890503,
+      "step": 11700
+    },
+    {
+      "epoch": 0.6483431231297795,
+      "eval_holoassist_loss": 0.3334883749485016,
+      "step": 11700
+    },
+    {
+      "epoch": 0.6483431231297795,
+      "eval_robofail_loss": 0.20650313794612885,
+      "step": 11700
+    },
+    {
+      "epoch": 0.6488972625512579,
+      "grad_norm": 1.4149101972579956,
+      "learning_rate": 2.7962978128730354e-07,
+      "loss": 0.5078,
+      "step": 11710
+    },
+    {
+      "epoch": 0.6494514019727363,
+      "grad_norm": 1.2849150896072388,
+      "learning_rate": 2.788408690702423e-07,
+      "loss": 0.5049,
+      "step": 11720
+    },
+    {
+      "epoch": 0.6500055413942147,
+      "grad_norm": 1.292339563369751,
+      "learning_rate": 2.7805264076241443e-07,
+      "loss": 0.5084,
+      "step": 11730
+    },
+    {
+      "epoch": 0.6505596808156933,
+      "grad_norm": 1.361863136291504,
+      "learning_rate": 2.772650988013253e-07,
+      "loss": 0.5013,
+      "step": 11740
+    },
+    {
+      "epoch": 0.6511138202371717,
+      "grad_norm": 1.3807986974716187,
+      "learning_rate": 2.7647824562235797e-07,
+      "loss": 0.5029,
+      "step": 11750
+    },
+    {
+      "epoch": 0.6516679596586501,
+      "grad_norm": 1.3825058937072754,
+      "learning_rate": 2.7569208365876626e-07,
+      "loss": 0.5085,
+      "step": 11760
+    },
+    {
+      "epoch": 0.6522220990801285,
+      "grad_norm": 2.2140603065490723,
+      "learning_rate": 2.7490661534166495e-07,
+      "loss": 0.514,
+      "step": 11770
+    },
+    {
+      "epoch": 0.652776238501607,
+      "grad_norm": 1.1538255214691162,
+      "learning_rate": 2.741218431000254e-07,
+      "loss": 0.5083,
+      "step": 11780
+    },
+    {
+      "epoch": 0.6533303779230855,
+      "grad_norm": 1.2560113668441772,
+      "learning_rate": 2.7333776936066503e-07,
+      "loss": 0.5094,
+      "step": 11790
+    },
+    {
+      "epoch": 0.6538845173445639,
+      "grad_norm": 1.2154790163040161,
+      "learning_rate": 2.7255439654824255e-07,
+      "loss": 0.5124,
+      "step": 11800
+    },
+    {
+      "epoch": 0.6538845173445639,
+      "eval_robovqa_loss": 0.07828789949417114,
+      "step": 11800
+    },
+    {
+      "epoch": 0.6538845173445639,
+      "eval_agibot_loss": 0.47191324830055237,
+      "step": 11800
+    },
+    {
+      "epoch": 0.6538845173445639,
+      "eval_holoassist_loss": 0.3396921157836914,
+      "step": 11800
+    },
+    {
+      "epoch": 0.6538845173445639,
+      "eval_robofail_loss": 0.2082146257162094,
+      "step": 11800
+    },
+    {
+      "epoch": 0.6544386567660423,
+      "grad_norm": 1.3167015314102173,
+      "learning_rate": 2.71771727085248e-07,
+      "loss": 0.5034,
+      "step": 11810
+    },
+    {
+      "epoch": 0.6549927961875208,
+      "grad_norm": 1.2237238883972168,
+      "learning_rate": 2.709897633919972e-07,
+      "loss": 0.5074,
+      "step": 11820
+    },
+    {
+      "epoch": 0.6555469356089992,
+      "grad_norm": 1.281510591506958,
+      "learning_rate": 2.70208507886623e-07,
+      "loss": 0.5098,
+      "step": 11830
+    },
+    {
+      "epoch": 0.6561010750304777,
+      "grad_norm": 1.3285861015319824,
+      "learning_rate": 2.694279629850684e-07,
+      "loss": 0.512,
+      "step": 11840
+    },
+    {
+      "epoch": 0.6566552144519561,
+      "grad_norm": 1.301756501197815,
+      "learning_rate": 2.68648131101079e-07,
+      "loss": 0.5112,
+      "step": 11850
+    },
+    {
+      "epoch": 0.6572093538734346,
+      "grad_norm": 1.3204463720321655,
+      "learning_rate": 2.678690146461953e-07,
+      "loss": 0.5049,
+      "step": 11860
+    },
+    {
+      "epoch": 0.657763493294913,
+      "grad_norm": 1.308545708656311,
+      "learning_rate": 2.670906160297456e-07,
+      "loss": 0.515,
+      "step": 11870
+    },
+    {
+      "epoch": 0.6583176327163914,
+      "grad_norm": 1.405219554901123,
+      "learning_rate": 2.663129376588381e-07,
+      "loss": 0.5092,
+      "step": 11880
+    },
+    {
+      "epoch": 0.6588717721378698,
+      "grad_norm": 1.3876863718032837,
+      "learning_rate": 2.655359819383544e-07,
+      "loss": 0.5054,
+      "step": 11890
+    },
+    {
+      "epoch": 0.6594259115593484,
+      "grad_norm": 1.4001253843307495,
+      "learning_rate": 2.647597512709403e-07,
+      "loss": 0.5162,
+      "step": 11900
+    },
+    {
+      "epoch": 0.6594259115593484,
+      "eval_robovqa_loss": 0.07834941148757935,
+      "step": 11900
+    },
+    {
+      "epoch": 0.6594259115593484,
+      "eval_agibot_loss": 0.46745842695236206,
+      "step": 11900
+    },
+    {
+      "epoch": 0.6594259115593484,
+      "eval_holoassist_loss": 0.3343101739883423,
+      "step": 11900
+    },
+    {
+      "epoch": 0.6594259115593484,
+      "eval_robofail_loss": 0.20790104568004608,
+      "step": 11900
+    },
+    {
+      "epoch": 0.6599800509808268,
+      "grad_norm": 1.190024495124817,
+      "learning_rate": 2.639842480570008e-07,
+      "loss": 0.5042,
+      "step": 11910
+    },
+    {
+      "epoch": 0.6605341904023052,
+      "grad_norm": 1.2486060857772827,
+      "learning_rate": 2.6320947469468966e-07,
+      "loss": 0.5079,
+      "step": 11920
+    },
+    {
+      "epoch": 0.6610883298237836,
+      "grad_norm": 1.2904120683670044,
+      "learning_rate": 2.624354335799054e-07,
+      "loss": 0.4991,
+      "step": 11930
+    },
+    {
+      "epoch": 0.6616424692452622,
+      "grad_norm": 1.2289960384368896,
+      "learning_rate": 2.61662127106281e-07,
+      "loss": 0.5073,
+      "step": 11940
+    },
+    {
+      "epoch": 0.6621966086667406,
+      "grad_norm": 1.2454793453216553,
+      "learning_rate": 2.6088955766517803e-07,
+      "loss": 0.5129,
+      "step": 11950
+    },
+    {
+      "epoch": 0.662750748088219,
+      "grad_norm": 1.3097715377807617,
+      "learning_rate": 2.6011772764567886e-07,
+      "loss": 0.5079,
+      "step": 11960
+    },
+    {
+      "epoch": 0.6633048875096974,
+      "grad_norm": 1.3632500171661377,
+      "learning_rate": 2.5934663943457913e-07,
+      "loss": 0.5134,
+      "step": 11970
+    },
+    {
+      "epoch": 0.6638590269311759,
+      "grad_norm": 1.3787610530853271,
+      "learning_rate": 2.5857629541638084e-07,
+      "loss": 0.5075,
+      "step": 11980
+    },
+    {
+      "epoch": 0.6644131663526543,
+      "grad_norm": 1.326090693473816,
+      "learning_rate": 2.5780669797328436e-07,
+      "loss": 0.5119,
+      "step": 11990
+    },
+    {
+      "epoch": 0.6649673057741328,
+      "grad_norm": 1.224238634109497,
+      "learning_rate": 2.570378494851815e-07,
+      "loss": 0.5066,
+      "step": 12000
+    },
+    {
+      "epoch": 0.6649673057741328,
+      "eval_robovqa_loss": 0.07904809713363647,
+      "step": 12000
+    },
+    {
+      "epoch": 0.6649673057741328,
+      "eval_agibot_loss": 0.47449058294296265,
+      "step": 12000
+    },
+    {
+      "epoch": 0.6649673057741328,
+      "eval_holoassist_loss": 0.3384062945842743,
+      "step": 12000
+    },
+    {
+      "epoch": 0.6649673057741328,
+      "eval_robofail_loss": 0.20950989425182343,
+      "step": 12000
+    },
+    {
+      "epoch": 0.6655214451956112,
+      "grad_norm": 1.2585564851760864,
+      "learning_rate": 2.5626975232964765e-07,
+      "loss": 0.5106,
+      "step": 12010
+    },
+    {
+      "epoch": 0.6660755846170897,
+      "grad_norm": 1.2485289573669434,
+      "learning_rate": 2.5550240888193587e-07,
+      "loss": 0.5109,
+      "step": 12020
+    },
+    {
+      "epoch": 0.6666297240385681,
+      "grad_norm": 1.313181757926941,
+      "learning_rate": 2.547358215149669e-07,
+      "loss": 0.5059,
+      "step": 12030
+    },
+    {
+      "epoch": 0.6671838634600465,
+      "grad_norm": 1.2604310512542725,
+      "learning_rate": 2.539699925993248e-07,
+      "loss": 0.5089,
+      "step": 12040
+    },
+    {
+      "epoch": 0.667738002881525,
+      "grad_norm": 1.3345324993133545,
+      "learning_rate": 2.532049245032473e-07,
+      "loss": 0.4985,
+      "step": 12050
+    },
+    {
+      "epoch": 0.6682921423030035,
+      "grad_norm": 1.2951722145080566,
+      "learning_rate": 2.5244061959261985e-07,
+      "loss": 0.5144,
+      "step": 12060
+    },
+    {
+      "epoch": 0.6688462817244819,
+      "grad_norm": 1.3322888612747192,
+      "learning_rate": 2.5167708023096776e-07,
+      "loss": 0.513,
+      "step": 12070
+    },
+    {
+      "epoch": 0.6694004211459603,
+      "grad_norm": 1.2499666213989258,
+      "learning_rate": 2.509143087794483e-07,
+      "loss": 0.5084,
+      "step": 12080
+    },
+    {
+      "epoch": 0.6699545605674387,
+      "grad_norm": 1.517769694328308,
+      "learning_rate": 2.5015230759684536e-07,
+      "loss": 0.5014,
+      "step": 12090
+    },
+    {
+      "epoch": 0.6705086999889172,
+      "grad_norm": 1.258423924446106,
+      "learning_rate": 2.493910790395598e-07,
+      "loss": 0.5043,
+      "step": 12100
+    },
+    {
+      "epoch": 0.6705086999889172,
+      "eval_robovqa_loss": 0.07928184419870377,
+      "step": 12100
+    },
+    {
+      "epoch": 0.6705086999889172,
+      "eval_agibot_loss": 0.4716735780239105,
+      "step": 12100
+    },
+    {
+      "epoch": 0.6705086999889172,
+      "eval_holoassist_loss": 0.34370583295822144,
+      "step": 12100
+    },
+    {
+      "epoch": 0.6705086999889172,
+      "eval_robofail_loss": 0.21027114987373352,
+      "step": 12100
+    },
+    {
+      "epoch": 0.6710628394103957,
+      "grad_norm": 1.257466435432434,
+      "learning_rate": 2.486306254616037e-07,
+      "loss": 0.5024,
+      "step": 12110
+    },
+    {
+      "epoch": 0.6716169788318741,
+      "grad_norm": 1.2684075832366943,
+      "learning_rate": 2.478709492145924e-07,
+      "loss": 0.5072,
+      "step": 12120
+    },
+    {
+      "epoch": 0.6721711182533525,
+      "grad_norm": 1.2231395244598389,
+      "learning_rate": 2.471120526477376e-07,
+      "loss": 0.503,
+      "step": 12130
+    },
+    {
+      "epoch": 0.6727252576748309,
+      "grad_norm": 1.316768765449524,
+      "learning_rate": 2.4635393810783955e-07,
+      "loss": 0.5021,
+      "step": 12140
+    },
+    {
+      "epoch": 0.6732793970963095,
+      "grad_norm": 1.3988139629364014,
+      "learning_rate": 2.4559660793928114e-07,
+      "loss": 0.5014,
+      "step": 12150
+    },
+    {
+      "epoch": 0.6738335365177879,
+      "grad_norm": 1.2170034646987915,
+      "learning_rate": 2.448400644840184e-07,
+      "loss": 0.5004,
+      "step": 12160
+    },
+    {
+      "epoch": 0.6743876759392663,
+      "grad_norm": 1.1897646188735962,
+      "learning_rate": 2.440843100815751e-07,
+      "loss": 0.506,
+      "step": 12170
+    },
+    {
+      "epoch": 0.6749418153607447,
+      "grad_norm": 1.2026582956314087,
+      "learning_rate": 2.4332934706903563e-07,
+      "loss": 0.5079,
+      "step": 12180
+    },
+    {
+      "epoch": 0.6754959547822232,
+      "grad_norm": 1.4329347610473633,
+      "learning_rate": 2.425751777810357e-07,
+      "loss": 0.509,
+      "step": 12190
+    },
+    {
+      "epoch": 0.6760500942037017,
+      "grad_norm": 1.2519539594650269,
+      "learning_rate": 2.418218045497581e-07,
+      "loss": 0.5012,
+      "step": 12200
+    },
+    {
+      "epoch": 0.6760500942037017,
+      "eval_robovqa_loss": 0.08088075369596481,
+      "step": 12200
+    },
+    {
+      "epoch": 0.6760500942037017,
+      "eval_agibot_loss": 0.47206321358680725,
+      "step": 12200
+    },
+    {
+      "epoch": 0.6760500942037017,
+      "eval_holoassist_loss": 0.34168729186058044,
+      "step": 12200
+    },
+    {
+      "epoch": 0.6760500942037017,
+      "eval_robofail_loss": 0.20752422511577606,
+      "step": 12200
+    },
+    {
+      "epoch": 0.6766042336251801,
+      "grad_norm": 1.3242433071136475,
+      "learning_rate": 2.4106922970492224e-07,
+      "loss": 0.4994,
+      "step": 12210
+    },
+    {
+      "epoch": 0.6771583730466585,
+      "grad_norm": 1.3746427297592163,
+      "learning_rate": 2.4031745557378e-07,
+      "loss": 0.5033,
+      "step": 12220
+    },
+    {
+      "epoch": 0.677712512468137,
+      "grad_norm": 1.3858458995819092,
+      "learning_rate": 2.3956648448110653e-07,
+      "loss": 0.4976,
+      "step": 12230
+    },
+    {
+      "epoch": 0.6782666518896154,
+      "grad_norm": 1.190671443939209,
+      "learning_rate": 2.3881631874919376e-07,
+      "loss": 0.514,
+      "step": 12240
+    },
+    {
+      "epoch": 0.6788207913110939,
+      "grad_norm": 1.2883816957473755,
+      "learning_rate": 2.380669606978431e-07,
+      "loss": 0.5077,
+      "step": 12250
+    },
+    {
+      "epoch": 0.6793749307325723,
+      "grad_norm": 1.3162753582000732,
+      "learning_rate": 2.373184126443583e-07,
+      "loss": 0.5031,
+      "step": 12260
+    },
+    {
+      "epoch": 0.6799290701540508,
+      "grad_norm": 1.2835661172866821,
+      "learning_rate": 2.3657067690353838e-07,
+      "loss": 0.5039,
+      "step": 12270
+    },
+    {
+      "epoch": 0.6804832095755292,
+      "grad_norm": 1.3551746606826782,
+      "learning_rate": 2.3582375578767022e-07,
+      "loss": 0.5077,
+      "step": 12280
+    },
+    {
+      "epoch": 0.6810373489970076,
+      "grad_norm": 1.5204124450683594,
+      "learning_rate": 2.3507765160652171e-07,
+      "loss": 0.5159,
+      "step": 12290
+    },
+    {
+      "epoch": 0.681591488418486,
+      "grad_norm": 1.2762117385864258,
+      "learning_rate": 2.3433236666733426e-07,
+      "loss": 0.4996,
+      "step": 12300
+    },
+    {
+      "epoch": 0.681591488418486,
+      "eval_robovqa_loss": 0.08026452362537384,
+      "step": 12300
+    },
+    {
+      "epoch": 0.681591488418486,
+      "eval_agibot_loss": 0.4723127782344818,
+      "step": 12300
+    },
+    {
+      "epoch": 0.681591488418486,
+      "eval_holoassist_loss": 0.3420204222202301,
+      "step": 12300
+    },
+    {
+      "epoch": 0.681591488418486,
+      "eval_robofail_loss": 0.2068939059972763,
+      "step": 12300
+    },
+    {
+      "epoch": 0.6821456278399646,
+      "grad_norm": 1.2783461809158325,
+      "learning_rate": 2.3358790327481654e-07,
+      "loss": 0.5014,
+      "step": 12310
+    },
+    {
+      "epoch": 0.682699767261443,
+      "grad_norm": 1.2947334051132202,
+      "learning_rate": 2.3284426373113535e-07,
+      "loss": 0.5134,
+      "step": 12320
+    },
+    {
+      "epoch": 0.6832539066829214,
+      "grad_norm": 1.3781307935714722,
+      "learning_rate": 2.321014503359115e-07,
+      "loss": 0.5044,
+      "step": 12330
+    },
+    {
+      "epoch": 0.6838080461043998,
+      "grad_norm": 1.3336130380630493,
+      "learning_rate": 2.3135946538620938e-07,
+      "loss": 0.4998,
+      "step": 12340
+    },
+    {
+      "epoch": 0.6843621855258784,
+      "grad_norm": 1.406515121459961,
+      "learning_rate": 2.3061831117653286e-07,
+      "loss": 0.509,
+      "step": 12350
+    },
+    {
+      "epoch": 0.6849163249473568,
+      "grad_norm": 1.2406295537948608,
+      "learning_rate": 2.2987798999881614e-07,
+      "loss": 0.5023,
+      "step": 12360
+    },
+    {
+      "epoch": 0.6854704643688352,
+      "grad_norm": 1.3135147094726562,
+      "learning_rate": 2.2913850414241737e-07,
+      "loss": 0.5095,
+      "step": 12370
+    },
+    {
+      "epoch": 0.6860246037903136,
+      "grad_norm": 1.2154548168182373,
+      "learning_rate": 2.2839985589411187e-07,
+      "loss": 0.5037,
+      "step": 12380
+    },
+    {
+      "epoch": 0.6865787432117921,
+      "grad_norm": 1.2988569736480713,
+      "learning_rate": 2.2766204753808448e-07,
+      "loss": 0.5052,
+      "step": 12390
+    },
+    {
+      "epoch": 0.6871328826332705,
+      "grad_norm": 1.2792890071868896,
+      "learning_rate": 2.2692508135592293e-07,
+      "loss": 0.5035,
+      "step": 12400
+    },
+    {
+      "epoch": 0.6871328826332705,
+      "eval_robovqa_loss": 0.08017577230930328,
+      "step": 12400
+    },
+    {
+      "epoch": 0.6871328826332705,
+      "eval_agibot_loss": 0.47120726108551025,
+      "step": 12400
+    },
+    {
+      "epoch": 0.6871328826332705,
+      "eval_holoassist_loss": 0.34770262241363525,
+      "step": 12400
+    },
+    {
+      "epoch": 0.6871328826332705,
+      "eval_robofail_loss": 0.20573784410953522,
+      "step": 12400
+    },
+    {
+      "epoch": 0.687687022054749,
+      "grad_norm": 1.2632423639297485,
+      "learning_rate": 2.261889596266105e-07,
+      "loss": 0.507,
+      "step": 12410
+    },
+    {
+      "epoch": 0.6882411614762274,
+      "grad_norm": 1.313950777053833,
+      "learning_rate": 2.2545368462651922e-07,
+      "loss": 0.509,
+      "step": 12420
+    },
+    {
+      "epoch": 0.6887953008977059,
+      "grad_norm": 1.19602632522583,
+      "learning_rate": 2.2471925862940244e-07,
+      "loss": 0.507,
+      "step": 12430
+    },
+    {
+      "epoch": 0.6893494403191843,
+      "grad_norm": 1.9964687824249268,
+      "learning_rate": 2.2398568390638888e-07,
+      "loss": 0.5039,
+      "step": 12440
+    },
+    {
+      "epoch": 0.6899035797406627,
+      "grad_norm": 1.262860655784607,
+      "learning_rate": 2.2325296272597337e-07,
+      "loss": 0.5068,
+      "step": 12450
+    },
+    {
+      "epoch": 0.6904577191621412,
+      "grad_norm": 1.281033992767334,
+      "learning_rate": 2.225210973540127e-07,
+      "loss": 0.5032,
+      "step": 12460
+    },
+    {
+      "epoch": 0.6910118585836197,
+      "grad_norm": 1.252382516860962,
+      "learning_rate": 2.2179009005371635e-07,
+      "loss": 0.5058,
+      "step": 12470
+    },
+    {
+      "epoch": 0.6915659980050981,
+      "grad_norm": 1.4163563251495361,
+      "learning_rate": 2.2105994308564046e-07,
+      "loss": 0.4976,
+      "step": 12480
+    },
+    {
+      "epoch": 0.6921201374265765,
+      "grad_norm": 1.3631582260131836,
+      "learning_rate": 2.2033065870768096e-07,
+      "loss": 0.5019,
+      "step": 12490
+    },
+    {
+      "epoch": 0.6926742768480549,
+      "grad_norm": 1.2653502225875854,
+      "learning_rate": 2.1960223917506554e-07,
+      "loss": 0.5054,
+      "step": 12500
+    },
+    {
+      "epoch": 0.6926742768480549,
+      "eval_robovqa_loss": 0.07985138148069382,
+      "step": 12500
+    },
+    {
+      "epoch": 0.6926742768480549,
+      "eval_agibot_loss": 0.4739623963832855,
+      "step": 12500
+    },
+    {
+      "epoch": 0.6926742768480549,
+      "eval_holoassist_loss": 0.3454645574092865,
+      "step": 12500
+    },
+    {
+      "epoch": 0.6926742768480549,
+      "eval_robofail_loss": 0.20615005493164062,
+      "step": 12500
+    },
+    {
+      "epoch": 0.6932284162695335,
+      "grad_norm": 1.2286384105682373,
+      "learning_rate": 2.1887468674034865e-07,
+      "loss": 0.5062,
+      "step": 12510
+    },
+    {
+      "epoch": 0.6937825556910119,
+      "grad_norm": 1.2688056230545044,
+      "learning_rate": 2.1814800365340237e-07,
+      "loss": 0.5108,
+      "step": 12520
+    },
+    {
+      "epoch": 0.6943366951124903,
+      "grad_norm": 1.2446410655975342,
+      "learning_rate": 2.1742219216141084e-07,
+      "loss": 0.5085,
+      "step": 12530
+    },
+    {
+      "epoch": 0.6948908345339687,
+      "grad_norm": 1.3683459758758545,
+      "learning_rate": 2.166972545088626e-07,
+      "loss": 0.508,
+      "step": 12540
+    },
+    {
+      "epoch": 0.6954449739554472,
+      "grad_norm": 1.2242019176483154,
+      "learning_rate": 2.159731929375448e-07,
+      "loss": 0.5148,
+      "step": 12550
+    },
+    {
+      "epoch": 0.6959991133769257,
+      "grad_norm": 1.2735531330108643,
+      "learning_rate": 2.152500096865339e-07,
+      "loss": 0.5085,
+      "step": 12560
+    },
+    {
+      "epoch": 0.6965532527984041,
+      "grad_norm": 1.4449024200439453,
+      "learning_rate": 2.1452770699219186e-07,
+      "loss": 0.5033,
+      "step": 12570
+    },
+    {
+      "epoch": 0.6971073922198825,
+      "grad_norm": 1.278161883354187,
+      "learning_rate": 2.1380628708815634e-07,
+      "loss": 0.4973,
+      "step": 12580
+    },
+    {
+      "epoch": 0.697661531641361,
+      "grad_norm": 1.3274133205413818,
+      "learning_rate": 2.130857522053357e-07,
+      "loss": 0.5097,
+      "step": 12590
+    },
+    {
+      "epoch": 0.6982156710628394,
+      "grad_norm": 1.197270154953003,
+      "learning_rate": 2.123661045719018e-07,
+      "loss": 0.5051,
+      "step": 12600
+    },
+    {
+      "epoch": 0.6982156710628394,
+      "eval_robovqa_loss": 0.08032442629337311,
+      "step": 12600
+    },
+    {
+      "epoch": 0.6982156710628394,
+      "eval_agibot_loss": 0.47164595127105713,
+      "step": 12600
+    },
+    {
+      "epoch": 0.6982156710628394,
+      "eval_holoassist_loss": 0.34333598613739014,
+      "step": 12600
+    },
+    {
+      "epoch": 0.6982156710628394,
+      "eval_robofail_loss": 0.20693282783031464,
+      "step": 12600
+    },
+    {
+      "epoch": 0.6987698104843179,
+      "grad_norm": 1.4889413118362427,
+      "learning_rate": 2.116473464132817e-07,
+      "loss": 0.5021,
+      "step": 12610
+    },
+    {
+      "epoch": 0.6993239499057963,
+      "grad_norm": 1.3277682065963745,
+      "learning_rate": 2.1092947995215322e-07,
+      "loss": 0.501,
+      "step": 12620
+    },
+    {
+      "epoch": 0.6998780893272747,
+      "grad_norm": 1.2460426092147827,
+      "learning_rate": 2.1021250740843528e-07,
+      "loss": 0.513,
+      "step": 12630
+    },
+    {
+      "epoch": 0.7004322287487532,
+      "grad_norm": 1.2420090436935425,
+      "learning_rate": 2.0949643099928377e-07,
+      "loss": 0.5046,
+      "step": 12640
+    },
+    {
+      "epoch": 0.7009863681702316,
+      "grad_norm": 1.3721883296966553,
+      "learning_rate": 2.0878125293908267e-07,
+      "loss": 0.5009,
+      "step": 12650
+    },
+    {
+      "epoch": 0.70154050759171,
+      "grad_norm": 1.241268277168274,
+      "learning_rate": 2.0806697543943796e-07,
+      "loss": 0.5091,
+      "step": 12660
+    },
+    {
+      "epoch": 0.7020946470131885,
+      "grad_norm": 1.2300727367401123,
+      "learning_rate": 2.073536007091709e-07,
+      "loss": 0.503,
+      "step": 12670
+    },
+    {
+      "epoch": 0.702648786434667,
+      "grad_norm": 1.3524351119995117,
+      "learning_rate": 2.0664113095431097e-07,
+      "loss": 0.5095,
+      "step": 12680
+    },
+    {
+      "epoch": 0.7032029258561454,
+      "grad_norm": 1.3212940692901611,
+      "learning_rate": 2.0592956837808916e-07,
+      "loss": 0.5114,
+      "step": 12690
+    },
+    {
+      "epoch": 0.7037570652776238,
+      "grad_norm": 1.3956350088119507,
+      "learning_rate": 2.0521891518093104e-07,
+      "loss": 0.4992,
+      "step": 12700
+    },
+    {
+      "epoch": 0.7037570652776238,
+      "eval_robovqa_loss": 0.07940621674060822,
+      "step": 12700
+    },
+    {
+      "epoch": 0.7037570652776238,
+      "eval_agibot_loss": 0.4638591408729553,
+      "step": 12700
+    },
+    {
+      "epoch": 0.7037570652776238,
+      "eval_holoassist_loss": 0.34395676851272583,
+      "step": 12700
+    },
+    {
+      "epoch": 0.7037570652776238,
+      "eval_robofail_loss": 0.2067268192768097,
+      "step": 12700
+    },
+    {
+      "epoch": 0.7043112046991022,
+      "grad_norm": 1.3695178031921387,
+      "learning_rate": 2.0450917356045006e-07,
+      "loss": 0.5024,
+      "step": 12710
+    },
+    {
+      "epoch": 0.7048653441205808,
+      "grad_norm": 1.2754809856414795,
+      "learning_rate": 2.0380034571144062e-07,
+      "loss": 0.503,
+      "step": 12720
+    },
+    {
+      "epoch": 0.7054194835420592,
+      "grad_norm": 1.2359682321548462,
+      "learning_rate": 2.0309243382587193e-07,
+      "loss": 0.5113,
+      "step": 12730
+    },
+    {
+      "epoch": 0.7059736229635376,
+      "grad_norm": 1.4112569093704224,
+      "learning_rate": 2.0238544009287965e-07,
+      "loss": 0.5027,
+      "step": 12740
+    },
+    {
+      "epoch": 0.706527762385016,
+      "grad_norm": 1.2389731407165527,
+      "learning_rate": 2.0167936669876128e-07,
+      "loss": 0.5073,
+      "step": 12750
+    },
+    {
+      "epoch": 0.7070819018064946,
+      "grad_norm": 1.2252289056777954,
+      "learning_rate": 2.0097421582696755e-07,
+      "loss": 0.5047,
+      "step": 12760
+    },
+    {
+      "epoch": 0.707636041227973,
+      "grad_norm": 1.2695038318634033,
+      "learning_rate": 2.0026998965809666e-07,
+      "loss": 0.5136,
+      "step": 12770
+    },
+    {
+      "epoch": 0.7081901806494514,
+      "grad_norm": 1.3047209978103638,
+      "learning_rate": 1.995666903698872e-07,
+      "loss": 0.511,
+      "step": 12780
+    },
+    {
+      "epoch": 0.7087443200709298,
+      "grad_norm": 1.4676003456115723,
+      "learning_rate": 1.988643201372115e-07,
+      "loss": 0.5008,
+      "step": 12790
+    },
+    {
+      "epoch": 0.7092984594924083,
+      "grad_norm": 1.31797456741333,
+      "learning_rate": 1.9816288113206887e-07,
+      "loss": 0.5034,
+      "step": 12800
+    },
+    {
+      "epoch": 0.7092984594924083,
+      "eval_robovqa_loss": 0.07861441373825073,
+      "step": 12800
+    },
+    {
+      "epoch": 0.7092984594924083,
+      "eval_agibot_loss": 0.4695967435836792,
+      "step": 12800
+    },
+    {
+      "epoch": 0.7092984594924083,
+      "eval_holoassist_loss": 0.3430720865726471,
+      "step": 12800
+    },
+    {
+      "epoch": 0.7092984594924083,
+      "eval_robofail_loss": 0.20867618918418884,
+      "step": 12800
+    },
+    {
+      "epoch": 0.7098525989138867,
+      "grad_norm": 1.3187572956085205,
+      "learning_rate": 1.9746237552357903e-07,
+      "loss": 0.5002,
+      "step": 12810
+    },
+    {
+      "epoch": 0.7104067383353652,
+      "grad_norm": 1.2811071872711182,
+      "learning_rate": 1.9676280547797514e-07,
+      "loss": 0.5037,
+      "step": 12820
+    },
+    {
+      "epoch": 0.7109608777568436,
+      "grad_norm": 1.3563631772994995,
+      "learning_rate": 1.9606417315859725e-07,
+      "loss": 0.5002,
+      "step": 12830
+    },
+    {
+      "epoch": 0.7115150171783221,
+      "grad_norm": 1.390297770500183,
+      "learning_rate": 1.953664807258857e-07,
+      "loss": 0.4985,
+      "step": 12840
+    },
+    {
+      "epoch": 0.7120691565998005,
+      "grad_norm": 1.3904591798782349,
+      "learning_rate": 1.946697303373741e-07,
+      "loss": 0.4984,
+      "step": 12850
+    },
+    {
+      "epoch": 0.7126232960212789,
+      "grad_norm": 1.5033354759216309,
+      "learning_rate": 1.9397392414768353e-07,
+      "loss": 0.506,
+      "step": 12860
+    },
+    {
+      "epoch": 0.7131774354427574,
+      "grad_norm": 1.2788035869598389,
+      "learning_rate": 1.9327906430851408e-07,
+      "loss": 0.5081,
+      "step": 12870
+    },
+    {
+      "epoch": 0.7137315748642359,
+      "grad_norm": 1.294557809829712,
+      "learning_rate": 1.925851529686407e-07,
+      "loss": 0.5033,
+      "step": 12880
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 1.3286951780319214,
+      "learning_rate": 1.9189219227390435e-07,
+      "loss": 0.5154,
+      "step": 12890
+    },
+    {
+      "epoch": 0.7148398537071927,
+      "grad_norm": 1.2231204509735107,
+      "learning_rate": 1.912001843672064e-07,
+      "loss": 0.5022,
+      "step": 12900
+    },
+    {
+      "epoch": 0.7148398537071927,
+      "eval_robovqa_loss": 0.07958105951547623,
+      "step": 12900
+    },
+    {
+      "epoch": 0.7148398537071927,
+      "eval_agibot_loss": 0.4711882770061493,
+      "step": 12900
+    },
+    {
+      "epoch": 0.7148398537071927,
+      "eval_holoassist_loss": 0.34485408663749695,
+      "step": 12900
+    },
+    {
+      "epoch": 0.7148398537071927,
+      "eval_robofail_loss": 0.20422405004501343,
+      "step": 12900
+    },
+    {
+      "epoch": 0.7153939931286711,
+      "grad_norm": 1.2375956773757935,
+      "learning_rate": 1.9050913138850193e-07,
+      "loss": 0.5157,
+      "step": 12910
+    },
+    {
+      "epoch": 0.7159481325501497,
+      "grad_norm": 1.3702161312103271,
+      "learning_rate": 1.89819035474793e-07,
+      "loss": 0.5037,
+      "step": 12920
+    },
+    {
+      "epoch": 0.7165022719716281,
+      "grad_norm": 1.4287394285202026,
+      "learning_rate": 1.8912989876012202e-07,
+      "loss": 0.5064,
+      "step": 12930
+    },
+    {
+      "epoch": 0.7170564113931065,
+      "grad_norm": 1.1849991083145142,
+      "learning_rate": 1.8844172337556513e-07,
+      "loss": 0.509,
+      "step": 12940
+    },
+    {
+      "epoch": 0.7176105508145849,
+      "grad_norm": 1.1852097511291504,
+      "learning_rate": 1.8775451144922583e-07,
+      "loss": 0.5004,
+      "step": 12950
+    },
+    {
+      "epoch": 0.7181646902360634,
+      "grad_norm": 1.2357486486434937,
+      "learning_rate": 1.8706826510622796e-07,
+      "loss": 0.5059,
+      "step": 12960
+    },
+    {
+      "epoch": 0.7187188296575419,
+      "grad_norm": 1.2942005395889282,
+      "learning_rate": 1.8638298646871003e-07,
+      "loss": 0.5068,
+      "step": 12970
+    },
+    {
+      "epoch": 0.7192729690790203,
+      "grad_norm": 1.3359532356262207,
+      "learning_rate": 1.8569867765581692e-07,
+      "loss": 0.5015,
+      "step": 12980
+    },
+    {
+      "epoch": 0.7198271085004987,
+      "grad_norm": 1.2774478197097778,
+      "learning_rate": 1.850153407836959e-07,
+      "loss": 0.5088,
+      "step": 12990
+    },
+    {
+      "epoch": 0.7203812479219772,
+      "grad_norm": 1.2135863304138184,
+      "learning_rate": 1.8433297796548713e-07,
+      "loss": 0.5134,
+      "step": 13000
+    },
+    {
+      "epoch": 0.7203812479219772,
+      "eval_robovqa_loss": 0.07999923080205917,
+      "step": 13000
+    },
+    {
+      "epoch": 0.7203812479219772,
+      "eval_agibot_loss": 0.4702279567718506,
+      "step": 13000
+    },
+    {
+      "epoch": 0.7203812479219772,
+      "eval_holoassist_loss": 0.3463514745235443,
+      "step": 13000
+    },
+    {
+      "epoch": 0.7203812479219772,
+      "eval_robofail_loss": 0.20676252245903015,
+      "step": 13000
+    },
+    {
+      "epoch": 0.7209353873434556,
+      "grad_norm": 1.2878004312515259,
+      "learning_rate": 1.8365159131131953e-07,
+      "loss": 0.4996,
+      "step": 13010
+    },
+    {
+      "epoch": 0.721489526764934,
+      "grad_norm": 1.2449074983596802,
+      "learning_rate": 1.8297118292830343e-07,
+      "loss": 0.5086,
+      "step": 13020
+    },
+    {
+      "epoch": 0.7220436661864125,
+      "grad_norm": 1.3004530668258667,
+      "learning_rate": 1.8229175492052318e-07,
+      "loss": 0.5035,
+      "step": 13030
+    },
+    {
+      "epoch": 0.722597805607891,
+      "grad_norm": 1.3262954950332642,
+      "learning_rate": 1.8161330938903234e-07,
+      "loss": 0.5029,
+      "step": 13040
+    },
+    {
+      "epoch": 0.7231519450293694,
+      "grad_norm": 1.2878361940383911,
+      "learning_rate": 1.8093584843184567e-07,
+      "loss": 0.5042,
+      "step": 13050
+    },
+    {
+      "epoch": 0.7237060844508478,
+      "grad_norm": 1.3664414882659912,
+      "learning_rate": 1.8025937414393356e-07,
+      "loss": 0.5036,
+      "step": 13060
+    },
+    {
+      "epoch": 0.7242602238723262,
+      "grad_norm": 1.342203974723816,
+      "learning_rate": 1.7958388861721495e-07,
+      "loss": 0.5076,
+      "step": 13070
+    },
+    {
+      "epoch": 0.7248143632938048,
+      "grad_norm": 1.3263356685638428,
+      "learning_rate": 1.7890939394055145e-07,
+      "loss": 0.5137,
+      "step": 13080
+    },
+    {
+      "epoch": 0.7253685027152832,
+      "grad_norm": 1.235404372215271,
+      "learning_rate": 1.782358921997404e-07,
+      "loss": 0.4972,
+      "step": 13090
+    },
+    {
+      "epoch": 0.7259226421367616,
+      "grad_norm": 1.3443561792373657,
+      "learning_rate": 1.775633854775086e-07,
+      "loss": 0.4943,
+      "step": 13100
+    },
+    {
+      "epoch": 0.7259226421367616,
+      "eval_robovqa_loss": 0.07895074039697647,
+      "step": 13100
+    },
+    {
+      "epoch": 0.7259226421367616,
+      "eval_agibot_loss": 0.4717673063278198,
+      "step": 13100
+    },
+    {
+      "epoch": 0.7259226421367616,
+      "eval_holoassist_loss": 0.3449384570121765,
+      "step": 13100
+    },
+    {
+      "epoch": 0.7259226421367616,
+      "eval_robofail_loss": 0.20671232044696808,
+      "step": 13100
+    },
+    {
+      "epoch": 0.72647678155824,
+      "grad_norm": 1.422790765762329,
+      "learning_rate": 1.7689187585350596e-07,
+      "loss": 0.5061,
+      "step": 13110
+    },
+    {
+      "epoch": 0.7270309209797184,
+      "grad_norm": 1.7094700336456299,
+      "learning_rate": 1.76221365404299e-07,
+      "loss": 0.5092,
+      "step": 13120
+    },
+    {
+      "epoch": 0.727585060401197,
+      "grad_norm": 1.218401551246643,
+      "learning_rate": 1.755518562033642e-07,
+      "loss": 0.503,
+      "step": 13130
+    },
+    {
+      "epoch": 0.7281391998226754,
+      "grad_norm": 1.2820569276809692,
+      "learning_rate": 1.748833503210818e-07,
+      "loss": 0.4995,
+      "step": 13140
+    },
+    {
+      "epoch": 0.7286933392441538,
+      "grad_norm": 1.1967058181762695,
+      "learning_rate": 1.7421584982473013e-07,
+      "loss": 0.5031,
+      "step": 13150
+    },
+    {
+      "epoch": 0.7292474786656322,
+      "grad_norm": 1.2413055896759033,
+      "learning_rate": 1.7354935677847715e-07,
+      "loss": 0.5046,
+      "step": 13160
+    },
+    {
+      "epoch": 0.7298016180871107,
+      "grad_norm": 1.242340326309204,
+      "learning_rate": 1.7288387324337662e-07,
+      "loss": 0.497,
+      "step": 13170
+    },
+    {
+      "epoch": 0.7303557575085892,
+      "grad_norm": 1.1847835779190063,
+      "learning_rate": 1.722194012773598e-07,
+      "loss": 0.5078,
+      "step": 13180
+    },
+    {
+      "epoch": 0.7309098969300676,
+      "grad_norm": 1.25177001953125,
+      "learning_rate": 1.7155594293523002e-07,
+      "loss": 0.4933,
+      "step": 13190
+    },
+    {
+      "epoch": 0.731464036351546,
+      "grad_norm": 1.2647806406021118,
+      "learning_rate": 1.7089350026865612e-07,
+      "loss": 0.5149,
+      "step": 13200
+    },
+    {
+      "epoch": 0.731464036351546,
+      "eval_robovqa_loss": 0.07882741838693619,
+      "step": 13200
+    },
+    {
+      "epoch": 0.731464036351546,
+      "eval_agibot_loss": 0.47131311893463135,
+      "step": 13200
+    },
+    {
+      "epoch": 0.731464036351546,
+      "eval_holoassist_loss": 0.34365206956863403,
+      "step": 13200
+    },
+    {
+      "epoch": 0.731464036351546,
+      "eval_robofail_loss": 0.20755870640277863,
+      "step": 13200
+    },
+    {
+      "epoch": 0.7320181757730245,
+      "grad_norm": 1.3093997240066528,
+      "learning_rate": 1.7023207532616596e-07,
+      "loss": 0.4962,
+      "step": 13210
+    },
+    {
+      "epoch": 0.7325723151945029,
+      "grad_norm": 1.3556628227233887,
+      "learning_rate": 1.6957167015314027e-07,
+      "loss": 0.5057,
+      "step": 13220
+    },
+    {
+      "epoch": 0.7331264546159814,
+      "grad_norm": 1.1387214660644531,
+      "learning_rate": 1.689122867918063e-07,
+      "loss": 0.4967,
+      "step": 13230
+    },
+    {
+      "epoch": 0.7336805940374598,
+      "grad_norm": 1.2340861558914185,
+      "learning_rate": 1.682539272812314e-07,
+      "loss": 0.508,
+      "step": 13240
+    },
+    {
+      "epoch": 0.7342347334589383,
+      "grad_norm": 1.3097808361053467,
+      "learning_rate": 1.6759659365731665e-07,
+      "loss": 0.5064,
+      "step": 13250
+    },
+    {
+      "epoch": 0.7347888728804167,
+      "grad_norm": 1.2376708984375,
+      "learning_rate": 1.669402879527914e-07,
+      "loss": 0.5015,
+      "step": 13260
+    },
+    {
+      "epoch": 0.7353430123018951,
+      "grad_norm": 1.2360093593597412,
+      "learning_rate": 1.6628501219720497e-07,
+      "loss": 0.5118,
+      "step": 13270
+    },
+    {
+      "epoch": 0.7358971517233736,
+      "grad_norm": 1.2686388492584229,
+      "learning_rate": 1.6563076841692318e-07,
+      "loss": 0.5124,
+      "step": 13280
+    },
+    {
+      "epoch": 0.7364512911448521,
+      "grad_norm": 1.2850980758666992,
+      "learning_rate": 1.6497755863511914e-07,
+      "loss": 0.5062,
+      "step": 13290
+    },
+    {
+      "epoch": 0.7370054305663305,
+      "grad_norm": 1.3701131343841553,
+      "learning_rate": 1.6432538487176962e-07,
+      "loss": 0.5077,
+      "step": 13300
+    },
+    {
+      "epoch": 0.7370054305663305,
+      "eval_robovqa_loss": 0.07887494564056396,
+      "step": 13300
+    },
+    {
+      "epoch": 0.7370054305663305,
+      "eval_agibot_loss": 0.47179874777793884,
+      "step": 13300
+    },
+    {
+      "epoch": 0.7370054305663305,
+      "eval_holoassist_loss": 0.3450358211994171,
+      "step": 13300
+    },
+    {
+      "epoch": 0.7370054305663305,
+      "eval_robofail_loss": 0.20648403465747833,
+      "step": 13300
+    },
+    {
+      "epoch": 0.7375595699878089,
+      "grad_norm": 1.20603609085083,
+      "learning_rate": 1.6367424914364704e-07,
+      "loss": 0.5107,
+      "step": 13310
+    },
+    {
+      "epoch": 0.7381137094092873,
+      "grad_norm": 1.2466236352920532,
+      "learning_rate": 1.6302415346431386e-07,
+      "loss": 0.5091,
+      "step": 13320
+    },
+    {
+      "epoch": 0.7386678488307659,
+      "grad_norm": 1.2673234939575195,
+      "learning_rate": 1.6237509984411636e-07,
+      "loss": 0.4981,
+      "step": 13330
+    },
+    {
+      "epoch": 0.7392219882522443,
+      "grad_norm": 1.3247064352035522,
+      "learning_rate": 1.6172709029017844e-07,
+      "loss": 0.5083,
+      "step": 13340
+    },
+    {
+      "epoch": 0.7397761276737227,
+      "grad_norm": 1.3026233911514282,
+      "learning_rate": 1.610801268063952e-07,
+      "loss": 0.4987,
+      "step": 13350
+    },
+    {
+      "epoch": 0.7403302670952011,
+      "grad_norm": 1.2993193864822388,
+      "learning_rate": 1.604342113934271e-07,
+      "loss": 0.5064,
+      "step": 13360
+    },
+    {
+      "epoch": 0.7408844065166796,
+      "grad_norm": 1.3599908351898193,
+      "learning_rate": 1.5978934604869326e-07,
+      "loss": 0.5061,
+      "step": 13370
+    },
+    {
+      "epoch": 0.7414385459381581,
+      "grad_norm": 1.283867359161377,
+      "learning_rate": 1.5914553276636578e-07,
+      "loss": 0.503,
+      "step": 13380
+    },
+    {
+      "epoch": 0.7419926853596365,
+      "grad_norm": 1.2798441648483276,
+      "learning_rate": 1.5850277353736368e-07,
+      "loss": 0.5034,
+      "step": 13390
+    },
+    {
+      "epoch": 0.7425468247811149,
+      "grad_norm": 1.3285945653915405,
+      "learning_rate": 1.5786107034934553e-07,
+      "loss": 0.5098,
+      "step": 13400
+    },
+    {
+      "epoch": 0.7425468247811149,
+      "eval_robovqa_loss": 0.07909495383501053,
+      "step": 13400
+    },
+    {
+      "epoch": 0.7425468247811149,
+      "eval_agibot_loss": 0.468485951423645,
+      "step": 13400
+    },
+    {
+      "epoch": 0.7425468247811149,
+      "eval_holoassist_loss": 0.3339272439479828,
+      "step": 13400
+    },
+    {
+      "epoch": 0.7425468247811149,
+      "eval_robofail_loss": 0.20811797678470612,
+      "step": 13400
+    },
+    {
+      "epoch": 0.7431009642025934,
+      "grad_norm": 1.201356053352356,
+      "learning_rate": 1.5722042518670547e-07,
+      "loss": 0.5039,
+      "step": 13410
+    },
+    {
+      "epoch": 0.7436551036240718,
+      "grad_norm": 1.349993109703064,
+      "learning_rate": 1.565808400305645e-07,
+      "loss": 0.5044,
+      "step": 13420
+    },
+    {
+      "epoch": 0.7442092430455503,
+      "grad_norm": 1.3256423473358154,
+      "learning_rate": 1.559423168587668e-07,
+      "loss": 0.4919,
+      "step": 13430
+    },
+    {
+      "epoch": 0.7447633824670287,
+      "grad_norm": 1.2907826900482178,
+      "learning_rate": 1.5530485764587198e-07,
+      "loss": 0.5096,
+      "step": 13440
+    },
+    {
+      "epoch": 0.7453175218885072,
+      "grad_norm": 1.272213339805603,
+      "learning_rate": 1.54668464363149e-07,
+      "loss": 0.5003,
+      "step": 13450
+    },
+    {
+      "epoch": 0.7458716613099856,
+      "grad_norm": 1.3435653448104858,
+      "learning_rate": 1.5403313897857162e-07,
+      "loss": 0.5058,
+      "step": 13460
+    },
+    {
+      "epoch": 0.746425800731464,
+      "grad_norm": 1.2408884763717651,
+      "learning_rate": 1.5339888345681046e-07,
+      "loss": 0.5062,
+      "step": 13470
+    },
+    {
+      "epoch": 0.7469799401529424,
+      "grad_norm": 1.291700839996338,
+      "learning_rate": 1.5276569975922792e-07,
+      "loss": 0.4967,
+      "step": 13480
+    },
+    {
+      "epoch": 0.747534079574421,
+      "grad_norm": 1.2178364992141724,
+      "learning_rate": 1.5213358984387197e-07,
+      "loss": 0.5001,
+      "step": 13490
+    },
+    {
+      "epoch": 0.7480882189958994,
+      "grad_norm": 1.2450575828552246,
+      "learning_rate": 1.5150255566547004e-07,
+      "loss": 0.5112,
+      "step": 13500
+    },
+    {
+      "epoch": 0.7480882189958994,
+      "eval_robovqa_loss": 0.08054637908935547,
+      "step": 13500
+    },
+    {
+      "epoch": 0.7480882189958994,
+      "eval_agibot_loss": 0.4682695269584656,
+      "step": 13500
+    },
+    {
+      "epoch": 0.7480882189958994,
+      "eval_holoassist_loss": 0.33922889828681946,
+      "step": 13500
+    },
+    {
+      "epoch": 0.7480882189958994,
+      "eval_robofail_loss": 0.20658965408802032,
+      "step": 13500
+    },
+    {
+      "epoch": 0.7486423584173778,
+      "grad_norm": 1.261865258216858,
+      "learning_rate": 1.5087259917542278e-07,
+      "loss": 0.495,
+      "step": 13510
+    },
+    {
+      "epoch": 0.7491964978388562,
+      "grad_norm": 1.2658618688583374,
+      "learning_rate": 1.5024372232179843e-07,
+      "loss": 0.5023,
+      "step": 13520
+    },
+    {
+      "epoch": 0.7497506372603348,
+      "grad_norm": 1.2656056880950928,
+      "learning_rate": 1.4961592704932647e-07,
+      "loss": 0.5061,
+      "step": 13530
+    },
+    {
+      "epoch": 0.7503047766818132,
+      "grad_norm": 1.2357923984527588,
+      "learning_rate": 1.4898921529939173e-07,
+      "loss": 0.5098,
+      "step": 13540
+    },
+    {
+      "epoch": 0.7508589161032916,
+      "grad_norm": 1.2018358707427979,
+      "learning_rate": 1.4836358901002843e-07,
+      "loss": 0.5037,
+      "step": 13550
+    },
+    {
+      "epoch": 0.75141305552477,
+      "grad_norm": 1.2614952325820923,
+      "learning_rate": 1.4773905011591386e-07,
+      "loss": 0.4946,
+      "step": 13560
+    },
+    {
+      "epoch": 0.7519671949462485,
+      "grad_norm": 1.1827198266983032,
+      "learning_rate": 1.4711560054836337e-07,
+      "loss": 0.5099,
+      "step": 13570
+    },
+    {
+      "epoch": 0.752521334367727,
+      "grad_norm": 1.305364727973938,
+      "learning_rate": 1.4649324223532255e-07,
+      "loss": 0.5043,
+      "step": 13580
+    },
+    {
+      "epoch": 0.7530754737892054,
+      "grad_norm": 1.2731246948242188,
+      "learning_rate": 1.4587197710136363e-07,
+      "loss": 0.5011,
+      "step": 13590
+    },
+    {
+      "epoch": 0.7536296132106838,
+      "grad_norm": 1.1577284336090088,
+      "learning_rate": 1.4525180706767748e-07,
+      "loss": 0.5012,
+      "step": 13600
+    },
+    {
+      "epoch": 0.7536296132106838,
+      "eval_robovqa_loss": 0.08010783791542053,
+      "step": 13600
+    },
+    {
+      "epoch": 0.7536296132106838,
+      "eval_agibot_loss": 0.4657536745071411,
+      "step": 13600
+    },
+    {
+      "epoch": 0.7536296132106838,
+      "eval_holoassist_loss": 0.34381595253944397,
+      "step": 13600
+    },
+    {
+      "epoch": 0.7536296132106838,
+      "eval_robofail_loss": 0.20509548485279083,
+      "step": 13600
+    },
+    {
+      "epoch": 0.7541837526321623,
+      "grad_norm": 1.2127923965454102,
+      "learning_rate": 1.4463273405206877e-07,
+      "loss": 0.507,
+      "step": 13610
+    },
+    {
+      "epoch": 0.7547378920536407,
+      "grad_norm": 1.3108220100402832,
+      "learning_rate": 1.4401475996894985e-07,
+      "loss": 0.505,
+      "step": 13620
+    },
+    {
+      "epoch": 0.7552920314751191,
+      "grad_norm": 1.1869585514068604,
+      "learning_rate": 1.4339788672933455e-07,
+      "loss": 0.5016,
+      "step": 13630
+    },
+    {
+      "epoch": 0.7558461708965976,
+      "grad_norm": 1.2783479690551758,
+      "learning_rate": 1.4278211624083263e-07,
+      "loss": 0.5068,
+      "step": 13640
+    },
+    {
+      "epoch": 0.756400310318076,
+      "grad_norm": 1.256670594215393,
+      "learning_rate": 1.421674504076436e-07,
+      "loss": 0.5051,
+      "step": 13650
+    },
+    {
+      "epoch": 0.7569544497395545,
+      "grad_norm": 1.2494417428970337,
+      "learning_rate": 1.4155389113055115e-07,
+      "loss": 0.5092,
+      "step": 13660
+    },
+    {
+      "epoch": 0.7575085891610329,
+      "grad_norm": 1.3013426065444946,
+      "learning_rate": 1.4094144030691667e-07,
+      "loss": 0.5002,
+      "step": 13670
+    },
+    {
+      "epoch": 0.7580627285825113,
+      "grad_norm": 1.3104485273361206,
+      "learning_rate": 1.4033009983067452e-07,
+      "loss": 0.511,
+      "step": 13680
+    },
+    {
+      "epoch": 0.7586168680039898,
+      "grad_norm": 1.2376840114593506,
+      "learning_rate": 1.397198715923244e-07,
+      "loss": 0.5043,
+      "step": 13690
+    },
+    {
+      "epoch": 0.7591710074254683,
+      "grad_norm": 1.306136131286621,
+      "learning_rate": 1.3911075747892765e-07,
+      "loss": 0.5064,
+      "step": 13700
+    },
+    {
+      "epoch": 0.7591710074254683,
+      "eval_robovqa_loss": 0.08139156550168991,
+      "step": 13700
+    },
+    {
+      "epoch": 0.7591710074254683,
+      "eval_agibot_loss": 0.4662005305290222,
+      "step": 13700
+    },
+    {
+      "epoch": 0.7591710074254683,
+      "eval_holoassist_loss": 0.34228360652923584,
+      "step": 13700
+    },
+    {
+      "epoch": 0.7591710074254683,
+      "eval_robofail_loss": 0.2066621631383896,
+      "step": 13700
+    },
+    {
+      "epoch": 0.7597251468469467,
+      "grad_norm": 1.2643458843231201,
+      "learning_rate": 1.3850275937409923e-07,
+      "loss": 0.5068,
+      "step": 13710
+    },
+    {
+      "epoch": 0.7602792862684251,
+      "grad_norm": 1.3452255725860596,
+      "learning_rate": 1.3789587915800393e-07,
+      "loss": 0.5029,
+      "step": 13720
+    },
+    {
+      "epoch": 0.7608334256899035,
+      "grad_norm": 1.3727076053619385,
+      "learning_rate": 1.3729011870734907e-07,
+      "loss": 0.4996,
+      "step": 13730
+    },
+    {
+      "epoch": 0.7613875651113821,
+      "grad_norm": 1.8032333850860596,
+      "learning_rate": 1.3668547989537926e-07,
+      "loss": 0.5016,
+      "step": 13740
+    },
+    {
+      "epoch": 0.7619417045328605,
+      "grad_norm": 1.2953962087631226,
+      "learning_rate": 1.3608196459187067e-07,
+      "loss": 0.4946,
+      "step": 13750
+    },
+    {
+      "epoch": 0.7624958439543389,
+      "grad_norm": 1.278609037399292,
+      "learning_rate": 1.3547957466312508e-07,
+      "loss": 0.502,
+      "step": 13760
+    },
+    {
+      "epoch": 0.7630499833758173,
+      "grad_norm": 1.348230242729187,
+      "learning_rate": 1.3487831197196431e-07,
+      "loss": 0.5013,
+      "step": 13770
+    },
+    {
+      "epoch": 0.7636041227972958,
+      "grad_norm": 1.3317800760269165,
+      "learning_rate": 1.3427817837772414e-07,
+      "loss": 0.4983,
+      "step": 13780
+    },
+    {
+      "epoch": 0.7641582622187743,
+      "grad_norm": 1.3179935216903687,
+      "learning_rate": 1.3367917573624894e-07,
+      "loss": 0.4947,
+      "step": 13790
+    },
+    {
+      "epoch": 0.7647124016402527,
+      "grad_norm": 1.2974077463150024,
+      "learning_rate": 1.3308130589988541e-07,
+      "loss": 0.5026,
+      "step": 13800
+    },
+    {
+      "epoch": 0.7647124016402527,
+      "eval_robovqa_loss": 0.07974833250045776,
+      "step": 13800
+    },
+    {
+      "epoch": 0.7647124016402527,
+      "eval_agibot_loss": 0.46453380584716797,
+      "step": 13800
+    },
+    {
+      "epoch": 0.7647124016402527,
+      "eval_holoassist_loss": 0.34514328837394714,
+      "step": 13800
+    },
+    {
+      "epoch": 0.7647124016402527,
+      "eval_robofail_loss": 0.2072557955980301,
+      "step": 13800
+    },
+    {
+      "epoch": 0.7652665410617311,
+      "grad_norm": 1.2241129875183105,
+      "learning_rate": 1.3248457071747787e-07,
+      "loss": 0.5067,
+      "step": 13810
+    },
+    {
+      "epoch": 0.7658206804832096,
+      "grad_norm": 1.2652555704116821,
+      "learning_rate": 1.3188897203436085e-07,
+      "loss": 0.4938,
+      "step": 13820
+    },
+    {
+      "epoch": 0.766374819904688,
+      "grad_norm": 1.244423270225525,
+      "learning_rate": 1.312945116923554e-07,
+      "loss": 0.505,
+      "step": 13830
+    },
+    {
+      "epoch": 0.7669289593261664,
+      "grad_norm": 1.2688522338867188,
+      "learning_rate": 1.3070119152976144e-07,
+      "loss": 0.5068,
+      "step": 13840
+    },
+    {
+      "epoch": 0.7674830987476449,
+      "grad_norm": 1.353855013847351,
+      "learning_rate": 1.3010901338135394e-07,
+      "loss": 0.5117,
+      "step": 13850
+    },
+    {
+      "epoch": 0.7680372381691234,
+      "grad_norm": 1.2045435905456543,
+      "learning_rate": 1.2951797907837576e-07,
+      "loss": 0.4977,
+      "step": 13860
+    },
+    {
+      "epoch": 0.7685913775906018,
+      "grad_norm": 1.1978533267974854,
+      "learning_rate": 1.289280904485323e-07,
+      "loss": 0.5046,
+      "step": 13870
+    },
+    {
+      "epoch": 0.7691455170120802,
+      "grad_norm": 1.3398493528366089,
+      "learning_rate": 1.2833934931598688e-07,
+      "loss": 0.503,
+      "step": 13880
+    },
+    {
+      "epoch": 0.7696996564335586,
+      "grad_norm": 1.4586983919143677,
+      "learning_rate": 1.277517575013537e-07,
+      "loss": 0.5035,
+      "step": 13890
+    },
+    {
+      "epoch": 0.7702537958550372,
+      "grad_norm": 1.2211514711380005,
+      "learning_rate": 1.2716531682169308e-07,
+      "loss": 0.5115,
+      "step": 13900
+    },
+    {
+      "epoch": 0.7702537958550372,
+      "eval_robovqa_loss": 0.08144582062959671,
+      "step": 13900
+    },
+    {
+      "epoch": 0.7702537958550372,
+      "eval_agibot_loss": 0.4634343981742859,
+      "step": 13900
+    },
+    {
+      "epoch": 0.7702537958550372,
+      "eval_holoassist_loss": 0.34193772077560425,
+      "step": 13900
+    },
+    {
+      "epoch": 0.7702537958550372,
+      "eval_robofail_loss": 0.20634889602661133,
+      "step": 13900
+    },
+    {
+      "epoch": 0.7708079352765156,
+      "grad_norm": 1.3954793214797974,
+      "learning_rate": 1.2658002909050548e-07,
+      "loss": 0.4978,
+      "step": 13910
+    },
+    {
+      "epoch": 0.771362074697994,
+      "grad_norm": 1.1939213275909424,
+      "learning_rate": 1.259958961177261e-07,
+      "loss": 0.4997,
+      "step": 13920
+    },
+    {
+      "epoch": 0.7719162141194724,
+      "grad_norm": 1.2899671792984009,
+      "learning_rate": 1.254129197097189e-07,
+      "loss": 0.4989,
+      "step": 13930
+    },
+    {
+      "epoch": 0.772470353540951,
+      "grad_norm": 1.2517800331115723,
+      "learning_rate": 1.2483110166927203e-07,
+      "loss": 0.5029,
+      "step": 13940
+    },
+    {
+      "epoch": 0.7730244929624294,
+      "grad_norm": 1.3517178297042847,
+      "learning_rate": 1.2425044379559053e-07,
+      "loss": 0.5022,
+      "step": 13950
+    },
+    {
+      "epoch": 0.7735786323839078,
+      "grad_norm": 1.2301511764526367,
+      "learning_rate": 1.2367094788429227e-07,
+      "loss": 0.5052,
+      "step": 13960
+    },
+    {
+      "epoch": 0.7741327718053862,
+      "grad_norm": 1.3182294368743896,
+      "learning_rate": 1.2309261572740226e-07,
+      "loss": 0.4928,
+      "step": 13970
+    },
+    {
+      "epoch": 0.7746869112268647,
+      "grad_norm": 1.3121981620788574,
+      "learning_rate": 1.2251544911334582e-07,
+      "loss": 0.4967,
+      "step": 13980
+    },
+    {
+      "epoch": 0.7752410506483431,
+      "grad_norm": 1.256043791770935,
+      "learning_rate": 1.2193944982694493e-07,
+      "loss": 0.5035,
+      "step": 13990
+    },
+    {
+      "epoch": 0.7757951900698216,
+      "grad_norm": 1.287541389465332,
+      "learning_rate": 1.213646196494108e-07,
+      "loss": 0.4981,
+      "step": 14000
+    },
+    {
+      "epoch": 0.7757951900698216,
+      "eval_robovqa_loss": 0.08003152161836624,
+      "step": 14000
+    },
+    {
+      "epoch": 0.7757951900698216,
+      "eval_agibot_loss": 0.46629753708839417,
+      "step": 14000
+    },
+    {
+      "epoch": 0.7757951900698216,
+      "eval_holoassist_loss": 0.3387264907360077,
+      "step": 14000
+    },
+    {
+      "epoch": 0.7757951900698216,
+      "eval_robofail_loss": 0.2035808116197586,
+      "step": 14000
+    },
+    {
+      "epoch": 0.7763493294913,
+      "grad_norm": 1.342185378074646,
+      "learning_rate": 1.2079096035834025e-07,
+      "loss": 0.4956,
+      "step": 14010
+    },
+    {
+      "epoch": 0.7769034689127785,
+      "grad_norm": 1.2186702489852905,
+      "learning_rate": 1.2021847372770855e-07,
+      "loss": 0.499,
+      "step": 14020
+    },
+    {
+      "epoch": 0.7774576083342569,
+      "grad_norm": 1.3145819902420044,
+      "learning_rate": 1.1964716152786496e-07,
+      "loss": 0.5047,
+      "step": 14030
+    },
+    {
+      "epoch": 0.7780117477557353,
+      "grad_norm": 1.3532967567443848,
+      "learning_rate": 1.1907702552552695e-07,
+      "loss": 0.5001,
+      "step": 14040
+    },
+    {
+      "epoch": 0.7785658871772138,
+      "grad_norm": 1.3208905458450317,
+      "learning_rate": 1.1850806748377467e-07,
+      "loss": 0.4995,
+      "step": 14050
+    },
+    {
+      "epoch": 0.7791200265986923,
+      "grad_norm": 1.1371922492980957,
+      "learning_rate": 1.1794028916204546e-07,
+      "loss": 0.5068,
+      "step": 14060
+    },
+    {
+      "epoch": 0.7796741660201707,
+      "grad_norm": 1.2505205869674683,
+      "learning_rate": 1.1737369231612876e-07,
+      "loss": 0.4952,
+      "step": 14070
+    },
+    {
+      "epoch": 0.7802283054416491,
+      "grad_norm": 1.3631892204284668,
+      "learning_rate": 1.1680827869816024e-07,
+      "loss": 0.5023,
+      "step": 14080
+    },
+    {
+      "epoch": 0.7807824448631275,
+      "grad_norm": 1.198555588722229,
+      "learning_rate": 1.1624405005661648e-07,
+      "loss": 0.5036,
+      "step": 14090
+    },
+    {
+      "epoch": 0.7813365842846061,
+      "grad_norm": 1.1785237789154053,
+      "learning_rate": 1.1568100813631021e-07,
+      "loss": 0.506,
+      "step": 14100
+    },
+    {
+      "epoch": 0.7813365842846061,
+      "eval_robovqa_loss": 0.0807831659913063,
+      "step": 14100
+    },
+    {
+      "epoch": 0.7813365842846061,
+      "eval_agibot_loss": 0.46831321716308594,
+      "step": 14100
+    },
+    {
+      "epoch": 0.7813365842846061,
+      "eval_holoassist_loss": 0.3384007215499878,
+      "step": 14100
+    },
+    {
+      "epoch": 0.7813365842846061,
+      "eval_robofail_loss": 0.20681263506412506,
+      "step": 14100
+    },
+    {
+      "epoch": 0.7818907237060845,
+      "grad_norm": 1.2407755851745605,
+      "learning_rate": 1.1511915467838346e-07,
+      "loss": 0.5026,
+      "step": 14110
+    },
+    {
+      "epoch": 0.7824448631275629,
+      "grad_norm": 1.275132417678833,
+      "learning_rate": 1.1455849142030411e-07,
+      "loss": 0.5145,
+      "step": 14120
+    },
+    {
+      "epoch": 0.7829990025490413,
+      "grad_norm": 1.2104378938674927,
+      "learning_rate": 1.1399902009585833e-07,
+      "loss": 0.4907,
+      "step": 14130
+    },
+    {
+      "epoch": 0.7835531419705197,
+      "grad_norm": 1.2846125364303589,
+      "learning_rate": 1.1344074243514751e-07,
+      "loss": 0.4985,
+      "step": 14140
+    },
+    {
+      "epoch": 0.7841072813919983,
+      "grad_norm": 1.3293678760528564,
+      "learning_rate": 1.1288366016458101e-07,
+      "loss": 0.511,
+      "step": 14150
+    },
+    {
+      "epoch": 0.7846614208134767,
+      "grad_norm": 1.3414421081542969,
+      "learning_rate": 1.1232777500687185e-07,
+      "loss": 0.5071,
+      "step": 14160
+    },
+    {
+      "epoch": 0.7852155602349551,
+      "grad_norm": 1.2684837579727173,
+      "learning_rate": 1.1177308868103113e-07,
+      "loss": 0.5059,
+      "step": 14170
+    },
+    {
+      "epoch": 0.7857696996564335,
+      "grad_norm": 1.3259629011154175,
+      "learning_rate": 1.1121960290236265e-07,
+      "loss": 0.496,
+      "step": 14180
+    },
+    {
+      "epoch": 0.786323839077912,
+      "grad_norm": 1.334373950958252,
+      "learning_rate": 1.1066731938245766e-07,
+      "loss": 0.511,
+      "step": 14190
+    },
+    {
+      "epoch": 0.7868779784993905,
+      "grad_norm": 1.3560504913330078,
+      "learning_rate": 1.101162398291895e-07,
+      "loss": 0.5072,
+      "step": 14200
+    },
+    {
+      "epoch": 0.7868779784993905,
+      "eval_robovqa_loss": 0.07974942028522491,
+      "step": 14200
+    },
+    {
+      "epoch": 0.7868779784993905,
+      "eval_agibot_loss": 0.47056543827056885,
+      "step": 14200
+    },
+    {
+      "epoch": 0.7868779784993905,
+      "eval_holoassist_loss": 0.3427981436252594,
+      "step": 14200
+    },
+    {
+      "epoch": 0.7868779784993905,
+      "eval_robofail_loss": 0.20664264261722565,
+      "step": 14200
+    },
+    {
+      "epoch": 0.7874321179208689,
+      "grad_norm": 1.286238431930542,
+      "learning_rate": 1.0956636594670853e-07,
+      "loss": 0.5123,
+      "step": 14210
+    },
+    {
+      "epoch": 0.7879862573423473,
+      "grad_norm": 1.3134667873382568,
+      "learning_rate": 1.0901769943543654e-07,
+      "loss": 0.4983,
+      "step": 14220
+    },
+    {
+      "epoch": 0.7885403967638258,
+      "grad_norm": 1.3173279762268066,
+      "learning_rate": 1.08470241992062e-07,
+      "loss": 0.5023,
+      "step": 14230
+    },
+    {
+      "epoch": 0.7890945361853042,
+      "grad_norm": 1.8566786050796509,
+      "learning_rate": 1.0792399530953383e-07,
+      "loss": 0.5078,
+      "step": 14240
+    },
+    {
+      "epoch": 0.7896486756067826,
+      "grad_norm": 1.2928568124771118,
+      "learning_rate": 1.0737896107705757e-07,
+      "loss": 0.5033,
+      "step": 14250
+    },
+    {
+      "epoch": 0.7902028150282611,
+      "grad_norm": 1.355677843093872,
+      "learning_rate": 1.0683514098008894e-07,
+      "loss": 0.5044,
+      "step": 14260
+    },
+    {
+      "epoch": 0.7907569544497396,
+      "grad_norm": 1.452815055847168,
+      "learning_rate": 1.0629253670032911e-07,
+      "loss": 0.506,
+      "step": 14270
+    },
+    {
+      "epoch": 0.791311093871218,
+      "grad_norm": 1.3867067098617554,
+      "learning_rate": 1.0575114991571954e-07,
+      "loss": 0.5105,
+      "step": 14280
+    },
+    {
+      "epoch": 0.7918652332926964,
+      "grad_norm": 1.5770423412322998,
+      "learning_rate": 1.0521098230043684e-07,
+      "loss": 0.5007,
+      "step": 14290
+    },
+    {
+      "epoch": 0.7924193727141748,
+      "grad_norm": 1.3939576148986816,
+      "learning_rate": 1.0467203552488729e-07,
+      "loss": 0.5048,
+      "step": 14300
+    },
+    {
+      "epoch": 0.7924193727141748,
+      "eval_robovqa_loss": 0.08056650310754776,
+      "step": 14300
+    },
+    {
+      "epoch": 0.7924193727141748,
+      "eval_agibot_loss": 0.46663153171539307,
+      "step": 14300
+    },
+    {
+      "epoch": 0.7924193727141748,
+      "eval_holoassist_loss": 0.34411799907684326,
+      "step": 14300
+    },
+    {
+      "epoch": 0.7924193727141748,
+      "eval_robofail_loss": 0.2068299502134323,
+      "step": 14300
+    },
+    {
+      "epoch": 0.7929735121356534,
+      "grad_norm": 1.3456461429595947,
+      "learning_rate": 1.0413431125570199e-07,
+      "loss": 0.4943,
+      "step": 14310
+    },
+    {
+      "epoch": 0.7935276515571318,
+      "grad_norm": 1.416671872138977,
+      "learning_rate": 1.0359781115573147e-07,
+      "loss": 0.5049,
+      "step": 14320
+    },
+    {
+      "epoch": 0.7940817909786102,
+      "grad_norm": 1.4450587034225464,
+      "learning_rate": 1.0306253688404082e-07,
+      "loss": 0.505,
+      "step": 14330
+    },
+    {
+      "epoch": 0.7946359304000886,
+      "grad_norm": 1.4007171392440796,
+      "learning_rate": 1.025284900959042e-07,
+      "loss": 0.499,
+      "step": 14340
+    },
+    {
+      "epoch": 0.7951900698215671,
+      "grad_norm": 1.371230125427246,
+      "learning_rate": 1.0199567244279994e-07,
+      "loss": 0.5063,
+      "step": 14350
+    },
+    {
+      "epoch": 0.7957442092430456,
+      "grad_norm": 1.200240135192871,
+      "learning_rate": 1.0146408557240593e-07,
+      "loss": 0.5069,
+      "step": 14360
+    },
+    {
+      "epoch": 0.796298348664524,
+      "grad_norm": 1.2665449380874634,
+      "learning_rate": 1.0093373112859305e-07,
+      "loss": 0.4986,
+      "step": 14370
+    },
+    {
+      "epoch": 0.7968524880860024,
+      "grad_norm": 1.2393771409988403,
+      "learning_rate": 1.0040461075142165e-07,
+      "loss": 0.4997,
+      "step": 14380
+    },
+    {
+      "epoch": 0.7974066275074809,
+      "grad_norm": 1.3559898138046265,
+      "learning_rate": 9.987672607713615e-08,
+      "loss": 0.5094,
+      "step": 14390
+    },
+    {
+      "epoch": 0.7979607669289593,
+      "grad_norm": 1.2005971670150757,
+      "learning_rate": 9.935007873815882e-08,
+      "loss": 0.4932,
+      "step": 14400
+    },
+    {
+      "epoch": 0.7979607669289593,
+      "eval_robovqa_loss": 0.08066917210817337,
+      "step": 14400
+    },
+    {
+      "epoch": 0.7979607669289593,
+      "eval_agibot_loss": 0.4637383222579956,
+      "step": 14400
+    },
+    {
+      "epoch": 0.7979607669289593,
+      "eval_holoassist_loss": 0.34583958983421326,
+      "step": 14400
+    },
+    {
+      "epoch": 0.7979607669289593,
+      "eval_robofail_loss": 0.20523740351200104,
+      "step": 14400
+    },
+    {
+      "epoch": 0.7985149063504378,
+      "grad_norm": 1.3395006656646729,
+      "learning_rate": 9.882467036308661e-08,
+      "loss": 0.4964,
+      "step": 14410
+    },
+    {
+      "epoch": 0.7990690457719162,
+      "grad_norm": 1.279321551322937,
+      "learning_rate": 9.83005025766841e-08,
+      "loss": 0.5034,
+      "step": 14420
+    },
+    {
+      "epoch": 0.7996231851933947,
+      "grad_norm": 1.330895185470581,
+      "learning_rate": 9.777757699988043e-08,
+      "loss": 0.5036,
+      "step": 14430
+    },
+    {
+      "epoch": 0.8001773246148731,
+      "grad_norm": 1.2493219375610352,
+      "learning_rate": 9.72558952497628e-08,
+      "loss": 0.4953,
+      "step": 14440
+    },
+    {
+      "epoch": 0.8007314640363515,
+      "grad_norm": 1.1940813064575195,
+      "learning_rate": 9.673545893957219e-08,
+      "loss": 0.4969,
+      "step": 14450
+    },
+    {
+      "epoch": 0.80128560345783,
+      "grad_norm": 1.4241101741790771,
+      "learning_rate": 9.621626967869806e-08,
+      "loss": 0.499,
+      "step": 14460
+    },
+    {
+      "epoch": 0.8018397428793085,
+      "grad_norm": 1.2931216955184937,
+      "learning_rate": 9.569832907267378e-08,
+      "loss": 0.5026,
+      "step": 14470
+    },
+    {
+      "epoch": 0.8023938823007869,
+      "grad_norm": 1.2512246370315552,
+      "learning_rate": 9.518163872317109e-08,
+      "loss": 0.5023,
+      "step": 14480
+    },
+    {
+      "epoch": 0.8029480217222653,
+      "grad_norm": 1.3005505800247192,
+      "learning_rate": 9.46662002279957e-08,
+      "loss": 0.5145,
+      "step": 14490
+    },
+    {
+      "epoch": 0.8035021611437437,
+      "grad_norm": 1.2436476945877075,
+      "learning_rate": 9.415201518108201e-08,
+      "loss": 0.509,
+      "step": 14500
+    },
+    {
+      "epoch": 0.8035021611437437,
+      "eval_robovqa_loss": 0.08058148622512817,
+      "step": 14500
+    },
+    {
+      "epoch": 0.8035021611437437,
+      "eval_agibot_loss": 0.46685054898262024,
+      "step": 14500
+    },
+    {
+      "epoch": 0.8035021611437437,
+      "eval_holoassist_loss": 0.34587132930755615,
+      "step": 14500
+    },
+    {
+      "epoch": 0.8035021611437437,
+      "eval_robofail_loss": 0.20620864629745483,
+      "step": 14500
+    },
+    {
+      "epoch": 0.8040563005652223,
+      "grad_norm": 1.3661483526229858,
+      "learning_rate": 9.363908517248803e-08,
+      "loss": 0.5048,
+      "step": 14510
+    },
+    {
+      "epoch": 0.8046104399867007,
+      "grad_norm": 1.3581483364105225,
+      "learning_rate": 9.312741178839145e-08,
+      "loss": 0.507,
+      "step": 14520
+    },
+    {
+      "epoch": 0.8051645794081791,
+      "grad_norm": 1.3007433414459229,
+      "learning_rate": 9.261699661108296e-08,
+      "loss": 0.4986,
+      "step": 14530
+    },
+    {
+      "epoch": 0.8057187188296575,
+      "grad_norm": 1.2804155349731445,
+      "learning_rate": 9.210784121896343e-08,
+      "loss": 0.4958,
+      "step": 14540
+    },
+    {
+      "epoch": 0.806272858251136,
+      "grad_norm": 1.3894696235656738,
+      "learning_rate": 9.159994718653691e-08,
+      "loss": 0.5023,
+      "step": 14550
+    },
+    {
+      "epoch": 0.8068269976726145,
+      "grad_norm": 1.2251273393630981,
+      "learning_rate": 9.109331608440784e-08,
+      "loss": 0.5117,
+      "step": 14560
+    },
+    {
+      "epoch": 0.8073811370940929,
+      "grad_norm": 1.3747328519821167,
+      "learning_rate": 9.058794947927467e-08,
+      "loss": 0.5026,
+      "step": 14570
+    },
+    {
+      "epoch": 0.8079352765155713,
+      "grad_norm": 1.286533236503601,
+      "learning_rate": 9.008384893392562e-08,
+      "loss": 0.5028,
+      "step": 14580
+    },
+    {
+      "epoch": 0.8084894159370498,
+      "grad_norm": 1.2131770849227905,
+      "learning_rate": 8.958101600723377e-08,
+      "loss": 0.5022,
+      "step": 14590
+    },
+    {
+      "epoch": 0.8090435553585282,
+      "grad_norm": 1.3336529731750488,
+      "learning_rate": 8.90794522541522e-08,
+      "loss": 0.5032,
+      "step": 14600
+    },
+    {
+      "epoch": 0.8090435553585282,
+      "eval_robovqa_loss": 0.07959491014480591,
+      "step": 14600
+    },
+    {
+      "epoch": 0.8090435553585282,
+      "eval_agibot_loss": 0.46647346019744873,
+      "step": 14600
+    },
+    {
+      "epoch": 0.8090435553585282,
+      "eval_holoassist_loss": 0.3430459499359131,
+      "step": 14600
+    },
+    {
+      "epoch": 0.8090435553585282,
+      "eval_robofail_loss": 0.20666061341762543,
+      "step": 14600
+    },
+    {
+      "epoch": 0.8095976947800066,
+      "grad_norm": 1.3067841529846191,
+      "learning_rate": 8.85791592257093e-08,
+      "loss": 0.4974,
+      "step": 14610
+    },
+    {
+      "epoch": 0.8101518342014851,
+      "grad_norm": 1.3319242000579834,
+      "learning_rate": 8.808013846900387e-08,
+      "loss": 0.5101,
+      "step": 14620
+    },
+    {
+      "epoch": 0.8107059736229635,
+      "grad_norm": 1.320974588394165,
+      "learning_rate": 8.758239152720027e-08,
+      "loss": 0.4994,
+      "step": 14630
+    },
+    {
+      "epoch": 0.811260113044442,
+      "grad_norm": 1.310084581375122,
+      "learning_rate": 8.708591993952374e-08,
+      "loss": 0.5018,
+      "step": 14640
+    },
+    {
+      "epoch": 0.8118142524659204,
+      "grad_norm": 1.309897780418396,
+      "learning_rate": 8.659072524125605e-08,
+      "loss": 0.505,
+      "step": 14650
+    },
+    {
+      "epoch": 0.8123683918873988,
+      "grad_norm": 1.3762198686599731,
+      "learning_rate": 8.609680896372946e-08,
+      "loss": 0.5009,
+      "step": 14660
+    },
+    {
+      "epoch": 0.8129225313088773,
+      "grad_norm": 1.2425806522369385,
+      "learning_rate": 8.560417263432374e-08,
+      "loss": 0.4962,
+      "step": 14670
+    },
+    {
+      "epoch": 0.8134766707303558,
+      "grad_norm": 1.228331208229065,
+      "learning_rate": 8.511281777646007e-08,
+      "loss": 0.4977,
+      "step": 14680
+    },
+    {
+      "epoch": 0.8140308101518342,
+      "grad_norm": 1.300062656402588,
+      "learning_rate": 8.462274590959705e-08,
+      "loss": 0.5002,
+      "step": 14690
+    },
+    {
+      "epoch": 0.8145849495733126,
+      "grad_norm": 1.565882682800293,
+      "learning_rate": 8.413395854922561e-08,
+      "loss": 0.5078,
+      "step": 14700
+    },
+    {
+      "epoch": 0.8145849495733126,
+      "eval_robovqa_loss": 0.07881201058626175,
+      "step": 14700
+    },
+    {
+      "epoch": 0.8145849495733126,
+      "eval_agibot_loss": 0.46805161237716675,
+      "step": 14700
+    },
+    {
+      "epoch": 0.8145849495733126,
+      "eval_holoassist_loss": 0.34516745805740356,
+      "step": 14700
+    },
+    {
+      "epoch": 0.8145849495733126,
+      "eval_robofail_loss": 0.20557105541229248,
+      "step": 14700
+    },
+    {
+      "epoch": 0.815139088994791,
+      "grad_norm": 1.2182979583740234,
+      "learning_rate": 8.364645720686465e-08,
+      "loss": 0.5068,
+      "step": 14710
+    },
+    {
+      "epoch": 0.8156932284162696,
+      "grad_norm": 1.2964673042297363,
+      "learning_rate": 8.3160243390056e-08,
+      "loss": 0.4959,
+      "step": 14720
+    },
+    {
+      "epoch": 0.816247367837748,
+      "grad_norm": 1.4023096561431885,
+      "learning_rate": 8.267531860236026e-08,
+      "loss": 0.5091,
+      "step": 14730
+    },
+    {
+      "epoch": 0.8168015072592264,
+      "grad_norm": 1.348850965499878,
+      "learning_rate": 8.219168434335149e-08,
+      "loss": 0.5034,
+      "step": 14740
+    },
+    {
+      "epoch": 0.8173556466807048,
+      "grad_norm": 1.2601499557495117,
+      "learning_rate": 8.170934210861319e-08,
+      "loss": 0.5029,
+      "step": 14750
+    },
+    {
+      "epoch": 0.8179097861021833,
+      "grad_norm": 1.3339990377426147,
+      "learning_rate": 8.122829338973363e-08,
+      "loss": 0.4932,
+      "step": 14760
+    },
+    {
+      "epoch": 0.8184639255236618,
+      "grad_norm": 1.3617079257965088,
+      "learning_rate": 8.07485396743004e-08,
+      "loss": 0.4993,
+      "step": 14770
+    },
+    {
+      "epoch": 0.8190180649451402,
+      "grad_norm": 1.4186921119689941,
+      "learning_rate": 8.027008244589712e-08,
+      "loss": 0.4967,
+      "step": 14780
+    },
+    {
+      "epoch": 0.8195722043666186,
+      "grad_norm": 1.3572496175765991,
+      "learning_rate": 7.979292318409764e-08,
+      "loss": 0.5055,
+      "step": 14790
+    },
+    {
+      "epoch": 0.8201263437880971,
+      "grad_norm": 1.245921015739441,
+      "learning_rate": 7.931706336446214e-08,
+      "loss": 0.5035,
+      "step": 14800
+    },
+    {
+      "epoch": 0.8201263437880971,
+      "eval_robovqa_loss": 0.07961060851812363,
+      "step": 14800
+    },
+    {
+      "epoch": 0.8201263437880971,
+      "eval_agibot_loss": 0.46477118134498596,
+      "step": 14800
+    },
+    {
+      "epoch": 0.8201263437880971,
+      "eval_holoassist_loss": 0.33696380257606506,
+      "step": 14800
+    },
+    {
+      "epoch": 0.8201263437880971,
+      "eval_robofail_loss": 0.2065405696630478,
+      "step": 14800
+    },
+    {
+      "epoch": 0.8206804832095755,
+      "grad_norm": 1.3446458578109741,
+      "learning_rate": 7.884250445853296e-08,
+      "loss": 0.4947,
+      "step": 14810
+    },
+    {
+      "epoch": 0.821234622631054,
+      "grad_norm": 1.4389073848724365,
+      "learning_rate": 7.836924793382848e-08,
+      "loss": 0.4994,
+      "step": 14820
+    },
+    {
+      "epoch": 0.8217887620525324,
+      "grad_norm": 1.2579436302185059,
+      "learning_rate": 7.789729525384075e-08,
+      "loss": 0.4999,
+      "step": 14830
+    },
+    {
+      "epoch": 0.8223429014740109,
+      "grad_norm": 1.2393964529037476,
+      "learning_rate": 7.742664787802871e-08,
+      "loss": 0.4991,
+      "step": 14840
+    },
+    {
+      "epoch": 0.8228970408954893,
+      "grad_norm": 1.2655634880065918,
+      "learning_rate": 7.695730726181581e-08,
+      "loss": 0.5104,
+      "step": 14850
+    },
+    {
+      "epoch": 0.8234511803169677,
+      "grad_norm": 1.381477952003479,
+      "learning_rate": 7.648927485658386e-08,
+      "loss": 0.5162,
+      "step": 14860
+    },
+    {
+      "epoch": 0.8240053197384462,
+      "grad_norm": 1.1565226316452026,
+      "learning_rate": 7.60225521096694e-08,
+      "loss": 0.5051,
+      "step": 14870
+    },
+    {
+      "epoch": 0.8245594591599247,
+      "grad_norm": 1.2606827020645142,
+      "learning_rate": 7.555714046435896e-08,
+      "loss": 0.503,
+      "step": 14880
+    },
+    {
+      "epoch": 0.8251135985814031,
+      "grad_norm": 1.312856912612915,
+      "learning_rate": 7.509304135988464e-08,
+      "loss": 0.4885,
+      "step": 14890
+    },
+    {
+      "epoch": 0.8256677380028815,
+      "grad_norm": 1.4741953611373901,
+      "learning_rate": 7.463025623141966e-08,
+      "loss": 0.4969,
+      "step": 14900
+    },
+    {
+      "epoch": 0.8256677380028815,
+      "eval_robovqa_loss": 0.08048444241285324,
+      "step": 14900
+    },
+    {
+      "epoch": 0.8256677380028815,
+      "eval_agibot_loss": 0.4689415991306305,
+      "step": 14900
+    },
+    {
+      "epoch": 0.8256677380028815,
+      "eval_holoassist_loss": 0.33760762214660645,
+      "step": 14900
+    },
+    {
+      "epoch": 0.8256677380028815,
+      "eval_robofail_loss": 0.20450977981090546,
+      "step": 14900
+    },
+    {
+      "epoch": 0.8262218774243599,
+      "grad_norm": 1.6763007640838623,
+      "learning_rate": 7.416878651007391e-08,
+      "loss": 0.5014,
+      "step": 14910
+    },
+    {
+      "epoch": 0.8267760168458385,
+      "grad_norm": 1.340054988861084,
+      "learning_rate": 7.370863362288959e-08,
+      "loss": 0.506,
+      "step": 14920
+    },
+    {
+      "epoch": 0.8273301562673169,
+      "grad_norm": 1.2200467586517334,
+      "learning_rate": 7.32497989928365e-08,
+      "loss": 0.4985,
+      "step": 14930
+    },
+    {
+      "epoch": 0.8278842956887953,
+      "grad_norm": 1.2595113515853882,
+      "learning_rate": 7.279228403880855e-08,
+      "loss": 0.5048,
+      "step": 14940
+    },
+    {
+      "epoch": 0.8284384351102737,
+      "grad_norm": 1.2719165086746216,
+      "learning_rate": 7.23360901756177e-08,
+      "loss": 0.5051,
+      "step": 14950
+    },
+    {
+      "epoch": 0.8289925745317522,
+      "grad_norm": 1.2548913955688477,
+      "learning_rate": 7.188121881399145e-08,
+      "loss": 0.5031,
+      "step": 14960
+    },
+    {
+      "epoch": 0.8295467139532307,
+      "grad_norm": 1.3111087083816528,
+      "learning_rate": 7.142767136056715e-08,
+      "loss": 0.4968,
+      "step": 14970
+    },
+    {
+      "epoch": 0.8301008533747091,
+      "grad_norm": 1.2431973218917847,
+      "learning_rate": 7.09754492178884e-08,
+      "loss": 0.4947,
+      "step": 14980
+    },
+    {
+      "epoch": 0.8306549927961875,
+      "grad_norm": 1.212968349456787,
+      "learning_rate": 7.052455378440009e-08,
+      "loss": 0.4996,
+      "step": 14990
+    },
+    {
+      "epoch": 0.831209132217666,
+      "grad_norm": 1.1992779970169067,
+      "learning_rate": 7.00749864544447e-08,
+      "loss": 0.4994,
+      "step": 15000
+    },
+    {
+      "epoch": 0.831209132217666,
+      "eval_robovqa_loss": 0.07818939536809921,
+      "step": 15000
+    },
+    {
+      "epoch": 0.831209132217666,
+      "eval_agibot_loss": 0.4664461612701416,
+      "step": 15000
+    },
+    {
+      "epoch": 0.831209132217666,
+      "eval_holoassist_loss": 0.337824285030365,
+      "step": 15000
+    },
+    {
+      "epoch": 0.831209132217666,
+      "eval_robofail_loss": 0.20724499225616455,
+      "step": 15000
+    },
+    {
+      "epoch": 0.8317632716391444,
+      "grad_norm": 1.3036961555480957,
+      "learning_rate": 6.962674861825762e-08,
+      "loss": 0.5,
+      "step": 15010
+    },
+    {
+      "epoch": 0.8323174110606228,
+      "grad_norm": 1.3443633317947388,
+      "learning_rate": 6.917984166196294e-08,
+      "loss": 0.5026,
+      "step": 15020
+    },
+    {
+      "epoch": 0.8328715504821013,
+      "grad_norm": 1.251579761505127,
+      "learning_rate": 6.873426696756907e-08,
+      "loss": 0.5025,
+      "step": 15030
+    },
+    {
+      "epoch": 0.8334256899035798,
+      "grad_norm": 1.3679530620574951,
+      "learning_rate": 6.829002591296462e-08,
+      "loss": 0.51,
+      "step": 15040
+    },
+    {
+      "epoch": 0.8339798293250582,
+      "grad_norm": 1.2519484758377075,
+      "learning_rate": 6.784711987191416e-08,
+      "loss": 0.5047,
+      "step": 15050
+    },
+    {
+      "epoch": 0.8345339687465366,
+      "grad_norm": 1.2893295288085938,
+      "learning_rate": 6.740555021405359e-08,
+      "loss": 0.5029,
+      "step": 15060
+    },
+    {
+      "epoch": 0.835088108168015,
+      "grad_norm": 1.2519820928573608,
+      "learning_rate": 6.696531830488678e-08,
+      "loss": 0.4986,
+      "step": 15070
+    },
+    {
+      "epoch": 0.8356422475894936,
+      "grad_norm": 1.397689938545227,
+      "learning_rate": 6.652642550577992e-08,
+      "loss": 0.5049,
+      "step": 15080
+    },
+    {
+      "epoch": 0.836196387010972,
+      "grad_norm": 1.349165439605713,
+      "learning_rate": 6.6088873173959e-08,
+      "loss": 0.5033,
+      "step": 15090
+    },
+    {
+      "epoch": 0.8367505264324504,
+      "grad_norm": 1.3300151824951172,
+      "learning_rate": 6.565266266250424e-08,
+      "loss": 0.506,
+      "step": 15100
+    },
+    {
+      "epoch": 0.8367505264324504,
+      "eval_robovqa_loss": 0.07980124652385712,
+      "step": 15100
+    },
+    {
+      "epoch": 0.8367505264324504,
+      "eval_agibot_loss": 0.4690721929073334,
+      "step": 15100
+    },
+    {
+      "epoch": 0.8367505264324504,
+      "eval_holoassist_loss": 0.33955150842666626,
+      "step": 15100
+    },
+    {
+      "epoch": 0.8367505264324504,
+      "eval_robofail_loss": 0.2068992555141449,
+      "step": 15100
+    },
+    {
+      "epoch": 0.8373046658539288,
+      "grad_norm": 1.2830733060836792,
+      "learning_rate": 6.521779532034666e-08,
+      "loss": 0.5066,
+      "step": 15110
+    },
+    {
+      "epoch": 0.8378588052754073,
+      "grad_norm": 1.3282312154769897,
+      "learning_rate": 6.478427249226353e-08,
+      "loss": 0.4936,
+      "step": 15120
+    },
+    {
+      "epoch": 0.8384129446968858,
+      "grad_norm": 1.2314647436141968,
+      "learning_rate": 6.43520955188745e-08,
+      "loss": 0.5027,
+      "step": 15130
+    },
+    {
+      "epoch": 0.8389670841183642,
+      "grad_norm": 1.2979685068130493,
+      "learning_rate": 6.392126573663725e-08,
+      "loss": 0.5028,
+      "step": 15140
+    },
+    {
+      "epoch": 0.8395212235398426,
+      "grad_norm": 1.2357728481292725,
+      "learning_rate": 6.349178447784343e-08,
+      "loss": 0.5021,
+      "step": 15150
+    },
+    {
+      "epoch": 0.840075362961321,
+      "grad_norm": 1.438501000404358,
+      "learning_rate": 6.306365307061457e-08,
+      "loss": 0.5043,
+      "step": 15160
+    },
+    {
+      "epoch": 0.8406295023827995,
+      "grad_norm": 1.271740436553955,
+      "learning_rate": 6.263687283889773e-08,
+      "loss": 0.506,
+      "step": 15170
+    },
+    {
+      "epoch": 0.841183641804278,
+      "grad_norm": 1.219208836555481,
+      "learning_rate": 6.221144510246229e-08,
+      "loss": 0.4971,
+      "step": 15180
+    },
+    {
+      "epoch": 0.8417377812257564,
+      "grad_norm": 1.1988638639450073,
+      "learning_rate": 6.178737117689409e-08,
+      "loss": 0.5043,
+      "step": 15190
+    },
+    {
+      "epoch": 0.8422919206472348,
+      "grad_norm": 1.1887098550796509,
+      "learning_rate": 6.136465237359368e-08,
+      "loss": 0.5052,
+      "step": 15200
+    },
+    {
+      "epoch": 0.8422919206472348,
+      "eval_robovqa_loss": 0.0789259672164917,
+      "step": 15200
+    },
+    {
+      "epoch": 0.8422919206472348,
+      "eval_agibot_loss": 0.47145771980285645,
+      "step": 15200
+    },
+    {
+      "epoch": 0.8422919206472348,
+      "eval_holoassist_loss": 0.3435930907726288,
+      "step": 15200
+    },
+    {
+      "epoch": 0.8422919206472348,
+      "eval_robofail_loss": 0.20610633492469788,
+      "step": 15200
+    },
+    {
+      "epoch": 0.8428460600687133,
+      "grad_norm": 1.3698608875274658,
+      "learning_rate": 6.094328999976988e-08,
+      "loss": 0.5017,
+      "step": 15210
+    },
+    {
+      "epoch": 0.8434001994901917,
+      "grad_norm": 1.305781364440918,
+      "learning_rate": 6.052328535843787e-08,
+      "loss": 0.5011,
+      "step": 15220
+    },
+    {
+      "epoch": 0.8439543389116702,
+      "grad_norm": 1.3222838640213013,
+      "learning_rate": 6.010463974841378e-08,
+      "loss": 0.5047,
+      "step": 15230
+    },
+    {
+      "epoch": 0.8445084783331486,
+      "grad_norm": 1.3168706893920898,
+      "learning_rate": 5.968735446431078e-08,
+      "loss": 0.502,
+      "step": 15240
+    },
+    {
+      "epoch": 0.8450626177546271,
+      "grad_norm": 1.2537970542907715,
+      "learning_rate": 5.9271430796536125e-08,
+      "loss": 0.499,
+      "step": 15250
+    },
+    {
+      "epoch": 0.8456167571761055,
+      "grad_norm": 1.2194982767105103,
+      "learning_rate": 5.8856870031285864e-08,
+      "loss": 0.496,
+      "step": 15260
+    },
+    {
+      "epoch": 0.8461708965975839,
+      "grad_norm": 1.162577748298645,
+      "learning_rate": 5.844367345054158e-08,
+      "loss": 0.5055,
+      "step": 15270
+    },
+    {
+      "epoch": 0.8467250360190623,
+      "grad_norm": 1.2621872425079346,
+      "learning_rate": 5.803184233206632e-08,
+      "loss": 0.5013,
+      "step": 15280
+    },
+    {
+      "epoch": 0.8472791754405409,
+      "grad_norm": 1.2299410104751587,
+      "learning_rate": 5.7621377949400474e-08,
+      "loss": 0.4944,
+      "step": 15290
+    },
+    {
+      "epoch": 0.8478333148620193,
+      "grad_norm": 1.311079978942871,
+      "learning_rate": 5.7212281571858144e-08,
+      "loss": 0.5026,
+      "step": 15300
+    },
+    {
+      "epoch": 0.8478333148620193,
+      "eval_robovqa_loss": 0.07984565198421478,
+      "step": 15300
+    },
+    {
+      "epoch": 0.8478333148620193,
+      "eval_agibot_loss": 0.4716016352176666,
+      "step": 15300
+    },
+    {
+      "epoch": 0.8478333148620193,
+      "eval_holoassist_loss": 0.339059978723526,
+      "step": 15300
+    },
+    {
+      "epoch": 0.8478333148620193,
+      "eval_robofail_loss": 0.20713838934898376,
+      "step": 15300
+    },
+    {
+      "epoch": 0.8483874542834977,
+      "grad_norm": 1.3020635843276978,
+      "learning_rate": 5.680455446452292e-08,
+      "loss": 0.4988,
+      "step": 15310
+    },
+    {
+      "epoch": 0.8489415937049761,
+      "grad_norm": 1.3074508905410767,
+      "learning_rate": 5.639819788824407e-08,
+      "loss": 0.5022,
+      "step": 15320
+    },
+    {
+      "epoch": 0.8494957331264547,
+      "grad_norm": 1.347007155418396,
+      "learning_rate": 5.5993213099632617e-08,
+      "loss": 0.4995,
+      "step": 15330
+    },
+    {
+      "epoch": 0.8500498725479331,
+      "grad_norm": 1.2432276010513306,
+      "learning_rate": 5.5589601351057615e-08,
+      "loss": 0.498,
+      "step": 15340
+    },
+    {
+      "epoch": 0.8506040119694115,
+      "grad_norm": 1.324554443359375,
+      "learning_rate": 5.518736389064199e-08,
+      "loss": 0.5066,
+      "step": 15350
+    },
+    {
+      "epoch": 0.8511581513908899,
+      "grad_norm": 1.3333237171173096,
+      "learning_rate": 5.47865019622592e-08,
+      "loss": 0.5064,
+      "step": 15360
+    },
+    {
+      "epoch": 0.8517122908123684,
+      "grad_norm": 1.228026032447815,
+      "learning_rate": 5.438701680552832e-08,
+      "loss": 0.5039,
+      "step": 15370
+    },
+    {
+      "epoch": 0.8522664302338469,
+      "grad_norm": 1.2469559907913208,
+      "learning_rate": 5.398890965581177e-08,
+      "loss": 0.5029,
+      "step": 15380
+    },
+    {
+      "epoch": 0.8528205696553253,
+      "grad_norm": 1.2912787199020386,
+      "learning_rate": 5.3592181744209986e-08,
+      "loss": 0.4999,
+      "step": 15390
+    },
+    {
+      "epoch": 0.8533747090768037,
+      "grad_norm": 1.283332109451294,
+      "learning_rate": 5.3196834297558555e-08,
+      "loss": 0.5005,
+      "step": 15400
+    },
+    {
+      "epoch": 0.8533747090768037,
+      "eval_robovqa_loss": 0.07939862459897995,
+      "step": 15400
+    },
+    {
+      "epoch": 0.8533747090768037,
+      "eval_agibot_loss": 0.4680336117744446,
+      "step": 15400
+    },
+    {
+      "epoch": 0.8533747090768037,
+      "eval_holoassist_loss": 0.3447457253932953,
+      "step": 15400
+    },
+    {
+      "epoch": 0.8533747090768037,
+      "eval_robofail_loss": 0.20662450790405273,
+      "step": 15400
+    },
+    {
+      "epoch": 0.8539288484982822,
+      "grad_norm": 1.1684739589691162,
+      "learning_rate": 5.280286853842419e-08,
+      "loss": 0.4956,
+      "step": 15410
+    },
+    {
+      "epoch": 0.8544829879197606,
+      "grad_norm": 1.2655730247497559,
+      "learning_rate": 5.241028568510064e-08,
+      "loss": 0.5027,
+      "step": 15420
+    },
+    {
+      "epoch": 0.855037127341239,
+      "grad_norm": 1.3228144645690918,
+      "learning_rate": 5.2019086951605397e-08,
+      "loss": 0.5057,
+      "step": 15430
+    },
+    {
+      "epoch": 0.8555912667627175,
+      "grad_norm": 1.263094186782837,
+      "learning_rate": 5.162927354767571e-08,
+      "loss": 0.5083,
+      "step": 15440
+    },
+    {
+      "epoch": 0.856145406184196,
+      "grad_norm": 1.283470869064331,
+      "learning_rate": 5.124084667876466e-08,
+      "loss": 0.5002,
+      "step": 15450
+    },
+    {
+      "epoch": 0.8566995456056744,
+      "grad_norm": 1.3667223453521729,
+      "learning_rate": 5.08538075460378e-08,
+      "loss": 0.5114,
+      "step": 15460
+    },
+    {
+      "epoch": 0.8572536850271528,
+      "grad_norm": 1.2814877033233643,
+      "learning_rate": 5.046815734636945e-08,
+      "loss": 0.5031,
+      "step": 15470
+    },
+    {
+      "epoch": 0.8578078244486312,
+      "grad_norm": 1.2532013654708862,
+      "learning_rate": 5.008389727233825e-08,
+      "loss": 0.4976,
+      "step": 15480
+    },
+    {
+      "epoch": 0.8583619638701098,
+      "grad_norm": 1.2411538362503052,
+      "learning_rate": 4.9701028512224684e-08,
+      "loss": 0.4999,
+      "step": 15490
+    },
+    {
+      "epoch": 0.8589161032915882,
+      "grad_norm": 1.3669066429138184,
+      "learning_rate": 4.9319552250006225e-08,
+      "loss": 0.5031,
+      "step": 15500
+    },
+    {
+      "epoch": 0.8589161032915882,
+      "eval_robovqa_loss": 0.07973505556583405,
+      "step": 15500
+    },
+    {
+      "epoch": 0.8589161032915882,
+      "eval_agibot_loss": 0.4694417119026184,
+      "step": 15500
+    },
+    {
+      "epoch": 0.8589161032915882,
+      "eval_holoassist_loss": 0.34358322620391846,
+      "step": 15500
+    },
+    {
+      "epoch": 0.8589161032915882,
+      "eval_robofail_loss": 0.20606116950511932,
+      "step": 15500
+    },
+    {
+      "epoch": 0.8594702427130666,
+      "grad_norm": 1.2971889972686768,
+      "learning_rate": 4.893946966535456e-08,
+      "loss": 0.5013,
+      "step": 15510
+    },
+    {
+      "epoch": 0.860024382134545,
+      "grad_norm": 1.226798415184021,
+      "learning_rate": 4.85607819336315e-08,
+      "loss": 0.5044,
+      "step": 15520
+    },
+    {
+      "epoch": 0.8605785215560235,
+      "grad_norm": 1.3179229497909546,
+      "learning_rate": 4.8183490225885356e-08,
+      "loss": 0.5015,
+      "step": 15530
+    },
+    {
+      "epoch": 0.861132660977502,
+      "grad_norm": 1.3810381889343262,
+      "learning_rate": 4.7807595708847404e-08,
+      "loss": 0.5034,
+      "step": 15540
+    },
+    {
+      "epoch": 0.8616868003989804,
+      "grad_norm": 1.2863613367080688,
+      "learning_rate": 4.7433099544928304e-08,
+      "loss": 0.5023,
+      "step": 15550
+    },
+    {
+      "epoch": 0.8622409398204588,
+      "grad_norm": 1.2590227127075195,
+      "learning_rate": 4.706000289221451e-08,
+      "loss": 0.4942,
+      "step": 15560
+    },
+    {
+      "epoch": 0.8627950792419373,
+      "grad_norm": 1.2334527969360352,
+      "learning_rate": 4.6688306904464566e-08,
+      "loss": 0.5051,
+      "step": 15570
+    },
+    {
+      "epoch": 0.8633492186634157,
+      "grad_norm": 1.3011952638626099,
+      "learning_rate": 4.631801273110575e-08,
+      "loss": 0.5005,
+      "step": 15580
+    },
+    {
+      "epoch": 0.8639033580848942,
+      "grad_norm": 1.2918076515197754,
+      "learning_rate": 4.5949121517230074e-08,
+      "loss": 0.5081,
+      "step": 15590
+    },
+    {
+      "epoch": 0.8644574975063726,
+      "grad_norm": 1.3559482097625732,
+      "learning_rate": 4.558163440359164e-08,
+      "loss": 0.4973,
+      "step": 15600
+    },
+    {
+      "epoch": 0.8644574975063726,
+      "eval_robovqa_loss": 0.07946397364139557,
+      "step": 15600
+    },
+    {
+      "epoch": 0.8644574975063726,
+      "eval_agibot_loss": 0.47012320160865784,
+      "step": 15600
+    },
+    {
+      "epoch": 0.8644574975063726,
+      "eval_holoassist_loss": 0.3427233099937439,
+      "step": 15600
+    },
+    {
+      "epoch": 0.8644574975063726,
+      "eval_robofail_loss": 0.20614294707775116,
+      "step": 15600
+    },
+    {
+      "epoch": 0.8650116369278511,
+      "grad_norm": 1.1788452863693237,
+      "learning_rate": 4.5215552526601755e-08,
+      "loss": 0.4985,
+      "step": 15610
+    },
+    {
+      "epoch": 0.8655657763493295,
+      "grad_norm": 1.2811709642410278,
+      "learning_rate": 4.485087701832707e-08,
+      "loss": 0.5033,
+      "step": 15620
+    },
+    {
+      "epoch": 0.8661199157708079,
+      "grad_norm": 1.4001015424728394,
+      "learning_rate": 4.4487609006484294e-08,
+      "loss": 0.4987,
+      "step": 15630
+    },
+    {
+      "epoch": 0.8666740551922864,
+      "grad_norm": 1.2912287712097168,
+      "learning_rate": 4.412574961443843e-08,
+      "loss": 0.5002,
+      "step": 15640
+    },
+    {
+      "epoch": 0.8672281946137648,
+      "grad_norm": 1.2593200206756592,
+      "learning_rate": 4.376529996119804e-08,
+      "loss": 0.5032,
+      "step": 15650
+    },
+    {
+      "epoch": 0.8677823340352433,
+      "grad_norm": 1.1706631183624268,
+      "learning_rate": 4.340626116141205e-08,
+      "loss": 0.4976,
+      "step": 15660
+    },
+    {
+      "epoch": 0.8683364734567217,
+      "grad_norm": 1.1834205389022827,
+      "learning_rate": 4.304863432536704e-08,
+      "loss": 0.4986,
+      "step": 15670
+    },
+    {
+      "epoch": 0.8688906128782001,
+      "grad_norm": 1.1847405433654785,
+      "learning_rate": 4.269242055898287e-08,
+      "loss": 0.502,
+      "step": 15680
+    },
+    {
+      "epoch": 0.8694447522996785,
+      "grad_norm": 1.2474462985992432,
+      "learning_rate": 4.2337620963809717e-08,
+      "loss": 0.4979,
+      "step": 15690
+    },
+    {
+      "epoch": 0.8699988917211571,
+      "grad_norm": 1.3208974599838257,
+      "learning_rate": 4.1984236637024664e-08,
+      "loss": 0.504,
+      "step": 15700
+    },
+    {
+      "epoch": 0.8699988917211571,
+      "eval_robovqa_loss": 0.07964864373207092,
+      "step": 15700
+    },
+    {
+      "epoch": 0.8699988917211571,
+      "eval_agibot_loss": 0.4706023335456848,
+      "step": 15700
+    },
+    {
+      "epoch": 0.8699988917211571,
+      "eval_holoassist_loss": 0.34353363513946533,
+      "step": 15700
+    },
+    {
+      "epoch": 0.8699988917211571,
+      "eval_robofail_loss": 0.20575182139873505,
+      "step": 15700
+    },
+    {
+      "epoch": 0.8705530311426355,
+      "grad_norm": 1.373943567276001,
+      "learning_rate": 4.163226867142816e-08,
+      "loss": 0.5106,
+      "step": 15710
+    },
+    {
+      "epoch": 0.8711071705641139,
+      "grad_norm": 1.3266154527664185,
+      "learning_rate": 4.1281718155440626e-08,
+      "loss": 0.51,
+      "step": 15720
+    },
+    {
+      "epoch": 0.8716613099855923,
+      "grad_norm": 1.3482214212417603,
+      "learning_rate": 4.0932586173099634e-08,
+      "loss": 0.5016,
+      "step": 15730
+    },
+    {
+      "epoch": 0.8722154494070709,
+      "grad_norm": 1.3130931854248047,
+      "learning_rate": 4.0584873804055453e-08,
+      "loss": 0.5056,
+      "step": 15740
+    },
+    {
+      "epoch": 0.8727695888285493,
+      "grad_norm": 1.4110873937606812,
+      "learning_rate": 4.023858212356867e-08,
+      "loss": 0.5027,
+      "step": 15750
+    },
+    {
+      "epoch": 0.8733237282500277,
+      "grad_norm": 1.291988492012024,
+      "learning_rate": 3.9893712202506924e-08,
+      "loss": 0.4983,
+      "step": 15760
+    },
+    {
+      "epoch": 0.8738778676715061,
+      "grad_norm": 1.2607611417770386,
+      "learning_rate": 3.9550265107340444e-08,
+      "loss": 0.5048,
+      "step": 15770
+    },
+    {
+      "epoch": 0.8744320070929846,
+      "grad_norm": 1.1833827495574951,
+      "learning_rate": 3.920824190014027e-08,
+      "loss": 0.5037,
+      "step": 15780
+    },
+    {
+      "epoch": 0.874986146514463,
+      "grad_norm": 1.4885532855987549,
+      "learning_rate": 3.88676436385737e-08,
+      "loss": 0.5026,
+      "step": 15790
+    },
+    {
+      "epoch": 0.8755402859359415,
+      "grad_norm": 1.2019237279891968,
+      "learning_rate": 3.8528471375901927e-08,
+      "loss": 0.499,
+      "step": 15800
+    },
+    {
+      "epoch": 0.8755402859359415,
+      "eval_robovqa_loss": 0.07928010076284409,
+      "step": 15800
+    },
+    {
+      "epoch": 0.8755402859359415,
+      "eval_agibot_loss": 0.46986114978790283,
+      "step": 15800
+    },
+    {
+      "epoch": 0.8755402859359415,
+      "eval_holoassist_loss": 0.34036532044410706,
+      "step": 15800
+    },
+    {
+      "epoch": 0.8755402859359415,
+      "eval_robofail_loss": 0.20599603652954102,
+      "step": 15800
+    },
+    {
+      "epoch": 0.8760944253574199,
+      "grad_norm": 1.400942325592041,
+      "learning_rate": 3.8190726160976236e-08,
+      "loss": 0.4956,
+      "step": 15810
+    },
+    {
+      "epoch": 0.8766485647788984,
+      "grad_norm": 1.3975341320037842,
+      "learning_rate": 3.785440903823489e-08,
+      "loss": 0.5102,
+      "step": 15820
+    },
+    {
+      "epoch": 0.8772027042003768,
+      "grad_norm": 1.3359533548355103,
+      "learning_rate": 3.751952104770001e-08,
+      "loss": 0.4995,
+      "step": 15830
+    },
+    {
+      "epoch": 0.8777568436218552,
+      "grad_norm": 1.315171241760254,
+      "learning_rate": 3.718606322497431e-08,
+      "loss": 0.4976,
+      "step": 15840
+    },
+    {
+      "epoch": 0.8783109830433337,
+      "grad_norm": 1.3162868022918701,
+      "learning_rate": 3.685403660123776e-08,
+      "loss": 0.5016,
+      "step": 15850
+    },
+    {
+      "epoch": 0.8788651224648122,
+      "grad_norm": 1.302438735961914,
+      "learning_rate": 3.652344220324471e-08,
+      "loss": 0.5043,
+      "step": 15860
+    },
+    {
+      "epoch": 0.8794192618862906,
+      "grad_norm": 1.2216514348983765,
+      "learning_rate": 3.619428105332029e-08,
+      "loss": 0.5009,
+      "step": 15870
+    },
+    {
+      "epoch": 0.879973401307769,
+      "grad_norm": 1.3376489877700806,
+      "learning_rate": 3.586655416935758e-08,
+      "loss": 0.502,
+      "step": 15880
+    },
+    {
+      "epoch": 0.8805275407292474,
+      "grad_norm": 1.3758950233459473,
+      "learning_rate": 3.554026256481457e-08,
+      "loss": 0.5024,
+      "step": 15890
+    },
+    {
+      "epoch": 0.881081680150726,
+      "grad_norm": 1.2839524745941162,
+      "learning_rate": 3.521540724871025e-08,
+      "loss": 0.4968,
+      "step": 15900
+    },
+    {
+      "epoch": 0.881081680150726,
+      "eval_robovqa_loss": 0.08014993369579315,
+      "step": 15900
+    },
+    {
+      "epoch": 0.881081680150726,
+      "eval_agibot_loss": 0.46950507164001465,
+      "step": 15900
+    },
+    {
+      "epoch": 0.881081680150726,
+      "eval_holoassist_loss": 0.33926185965538025,
+      "step": 15900
+    },
+    {
+      "epoch": 0.881081680150726,
+      "eval_robofail_loss": 0.20595195889472961,
+      "step": 15900
+    },
+    {
+      "epoch": 0.8816358195722044,
+      "grad_norm": 1.228919506072998,
+      "learning_rate": 3.489198922562281e-08,
+      "loss": 0.497,
+      "step": 15910
+    },
+    {
+      "epoch": 0.8821899589936828,
+      "grad_norm": 1.357208251953125,
+      "learning_rate": 3.457000949568495e-08,
+      "loss": 0.5011,
+      "step": 15920
+    },
+    {
+      "epoch": 0.8827440984151612,
+      "grad_norm": 1.3176189661026,
+      "learning_rate": 3.424946905458237e-08,
+      "loss": 0.5015,
+      "step": 15930
+    },
+    {
+      "epoch": 0.8832982378366397,
+      "grad_norm": 1.1953414678573608,
+      "learning_rate": 3.393036889354944e-08,
+      "loss": 0.5067,
+      "step": 15940
+    },
+    {
+      "epoch": 0.8838523772581182,
+      "grad_norm": 1.3596019744873047,
+      "learning_rate": 3.361270999936688e-08,
+      "loss": 0.5103,
+      "step": 15950
+    },
+    {
+      "epoch": 0.8844065166795966,
+      "grad_norm": 1.3599255084991455,
+      "learning_rate": 3.32964933543583e-08,
+      "loss": 0.4992,
+      "step": 15960
+    },
+    {
+      "epoch": 0.884960656101075,
+      "grad_norm": 1.2909244298934937,
+      "learning_rate": 3.2981719936387376e-08,
+      "loss": 0.4973,
+      "step": 15970
+    },
+    {
+      "epoch": 0.8855147955225535,
+      "grad_norm": 1.4103670120239258,
+      "learning_rate": 3.26683907188548e-08,
+      "loss": 0.4995,
+      "step": 15980
+    },
+    {
+      "epoch": 0.8860689349440319,
+      "grad_norm": 1.277976393699646,
+      "learning_rate": 3.2356506670695224e-08,
+      "loss": 0.5054,
+      "step": 15990
+    },
+    {
+      "epoch": 0.8866230743655104,
+      "grad_norm": 1.3098328113555908,
+      "learning_rate": 3.204606875637433e-08,
+      "loss": 0.4957,
+      "step": 16000
+    },
+    {
+      "epoch": 0.8866230743655104,
+      "eval_robovqa_loss": 0.08066313713788986,
+      "step": 16000
+    },
+    {
+      "epoch": 0.8866230743655104,
+      "eval_agibot_loss": 0.4705943763256073,
+      "step": 16000
+    },
+    {
+      "epoch": 0.8866230743655104,
+      "eval_holoassist_loss": 0.3439936339855194,
+      "step": 16000
+    },
+    {
+      "epoch": 0.8866230743655104,
+      "eval_robofail_loss": 0.20622745156288147,
+      "step": 16000
+    },
+    {
+      "epoch": 0.8871772137869888,
+      "grad_norm": 1.235158920288086,
+      "learning_rate": 3.1737077935885595e-08,
+      "loss": 0.5051,
+      "step": 16010
+    },
+    {
+      "epoch": 0.8877313532084673,
+      "grad_norm": 1.2857601642608643,
+      "learning_rate": 3.1429535164747946e-08,
+      "loss": 0.4915,
+      "step": 16020
+    },
+    {
+      "epoch": 0.8882854926299457,
+      "grad_norm": 1.2705779075622559,
+      "learning_rate": 3.112344139400175e-08,
+      "loss": 0.5091,
+      "step": 16030
+    },
+    {
+      "epoch": 0.8888396320514241,
+      "grad_norm": 2.750047445297241,
+      "learning_rate": 3.0818797570207225e-08,
+      "loss": 0.5082,
+      "step": 16040
+    },
+    {
+      "epoch": 0.8893937714729026,
+      "grad_norm": 1.32404625415802,
+      "learning_rate": 3.051560463544006e-08,
+      "loss": 0.4942,
+      "step": 16050
+    },
+    {
+      "epoch": 0.8899479108943811,
+      "grad_norm": 1.2928636074066162,
+      "learning_rate": 3.0213863527289995e-08,
+      "loss": 0.5037,
+      "step": 16060
+    },
+    {
+      "epoch": 0.8905020503158595,
+      "grad_norm": 1.3412522077560425,
+      "learning_rate": 2.991357517885651e-08,
+      "loss": 0.4954,
+      "step": 16070
+    },
+    {
+      "epoch": 0.8910561897373379,
+      "grad_norm": 1.2603336572647095,
+      "learning_rate": 2.9614740518747015e-08,
+      "loss": 0.501,
+      "step": 16080
+    },
+    {
+      "epoch": 0.8916103291588163,
+      "grad_norm": 1.224798560142517,
+      "learning_rate": 2.931736047107336e-08,
+      "loss": 0.5061,
+      "step": 16090
+    },
+    {
+      "epoch": 0.8921644685802949,
+      "grad_norm": 1.232436180114746,
+      "learning_rate": 2.9021435955449146e-08,
+      "loss": 0.5095,
+      "step": 16100
+    },
+    {
+      "epoch": 0.8921644685802949,
+      "eval_robovqa_loss": 0.07940918952226639,
+      "step": 16100
+    },
+    {
+      "epoch": 0.8921644685802949,
+      "eval_agibot_loss": 0.46845871210098267,
+      "step": 16100
+    },
+    {
+      "epoch": 0.8921644685802949,
+      "eval_holoassist_loss": 0.34106385707855225,
+      "step": 16100
+    },
+    {
+      "epoch": 0.8921644685802949,
+      "eval_robofail_loss": 0.2060997635126114,
+      "step": 16100
+    },
+    {
+      "epoch": 0.8927186080017733,
+      "grad_norm": 1.236258864402771,
+      "learning_rate": 2.8726967886987087e-08,
+      "loss": 0.4998,
+      "step": 16110
+    },
+    {
+      "epoch": 0.8932727474232517,
+      "grad_norm": 1.2110929489135742,
+      "learning_rate": 2.843395717629582e-08,
+      "loss": 0.4958,
+      "step": 16120
+    },
+    {
+      "epoch": 0.8938268868447301,
+      "grad_norm": 1.3992581367492676,
+      "learning_rate": 2.814240472947732e-08,
+      "loss": 0.5105,
+      "step": 16130
+    },
+    {
+      "epoch": 0.8943810262662085,
+      "grad_norm": 1.2504703998565674,
+      "learning_rate": 2.785231144812411e-08,
+      "loss": 0.4961,
+      "step": 16140
+    },
+    {
+      "epoch": 0.894935165687687,
+      "grad_norm": 1.2840800285339355,
+      "learning_rate": 2.756367822931649e-08,
+      "loss": 0.5009,
+      "step": 16150
+    },
+    {
+      "epoch": 0.8954893051091655,
+      "grad_norm": 1.8613845109939575,
+      "learning_rate": 2.7276505965619424e-08,
+      "loss": 0.5007,
+      "step": 16160
+    },
+    {
+      "epoch": 0.8960434445306439,
+      "grad_norm": 1.3030331134796143,
+      "learning_rate": 2.6990795545080102e-08,
+      "loss": 0.4982,
+      "step": 16170
+    },
+    {
+      "epoch": 0.8965975839521223,
+      "grad_norm": 1.3419381380081177,
+      "learning_rate": 2.6706547851225502e-08,
+      "loss": 0.5008,
+      "step": 16180
+    },
+    {
+      "epoch": 0.8971517233736008,
+      "grad_norm": 1.294830322265625,
+      "learning_rate": 2.642376376305866e-08,
+      "loss": 0.5063,
+      "step": 16190
+    },
+    {
+      "epoch": 0.8977058627950792,
+      "grad_norm": 1.390956997871399,
+      "learning_rate": 2.6142444155057232e-08,
+      "loss": 0.4959,
+      "step": 16200
+    },
+    {
+      "epoch": 0.8977058627950792,
+      "eval_robovqa_loss": 0.07957933098077774,
+      "step": 16200
+    },
+    {
+      "epoch": 0.8977058627950792,
+      "eval_agibot_loss": 0.46854168176651,
+      "step": 16200
+    },
+    {
+      "epoch": 0.8977058627950792,
+      "eval_holoassist_loss": 0.3405410647392273,
+      "step": 16200
+    },
+    {
+      "epoch": 0.8977058627950792,
+      "eval_robofail_loss": 0.20521175861358643,
+      "step": 16200
+    },
+    {
+      "epoch": 0.8982600022165577,
+      "grad_norm": 1.523853063583374,
+      "learning_rate": 2.5862589897169495e-08,
+      "loss": 0.5074,
+      "step": 16210
+    },
+    {
+      "epoch": 0.8988141416380361,
+      "grad_norm": 1.282938838005066,
+      "learning_rate": 2.5584201854812916e-08,
+      "loss": 0.4958,
+      "step": 16220
+    },
+    {
+      "epoch": 0.8993682810595146,
+      "grad_norm": 1.2646970748901367,
+      "learning_rate": 2.530728088887041e-08,
+      "loss": 0.5002,
+      "step": 16230
+    },
+    {
+      "epoch": 0.899922420480993,
+      "grad_norm": 1.2770237922668457,
+      "learning_rate": 2.503182785568836e-08,
+      "loss": 0.4968,
+      "step": 16240
+    },
+    {
+      "epoch": 0.9004765599024714,
+      "grad_norm": 1.259892225265503,
+      "learning_rate": 2.475784360707367e-08,
+      "loss": 0.5041,
+      "step": 16250
+    },
+    {
+      "epoch": 0.9010306993239499,
+      "grad_norm": 1.1669116020202637,
+      "learning_rate": 2.44853289902911e-08,
+      "loss": 0.5008,
+      "step": 16260
+    },
+    {
+      "epoch": 0.9015848387454284,
+      "grad_norm": 1.2227928638458252,
+      "learning_rate": 2.421428484806093e-08,
+      "loss": 0.5012,
+      "step": 16270
+    },
+    {
+      "epoch": 0.9021389781669068,
+      "grad_norm": 1.173886775970459,
+      "learning_rate": 2.394471201855597e-08,
+      "loss": 0.4962,
+      "step": 16280
+    },
+    {
+      "epoch": 0.9026931175883852,
+      "grad_norm": 1.2956697940826416,
+      "learning_rate": 2.367661133539933e-08,
+      "loss": 0.505,
+      "step": 16290
+    },
+    {
+      "epoch": 0.9032472570098636,
+      "grad_norm": 1.2023411989212036,
+      "learning_rate": 2.34099836276615e-08,
+      "loss": 0.5028,
+      "step": 16300
+    },
+    {
+      "epoch": 0.9032472570098636,
+      "eval_robovqa_loss": 0.0790875256061554,
+      "step": 16300
+    },
+    {
+      "epoch": 0.9032472570098636,
+      "eval_agibot_loss": 0.4672633707523346,
+      "step": 16300
+    },
+    {
+      "epoch": 0.9032472570098636,
+      "eval_holoassist_loss": 0.3404013514518738,
+      "step": 16300
+    },
+    {
+      "epoch": 0.9032472570098636,
+      "eval_robofail_loss": 0.20556989312171936,
+      "step": 16300
+    },
+    {
+      "epoch": 0.9038013964313422,
+      "grad_norm": 1.2203186750411987,
+      "learning_rate": 2.3144829719858317e-08,
+      "loss": 0.5014,
+      "step": 16310
+    },
+    {
+      "epoch": 0.9043555358528206,
+      "grad_norm": 1.1849100589752197,
+      "learning_rate": 2.2881150431947492e-08,
+      "loss": 0.5012,
+      "step": 16320
+    },
+    {
+      "epoch": 0.904909675274299,
+      "grad_norm": 1.2506835460662842,
+      "learning_rate": 2.261894657932728e-08,
+      "loss": 0.5002,
+      "step": 16330
+    },
+    {
+      "epoch": 0.9054638146957774,
+      "grad_norm": 1.189805030822754,
+      "learning_rate": 2.2358218972832687e-08,
+      "loss": 0.5012,
+      "step": 16340
+    },
+    {
+      "epoch": 0.9060179541172559,
+      "grad_norm": 1.3223282098770142,
+      "learning_rate": 2.209896841873421e-08,
+      "loss": 0.4986,
+      "step": 16350
+    },
+    {
+      "epoch": 0.9065720935387344,
+      "grad_norm": 1.2324782609939575,
+      "learning_rate": 2.1841195718734216e-08,
+      "loss": 0.5115,
+      "step": 16360
+    },
+    {
+      "epoch": 0.9071262329602128,
+      "grad_norm": 1.2871334552764893,
+      "learning_rate": 2.158490166996535e-08,
+      "loss": 0.5097,
+      "step": 16370
+    },
+    {
+      "epoch": 0.9076803723816912,
+      "grad_norm": 1.2659308910369873,
+      "learning_rate": 2.1330087064987456e-08,
+      "loss": 0.5037,
+      "step": 16380
+    },
+    {
+      "epoch": 0.9082345118031697,
+      "grad_norm": 1.2907123565673828,
+      "learning_rate": 2.1076752691785494e-08,
+      "loss": 0.493,
+      "step": 16390
+    },
+    {
+      "epoch": 0.9087886512246481,
+      "grad_norm": 2.1892504692077637,
+      "learning_rate": 2.0824899333766964e-08,
+      "loss": 0.5029,
+      "step": 16400
+    },
+    {
+      "epoch": 0.9087886512246481,
+      "eval_robovqa_loss": 0.0789080336689949,
+      "step": 16400
+    },
+    {
+      "epoch": 0.9087886512246481,
+      "eval_agibot_loss": 0.46751609444618225,
+      "step": 16400
+    },
+    {
+      "epoch": 0.9087886512246481,
+      "eval_holoassist_loss": 0.3404904901981354,
+      "step": 16400
+    },
+    {
+      "epoch": 0.9087886512246481,
+      "eval_robofail_loss": 0.20692668855190277,
+      "step": 16400
+    },
+    {
+      "epoch": 0.9093427906461266,
+      "grad_norm": 1.2983829975128174,
+      "learning_rate": 2.057452776975943e-08,
+      "loss": 0.4982,
+      "step": 16410
+    },
+    {
+      "epoch": 0.909896930067605,
+      "grad_norm": 1.2150001525878906,
+      "learning_rate": 2.0325638774008225e-08,
+      "loss": 0.4921,
+      "step": 16420
+    },
+    {
+      "epoch": 0.9104510694890835,
+      "grad_norm": 1.3113360404968262,
+      "learning_rate": 2.0078233116173958e-08,
+      "loss": 0.5032,
+      "step": 16430
+    },
+    {
+      "epoch": 0.9110052089105619,
+      "grad_norm": 1.2693955898284912,
+      "learning_rate": 1.9832311561330417e-08,
+      "loss": 0.5044,
+      "step": 16440
+    },
+    {
+      "epoch": 0.9115593483320403,
+      "grad_norm": 1.3610048294067383,
+      "learning_rate": 1.9587874869961553e-08,
+      "loss": 0.5009,
+      "step": 16450
+    },
+    {
+      "epoch": 0.9121134877535187,
+      "grad_norm": 1.3244050741195679,
+      "learning_rate": 1.9344923797959932e-08,
+      "loss": 0.5009,
+      "step": 16460
+    },
+    {
+      "epoch": 0.9126676271749973,
+      "grad_norm": 1.3783842325210571,
+      "learning_rate": 1.9103459096623864e-08,
+      "loss": 0.5076,
+      "step": 16470
+    },
+    {
+      "epoch": 0.9132217665964757,
+      "grad_norm": 1.4227899312973022,
+      "learning_rate": 1.886348151265521e-08,
+      "loss": 0.5055,
+      "step": 16480
+    },
+    {
+      "epoch": 0.9137759060179541,
+      "grad_norm": 1.330203890800476,
+      "learning_rate": 1.8624991788157074e-08,
+      "loss": 0.4957,
+      "step": 16490
+    },
+    {
+      "epoch": 0.9143300454394325,
+      "grad_norm": 1.3594955205917358,
+      "learning_rate": 1.838799066063157e-08,
+      "loss": 0.5069,
+      "step": 16500
+    },
+    {
+      "epoch": 0.9143300454394325,
+      "eval_robovqa_loss": 0.07888428866863251,
+      "step": 16500
+    },
+    {
+      "epoch": 0.9143300454394325,
+      "eval_agibot_loss": 0.46826931834220886,
+      "step": 16500
+    },
+    {
+      "epoch": 0.9143300454394325,
+      "eval_holoassist_loss": 0.3428939878940582,
+      "step": 16500
+    },
+    {
+      "epoch": 0.9143300454394325,
+      "eval_robofail_loss": 0.20727530121803284,
+      "step": 16500
+    },
+    {
+      "epoch": 0.914884184860911,
+      "grad_norm": 1.254654884338379,
+      "learning_rate": 1.8152478862977548e-08,
+      "loss": 0.509,
+      "step": 16510
+    },
+    {
+      "epoch": 0.9154383242823895,
+      "grad_norm": 1.186046838760376,
+      "learning_rate": 1.7918457123488105e-08,
+      "loss": 0.4965,
+      "step": 16520
+    },
+    {
+      "epoch": 0.9159924637038679,
+      "grad_norm": 1.2767221927642822,
+      "learning_rate": 1.768592616584874e-08,
+      "loss": 0.5058,
+      "step": 16530
+    },
+    {
+      "epoch": 0.9165466031253463,
+      "grad_norm": 1.4271248579025269,
+      "learning_rate": 1.7454886709134696e-08,
+      "loss": 0.514,
+      "step": 16540
+    },
+    {
+      "epoch": 0.9171007425468248,
+      "grad_norm": 1.3460348844528198,
+      "learning_rate": 1.7225339467808964e-08,
+      "loss": 0.4948,
+      "step": 16550
+    },
+    {
+      "epoch": 0.9176548819683032,
+      "grad_norm": 1.373888373374939,
+      "learning_rate": 1.699728515171994e-08,
+      "loss": 0.5032,
+      "step": 16560
+    },
+    {
+      "epoch": 0.9182090213897817,
+      "grad_norm": 1.3361270427703857,
+      "learning_rate": 1.6770724466099728e-08,
+      "loss": 0.5046,
+      "step": 16570
+    },
+    {
+      "epoch": 0.9187631608112601,
+      "grad_norm": 1.2750800848007202,
+      "learning_rate": 1.6545658111560944e-08,
+      "loss": 0.5072,
+      "step": 16580
+    },
+    {
+      "epoch": 0.9193173002327386,
+      "grad_norm": 1.327015995979309,
+      "learning_rate": 1.6322086784095633e-08,
+      "loss": 0.5049,
+      "step": 16590
+    },
+    {
+      "epoch": 0.919871439654217,
+      "grad_norm": 1.3848681449890137,
+      "learning_rate": 1.6100011175072537e-08,
+      "loss": 0.5115,
+      "step": 16600
+    },
+    {
+      "epoch": 0.919871439654217,
+      "eval_robovqa_loss": 0.07932386547327042,
+      "step": 16600
+    },
+    {
+      "epoch": 0.919871439654217,
+      "eval_agibot_loss": 0.47386589646339417,
+      "step": 16600
+    },
+    {
+      "epoch": 0.919871439654217,
+      "eval_holoassist_loss": 0.3420974314212799,
+      "step": 16600
+    },
+    {
+      "epoch": 0.919871439654217,
+      "eval_robofail_loss": 0.20680823922157288,
+      "step": 16600
+    },
+    {
+      "epoch": 0.9204255790756954,
+      "grad_norm": 1.335477352142334,
+      "learning_rate": 1.5879431971234703e-08,
+      "loss": 0.5054,
+      "step": 16610
+    },
+    {
+      "epoch": 0.9209797184971739,
+      "grad_norm": 1.2752649784088135,
+      "learning_rate": 1.566034985469833e-08,
+      "loss": 0.4978,
+      "step": 16620
+    },
+    {
+      "epoch": 0.9215338579186524,
+      "grad_norm": 1.310696005821228,
+      "learning_rate": 1.544276550294943e-08,
+      "loss": 0.4962,
+      "step": 16630
+    },
+    {
+      "epoch": 0.9220879973401308,
+      "grad_norm": 1.2542223930358887,
+      "learning_rate": 1.5226679588842827e-08,
+      "loss": 0.4993,
+      "step": 16640
+    },
+    {
+      "epoch": 0.9226421367616092,
+      "grad_norm": 1.46198570728302,
+      "learning_rate": 1.5012092780599394e-08,
+      "loss": 0.5034,
+      "step": 16650
+    },
+    {
+      "epoch": 0.9231962761830876,
+      "grad_norm": 1.2647993564605713,
+      "learning_rate": 1.4799005741804205e-08,
+      "loss": 0.5061,
+      "step": 16660
+    },
+    {
+      "epoch": 0.923750415604566,
+      "grad_norm": 1.3011325597763062,
+      "learning_rate": 1.4587419131404433e-08,
+      "loss": 0.5057,
+      "step": 16670
+    },
+    {
+      "epoch": 0.9243045550260446,
+      "grad_norm": 1.3477301597595215,
+      "learning_rate": 1.4377333603707464e-08,
+      "loss": 0.4925,
+      "step": 16680
+    },
+    {
+      "epoch": 0.924858694447523,
+      "grad_norm": 1.2643275260925293,
+      "learning_rate": 1.4168749808378566e-08,
+      "loss": 0.5096,
+      "step": 16690
+    },
+    {
+      "epoch": 0.9254128338690014,
+      "grad_norm": 1.3472421169281006,
+      "learning_rate": 1.3961668390439329e-08,
+      "loss": 0.5,
+      "step": 16700
+    },
+    {
+      "epoch": 0.9254128338690014,
+      "eval_robovqa_loss": 0.0792040303349495,
+      "step": 16700
+    },
+    {
+      "epoch": 0.9254128338690014,
+      "eval_agibot_loss": 0.46707916259765625,
+      "step": 16700
+    },
+    {
+      "epoch": 0.9254128338690014,
+      "eval_holoassist_loss": 0.34220725297927856,
+      "step": 16700
+    },
+    {
+      "epoch": 0.9254128338690014,
+      "eval_robofail_loss": 0.20692749321460724,
+      "step": 16700
+    },
+    {
+      "epoch": 0.9259669732904798,
+      "grad_norm": 1.2197116613388062,
+      "learning_rate": 1.3756089990265118e-08,
+      "loss": 0.5036,
+      "step": 16710
+    },
+    {
+      "epoch": 0.9265211127119584,
+      "grad_norm": 1.2766560316085815,
+      "learning_rate": 1.3552015243583681e-08,
+      "loss": 0.4924,
+      "step": 16720
+    },
+    {
+      "epoch": 0.9270752521334368,
+      "grad_norm": 1.303289771080017,
+      "learning_rate": 1.3349444781472763e-08,
+      "loss": 0.5019,
+      "step": 16730
+    },
+    {
+      "epoch": 0.9276293915549152,
+      "grad_norm": 1.2597707509994507,
+      "learning_rate": 1.3148379230358053e-08,
+      "loss": 0.5059,
+      "step": 16740
+    },
+    {
+      "epoch": 0.9281835309763936,
+      "grad_norm": 1.533216953277588,
+      "learning_rate": 1.2948819212012018e-08,
+      "loss": 0.4993,
+      "step": 16750
+    },
+    {
+      "epoch": 0.9287376703978721,
+      "grad_norm": 1.2297483682632446,
+      "learning_rate": 1.2750765343550852e-08,
+      "loss": 0.5028,
+      "step": 16760
+    },
+    {
+      "epoch": 0.9292918098193506,
+      "grad_norm": 1.4074506759643555,
+      "learning_rate": 1.2554218237433578e-08,
+      "loss": 0.5009,
+      "step": 16770
+    },
+    {
+      "epoch": 0.929845949240829,
+      "grad_norm": 1.348426342010498,
+      "learning_rate": 1.2359178501459455e-08,
+      "loss": 0.4921,
+      "step": 16780
+    },
+    {
+      "epoch": 0.9304000886623074,
+      "grad_norm": 1.2686290740966797,
+      "learning_rate": 1.2165646738766577e-08,
+      "loss": 0.4989,
+      "step": 16790
+    },
+    {
+      "epoch": 0.9309542280837859,
+      "grad_norm": 1.4175724983215332,
+      "learning_rate": 1.1973623547829604e-08,
+      "loss": 0.497,
+      "step": 16800
+    },
+    {
+      "epoch": 0.9309542280837859,
+      "eval_robovqa_loss": 0.07914873212575912,
+      "step": 16800
+    },
+    {
+      "epoch": 0.9309542280837859,
+      "eval_agibot_loss": 0.4693979024887085,
+      "step": 16800
+    },
+    {
+      "epoch": 0.9309542280837859,
+      "eval_holoassist_loss": 0.3408783972263336,
+      "step": 16800
+    },
+    {
+      "epoch": 0.9309542280837859,
+      "eval_robofail_loss": 0.20842097699642181,
+      "step": 16800
+    },
+    {
+      "epoch": 0.9315083675052643,
+      "grad_norm": 1.2252469062805176,
+      "learning_rate": 1.1783109522458311e-08,
+      "loss": 0.4994,
+      "step": 16810
+    },
+    {
+      "epoch": 0.9320625069267428,
+      "grad_norm": 1.2851723432540894,
+      "learning_rate": 1.159410525179544e-08,
+      "loss": 0.5008,
+      "step": 16820
+    },
+    {
+      "epoch": 0.9326166463482212,
+      "grad_norm": 1.3205325603485107,
+      "learning_rate": 1.1406611320315008e-08,
+      "loss": 0.4914,
+      "step": 16830
+    },
+    {
+      "epoch": 0.9331707857696997,
+      "grad_norm": 1.346598744392395,
+      "learning_rate": 1.12206283078205e-08,
+      "loss": 0.5065,
+      "step": 16840
+    },
+    {
+      "epoch": 0.9337249251911781,
+      "grad_norm": 1.2737990617752075,
+      "learning_rate": 1.1036156789442918e-08,
+      "loss": 0.5015,
+      "step": 16850
+    },
+    {
+      "epoch": 0.9342790646126565,
+      "grad_norm": 1.340154767036438,
+      "learning_rate": 1.0853197335639442e-08,
+      "loss": 0.5025,
+      "step": 16860
+    },
+    {
+      "epoch": 0.934833204034135,
+      "grad_norm": 1.242355227470398,
+      "learning_rate": 1.0671750512191047e-08,
+      "loss": 0.5108,
+      "step": 16870
+    },
+    {
+      "epoch": 0.9353873434556135,
+      "grad_norm": 1.2495098114013672,
+      "learning_rate": 1.0491816880201288e-08,
+      "loss": 0.5,
+      "step": 16880
+    },
+    {
+      "epoch": 0.9359414828770919,
+      "grad_norm": 1.344063639640808,
+      "learning_rate": 1.0313396996094237e-08,
+      "loss": 0.5022,
+      "step": 16890
+    },
+    {
+      "epoch": 0.9364956222985703,
+      "grad_norm": 1.3593281507492065,
+      "learning_rate": 1.013649141161299e-08,
+      "loss": 0.5036,
+      "step": 16900
+    },
+    {
+      "epoch": 0.9364956222985703,
+      "eval_robovqa_loss": 0.0798121765255928,
+      "step": 16900
+    },
+    {
+      "epoch": 0.9364956222985703,
+      "eval_agibot_loss": 0.46762222051620483,
+      "step": 16900
+    },
+    {
+      "epoch": 0.9364956222985703,
+      "eval_holoassist_loss": 0.34229886531829834,
+      "step": 16900
+    },
+    {
+      "epoch": 0.9364956222985703,
+      "eval_robofail_loss": 0.20631921291351318,
+      "step": 16900
+    },
+    {
+      "epoch": 0.9370497617200487,
+      "grad_norm": 1.2739533185958862,
+      "learning_rate": 9.961100673817669e-09,
+      "loss": 0.4973,
+      "step": 16910
+    },
+    {
+      "epoch": 0.9376039011415273,
+      "grad_norm": 1.2747890949249268,
+      "learning_rate": 9.787225325084025e-09,
+      "loss": 0.5023,
+      "step": 16920
+    },
+    {
+      "epoch": 0.9381580405630057,
+      "grad_norm": 1.2477951049804688,
+      "learning_rate": 9.61486590310162e-09,
+      "loss": 0.4996,
+      "step": 16930
+    },
+    {
+      "epoch": 0.9387121799844841,
+      "grad_norm": 1.2469936609268188,
+      "learning_rate": 9.444022940872265e-09,
+      "loss": 0.5071,
+      "step": 16940
+    },
+    {
+      "epoch": 0.9392663194059625,
+      "grad_norm": 1.3583348989486694,
+      "learning_rate": 9.274696966708129e-09,
+      "loss": 0.5037,
+      "step": 16950
+    },
+    {
+      "epoch": 0.939820458827441,
+      "grad_norm": 1.193283200263977,
+      "learning_rate": 9.106888504230304e-09,
+      "loss": 0.4929,
+      "step": 16960
+    },
+    {
+      "epoch": 0.9403745982489194,
+      "grad_norm": 1.3027455806732178,
+      "learning_rate": 8.940598072367356e-09,
+      "loss": 0.4944,
+      "step": 16970
+    },
+    {
+      "epoch": 0.9409287376703979,
+      "grad_norm": 1.280336618423462,
+      "learning_rate": 8.775826185353275e-09,
+      "loss": 0.4983,
+      "step": 16980
+    },
+    {
+      "epoch": 0.9414828770918763,
+      "grad_norm": 1.2994499206542969,
+      "learning_rate": 8.61257335272625e-09,
+      "loss": 0.4972,
+      "step": 16990
+    },
+    {
+      "epoch": 0.9420370165133548,
+      "grad_norm": 1.2502349615097046,
+      "learning_rate": 8.450840079326837e-09,
+      "loss": 0.5051,
+      "step": 17000
+    },
+    {
+      "epoch": 0.9420370165133548,
+      "eval_robovqa_loss": 0.07945544272661209,
+      "step": 17000
+    },
+    {
+      "epoch": 0.9420370165133548,
+      "eval_agibot_loss": 0.4695725739002228,
+      "step": 17000
+    },
+    {
+      "epoch": 0.9420370165133548,
+      "eval_holoassist_loss": 0.34149113297462463,
+      "step": 17000
+    },
+    {
+      "epoch": 0.9420370165133548,
+      "eval_robofail_loss": 0.20614288747310638,
+      "step": 17000
+    },
+    {
+      "epoch": 0.9425911559348332,
+      "grad_norm": 1.3239022493362427,
+      "learning_rate": 8.290626865296746e-09,
+      "loss": 0.4987,
+      "step": 17010
+    },
+    {
+      "epoch": 0.9431452953563116,
+      "grad_norm": 1.264885425567627,
+      "learning_rate": 8.13193420607694e-09,
+      "loss": 0.4971,
+      "step": 17020
+    },
+    {
+      "epoch": 0.9436994347777901,
+      "grad_norm": 1.2339860200881958,
+      "learning_rate": 7.974762592406092e-09,
+      "loss": 0.4965,
+      "step": 17030
+    },
+    {
+      "epoch": 0.9442535741992686,
+      "grad_norm": 1.328784704208374,
+      "learning_rate": 7.819112510319637e-09,
+      "loss": 0.5097,
+      "step": 17040
+    },
+    {
+      "epoch": 0.944807713620747,
+      "grad_norm": 1.3292211294174194,
+      "learning_rate": 7.664984441147381e-09,
+      "loss": 0.5013,
+      "step": 17050
+    },
+    {
+      "epoch": 0.9453618530422254,
+      "grad_norm": 1.3001716136932373,
+      "learning_rate": 7.512378861512791e-09,
+      "loss": 0.4996,
+      "step": 17060
+    },
+    {
+      "epoch": 0.9459159924637038,
+      "grad_norm": 1.2914053201675415,
+      "learning_rate": 7.36129624333115e-09,
+      "loss": 0.4981,
+      "step": 17070
+    },
+    {
+      "epoch": 0.9464701318851824,
+      "grad_norm": 1.3653786182403564,
+      "learning_rate": 7.211737053808009e-09,
+      "loss": 0.5046,
+      "step": 17080
+    },
+    {
+      "epoch": 0.9470242713066608,
+      "grad_norm": 1.3027645349502563,
+      "learning_rate": 7.0637017554380786e-09,
+      "loss": 0.4991,
+      "step": 17090
+    },
+    {
+      "epoch": 0.9475784107281392,
+      "grad_norm": 1.2498081922531128,
+      "learning_rate": 6.917190806003448e-09,
+      "loss": 0.4973,
+      "step": 17100
+    },
+    {
+      "epoch": 0.9475784107281392,
+      "eval_robovqa_loss": 0.07907669246196747,
+      "step": 17100
+    },
+    {
+      "epoch": 0.9475784107281392,
+      "eval_agibot_loss": 0.4675128161907196,
+      "step": 17100
+    },
+    {
+      "epoch": 0.9475784107281392,
+      "eval_holoassist_loss": 0.3421860635280609,
+      "step": 17100
+    },
+    {
+      "epoch": 0.9475784107281392,
+      "eval_robofail_loss": 0.20653896033763885,
+      "step": 17100
+    },
+    {
+      "epoch": 0.9481325501496176,
+      "grad_norm": 1.2087945938110352,
+      "learning_rate": 6.7722046585725335e-09,
+      "loss": 0.4989,
+      "step": 17110
+    },
+    {
+      "epoch": 0.9486866895710961,
+      "grad_norm": 1.1669977903366089,
+      "learning_rate": 6.628743761498245e-09,
+      "loss": 0.5071,
+      "step": 17120
+    },
+    {
+      "epoch": 0.9492408289925746,
+      "grad_norm": 1.3574289083480835,
+      "learning_rate": 6.486808558417045e-09,
+      "loss": 0.5034,
+      "step": 17130
+    },
+    {
+      "epoch": 0.949794968414053,
+      "grad_norm": 1.2516398429870605,
+      "learning_rate": 6.346399488247167e-09,
+      "loss": 0.508,
+      "step": 17140
+    },
+    {
+      "epoch": 0.9503491078355314,
+      "grad_norm": 1.2950613498687744,
+      "learning_rate": 6.207516985187622e-09,
+      "loss": 0.4986,
+      "step": 17150
+    },
+    {
+      "epoch": 0.9509032472570098,
+      "grad_norm": 1.2167352437973022,
+      "learning_rate": 6.070161478716529e-09,
+      "loss": 0.4884,
+      "step": 17160
+    },
+    {
+      "epoch": 0.9514573866784883,
+      "grad_norm": 1.2678173780441284,
+      "learning_rate": 5.934333393590119e-09,
+      "loss": 0.5046,
+      "step": 17170
+    },
+    {
+      "epoch": 0.9520115260999668,
+      "grad_norm": 1.3045835494995117,
+      "learning_rate": 5.800033149841121e-09,
+      "loss": 0.5011,
+      "step": 17180
+    },
+    {
+      "epoch": 0.9525656655214452,
+      "grad_norm": 1.3101907968521118,
+      "learning_rate": 5.667261162777659e-09,
+      "loss": 0.5006,
+      "step": 17190
+    },
+    {
+      "epoch": 0.9531198049429236,
+      "grad_norm": 1.2916492223739624,
+      "learning_rate": 5.536017842981799e-09,
+      "loss": 0.4948,
+      "step": 17200
+    },
+    {
+      "epoch": 0.9531198049429236,
+      "eval_robovqa_loss": 0.08043500781059265,
+      "step": 17200
+    },
+    {
+      "epoch": 0.9531198049429236,
+      "eval_agibot_loss": 0.46886998414993286,
+      "step": 17200
+    },
+    {
+      "epoch": 0.9531198049429236,
+      "eval_holoassist_loss": 0.3440666198730469,
+      "step": 17200
+    },
+    {
+      "epoch": 0.9531198049429236,
+      "eval_robofail_loss": 0.20499835908412933,
+      "step": 17200
+    },
+    {
+      "epoch": 0.9536739443644021,
+      "grad_norm": 1.2922749519348145,
+      "learning_rate": 5.406303596308504e-09,
+      "loss": 0.4923,
+      "step": 17210
+    },
+    {
+      "epoch": 0.9542280837858805,
+      "grad_norm": 1.1586236953735352,
+      "learning_rate": 5.278118823884181e-09,
+      "loss": 0.4985,
+      "step": 17220
+    },
+    {
+      "epoch": 0.954782223207359,
+      "grad_norm": 1.3929685354232788,
+      "learning_rate": 5.15146392210547e-09,
+      "loss": 0.5052,
+      "step": 17230
+    },
+    {
+      "epoch": 0.9553363626288374,
+      "grad_norm": 1.3838748931884766,
+      "learning_rate": 5.026339282638237e-09,
+      "loss": 0.4946,
+      "step": 17240
+    },
+    {
+      "epoch": 0.9558905020503159,
+      "grad_norm": 1.3531242609024048,
+      "learning_rate": 4.9027452924159664e-09,
+      "loss": 0.4973,
+      "step": 17250
+    },
+    {
+      "epoch": 0.9564446414717943,
+      "grad_norm": 1.434062123298645,
+      "learning_rate": 4.78068233363893e-09,
+      "loss": 0.5008,
+      "step": 17260
+    },
+    {
+      "epoch": 0.9569987808932727,
+      "grad_norm": 1.2690917253494263,
+      "learning_rate": 4.660150783772854e-09,
+      "loss": 0.5043,
+      "step": 17270
+    },
+    {
+      "epoch": 0.9575529203147511,
+      "grad_norm": 1.2824785709381104,
+      "learning_rate": 4.541151015547695e-09,
+      "loss": 0.5015,
+      "step": 17280
+    },
+    {
+      "epoch": 0.9581070597362297,
+      "grad_norm": 1.3362292051315308,
+      "learning_rate": 4.423683396956535e-09,
+      "loss": 0.4949,
+      "step": 17290
+    },
+    {
+      "epoch": 0.9586611991577081,
+      "grad_norm": 1.2386268377304077,
+      "learning_rate": 4.307748291254576e-09,
+      "loss": 0.4996,
+      "step": 17300
+    },
+    {
+      "epoch": 0.9586611991577081,
+      "eval_robovqa_loss": 0.0792182981967926,
+      "step": 17300
+    },
+    {
+      "epoch": 0.9586611991577081,
+      "eval_agibot_loss": 0.46768736839294434,
+      "step": 17300
+    },
+    {
+      "epoch": 0.9586611991577081,
+      "eval_holoassist_loss": 0.3400769829750061,
+      "step": 17300
+    },
+    {
+      "epoch": 0.9586611991577081,
+      "eval_robofail_loss": 0.20800350606441498,
+      "step": 17300
+    },
+    {
+      "epoch": 0.9592153385791865,
+      "grad_norm": 1.2726554870605469,
+      "learning_rate": 4.193346056957758e-09,
+      "loss": 0.5017,
+      "step": 17310
+    },
+    {
+      "epoch": 0.9597694780006649,
+      "grad_norm": 1.2675071954727173,
+      "learning_rate": 4.080477047841979e-09,
+      "loss": 0.5008,
+      "step": 17320
+    },
+    {
+      "epoch": 0.9603236174221434,
+      "grad_norm": 1.2689076662063599,
+      "learning_rate": 3.969141612941596e-09,
+      "loss": 0.5109,
+      "step": 17330
+    },
+    {
+      "epoch": 0.9608777568436219,
+      "grad_norm": 1.2972774505615234,
+      "learning_rate": 3.859340096548647e-09,
+      "loss": 0.5014,
+      "step": 17340
+    },
+    {
+      "epoch": 0.9614318962651003,
+      "grad_norm": 1.246842384338379,
+      "learning_rate": 3.7510728382117995e-09,
+      "loss": 0.5137,
+      "step": 17350
+    },
+    {
+      "epoch": 0.9619860356865787,
+      "grad_norm": 1.2148000001907349,
+      "learning_rate": 3.6443401727351277e-09,
+      "loss": 0.5015,
+      "step": 17360
+    },
+    {
+      "epoch": 0.9625401751080572,
+      "grad_norm": 1.238539457321167,
+      "learning_rate": 3.5391424301770556e-09,
+      "loss": 0.5019,
+      "step": 17370
+    },
+    {
+      "epoch": 0.9630943145295356,
+      "grad_norm": 1.3080482482910156,
+      "learning_rate": 3.4354799358495836e-09,
+      "loss": 0.5006,
+      "step": 17380
+    },
+    {
+      "epoch": 0.9636484539510141,
+      "grad_norm": 1.3060914278030396,
+      "learning_rate": 3.333353010317175e-09,
+      "loss": 0.504,
+      "step": 17390
+    },
+    {
+      "epoch": 0.9642025933724925,
+      "grad_norm": 1.1363999843597412,
+      "learning_rate": 3.2327619693954255e-09,
+      "loss": 0.5067,
+      "step": 17400
+    },
+    {
+      "epoch": 0.9642025933724925,
+      "eval_robovqa_loss": 0.07942011207342148,
+      "step": 17400
+    },
+    {
+      "epoch": 0.9642025933724925,
+      "eval_agibot_loss": 0.4680394232273102,
+      "step": 17400
+    },
+    {
+      "epoch": 0.9642025933724925,
+      "eval_holoassist_loss": 0.34095799922943115,
+      "step": 17400
+    },
+    {
+      "epoch": 0.9642025933724925,
+      "eval_robofail_loss": 0.2056218534708023,
+      "step": 17400
+    },
+    {
+      "epoch": 0.964756732793971,
+      "grad_norm": 1.1936854124069214,
+      "learning_rate": 3.133707124150786e-09,
+      "loss": 0.4893,
+      "step": 17410
+    },
+    {
+      "epoch": 0.9653108722154494,
+      "grad_norm": 1.371995449066162,
+      "learning_rate": 3.0361887808988963e-09,
+      "loss": 0.5018,
+      "step": 17420
+    },
+    {
+      "epoch": 0.9658650116369278,
+      "grad_norm": 1.2288304567337036,
+      "learning_rate": 2.940207241204029e-09,
+      "loss": 0.5045,
+      "step": 17430
+    },
+    {
+      "epoch": 0.9664191510584063,
+      "grad_norm": 1.324191689491272,
+      "learning_rate": 2.8457628018781487e-09,
+      "loss": 0.4986,
+      "step": 17440
+    },
+    {
+      "epoch": 0.9669732904798848,
+      "grad_norm": 1.2541767358779907,
+      "learning_rate": 2.7528557549797437e-09,
+      "loss": 0.4971,
+      "step": 17450
+    },
+    {
+      "epoch": 0.9675274299013632,
+      "grad_norm": 1.2267447710037231,
+      "learning_rate": 2.6614863878133277e-09,
+      "loss": 0.495,
+      "step": 17460
+    },
+    {
+      "epoch": 0.9680815693228416,
+      "grad_norm": 1.4325288534164429,
+      "learning_rate": 2.5716549829280508e-09,
+      "loss": 0.507,
+      "step": 17470
+    },
+    {
+      "epoch": 0.96863570874432,
+      "grad_norm": 1.2759631872177124,
+      "learning_rate": 2.483361818117313e-09,
+      "loss": 0.5029,
+      "step": 17480
+    },
+    {
+      "epoch": 0.9691898481657986,
+      "grad_norm": 1.2637205123901367,
+      "learning_rate": 2.396607166417597e-09,
+      "loss": 0.5006,
+      "step": 17490
+    },
+    {
+      "epoch": 0.969743987587277,
+      "grad_norm": 1.2998336553573608,
+      "learning_rate": 2.3113912961076344e-09,
+      "loss": 0.4989,
+      "step": 17500
+    },
+    {
+      "epoch": 0.969743987587277,
+      "eval_robovqa_loss": 0.07898278534412384,
+      "step": 17500
+    },
+    {
+      "epoch": 0.969743987587277,
+      "eval_agibot_loss": 0.46965765953063965,
+      "step": 17500
+    },
+    {
+      "epoch": 0.969743987587277,
+      "eval_holoassist_loss": 0.341774582862854,
+      "step": 17500
+    },
+    {
+      "epoch": 0.969743987587277,
+      "eval_robofail_loss": 0.20588012039661407,
+      "step": 17500
+    },
+    {
+      "epoch": 0.9702981270087554,
+      "grad_norm": 1.4253872632980347,
+      "learning_rate": 2.2277144707077976e-09,
+      "loss": 0.5015,
+      "step": 17510
+    },
+    {
+      "epoch": 0.9708522664302338,
+      "grad_norm": 1.2930989265441895,
+      "learning_rate": 2.1455769489790443e-09,
+      "loss": 0.5034,
+      "step": 17520
+    },
+    {
+      "epoch": 0.9714064058517123,
+      "grad_norm": 1.208019495010376,
+      "learning_rate": 2.0649789849221945e-09,
+      "loss": 0.5032,
+      "step": 17530
+    },
+    {
+      "epoch": 0.9719605452731908,
+      "grad_norm": 1.245316505432129,
+      "learning_rate": 1.9859208277772655e-09,
+      "loss": 0.5033,
+      "step": 17540
+    },
+    {
+      "epoch": 0.9725146846946692,
+      "grad_norm": 1.3489941358566284,
+      "learning_rate": 1.9084027220225284e-09,
+      "loss": 0.4925,
+      "step": 17550
+    },
+    {
+      "epoch": 0.9730688241161476,
+      "grad_norm": 1.395564317703247,
+      "learning_rate": 1.8324249073737863e-09,
+      "loss": 0.5011,
+      "step": 17560
+    },
+    {
+      "epoch": 0.9736229635376261,
+      "grad_norm": 1.3619478940963745,
+      "learning_rate": 1.7579876187837072e-09,
+      "loss": 0.5013,
+      "step": 17570
+    },
+    {
+      "epoch": 0.9741771029591045,
+      "grad_norm": 1.3213292360305786,
+      "learning_rate": 1.685091086441104e-09,
+      "loss": 0.5048,
+      "step": 17580
+    },
+    {
+      "epoch": 0.974731242380583,
+      "grad_norm": 1.335237979888916,
+      "learning_rate": 1.6137355357701555e-09,
+      "loss": 0.4997,
+      "step": 17590
+    },
+    {
+      "epoch": 0.9752853818020614,
+      "grad_norm": 1.3223797082901,
+      "learning_rate": 1.5439211874296865e-09,
+      "loss": 0.5024,
+      "step": 17600
+    },
+    {
+      "epoch": 0.9752853818020614,
+      "eval_robovqa_loss": 0.07963233441114426,
+      "step": 17600
+    },
+    {
+      "epoch": 0.9752853818020614,
+      "eval_agibot_loss": 0.4683992266654968,
+      "step": 17600
+    },
+    {
+      "epoch": 0.9752853818020614,
+      "eval_holoassist_loss": 0.341344952583313,
+      "step": 17600
+    },
+    {
+      "epoch": 0.9752853818020614,
+      "eval_robofail_loss": 0.2080458104610443,
+      "step": 17600
+    },
+    {
+      "epoch": 0.9758395212235399,
+      "grad_norm": 1.3255541324615479,
+      "learning_rate": 1.4756482573125562e-09,
+      "loss": 0.4962,
+      "step": 17610
+    },
+    {
+      "epoch": 0.9763936606450183,
+      "grad_norm": 1.195367455482483,
+      "learning_rate": 1.4089169565449365e-09,
+      "loss": 0.5021,
+      "step": 17620
+    },
+    {
+      "epoch": 0.9769478000664967,
+      "grad_norm": 1.2972179651260376,
+      "learning_rate": 1.3437274914857577e-09,
+      "loss": 0.5015,
+      "step": 17630
+    },
+    {
+      "epoch": 0.9775019394879751,
+      "grad_norm": 1.2350332736968994,
+      "learning_rate": 1.28008006372593e-09,
+      "loss": 0.4945,
+      "step": 17640
+    },
+    {
+      "epoch": 0.9780560789094536,
+      "grad_norm": 1.2943631410598755,
+      "learning_rate": 1.217974870087901e-09,
+      "loss": 0.4998,
+      "step": 17650
+    },
+    {
+      "epoch": 0.9786102183309321,
+      "grad_norm": 1.2925631999969482,
+      "learning_rate": 1.1574121026247108e-09,
+      "loss": 0.4991,
+      "step": 17660
+    },
+    {
+      "epoch": 0.9791643577524105,
+      "grad_norm": 1.286503553390503,
+      "learning_rate": 1.0983919486198812e-09,
+      "loss": 0.5059,
+      "step": 17670
+    },
+    {
+      "epoch": 0.9797184971738889,
+      "grad_norm": 1.299133539199829,
+      "learning_rate": 1.0409145905864726e-09,
+      "loss": 0.5055,
+      "step": 17680
+    },
+    {
+      "epoch": 0.9802726365953673,
+      "grad_norm": 2.1437478065490723,
+      "learning_rate": 9.84980206266528e-10,
+      "loss": 0.4981,
+      "step": 17690
+    },
+    {
+      "epoch": 0.9808267760168459,
+      "grad_norm": 1.328802227973938,
+      "learning_rate": 9.305889686307966e-10,
+      "loss": 0.4974,
+      "step": 17700
+    },
+    {
+      "epoch": 0.9808267760168459,
+      "eval_robovqa_loss": 0.07989518344402313,
+      "step": 17700
+    },
+    {
+      "epoch": 0.9808267760168459,
+      "eval_agibot_loss": 0.46887147426605225,
+      "step": 17700
+    },
+    {
+      "epoch": 0.9808267760168459,
+      "eval_holoassist_loss": 0.3394647538661957,
+      "step": 17700
+    },
+    {
+      "epoch": 0.9808267760168459,
+      "eval_robofail_loss": 0.20716767013072968,
+      "step": 17700
+    },
+    {
+      "epoch": 0.9813809154383243,
+      "grad_norm": 1.3513365983963013,
+      "learning_rate": 8.777410458779e-10,
+      "loss": 0.4976,
+      "step": 17710
+    },
+    {
+      "epoch": 0.9819350548598027,
+      "grad_norm": 1.3474825620651245,
+      "learning_rate": 8.264366014339996e-10,
+      "loss": 0.5026,
+      "step": 17720
+    },
+    {
+      "epoch": 0.9824891942812811,
+      "grad_norm": 1.3018909692764282,
+      "learning_rate": 7.766757939521306e-10,
+      "loss": 0.5004,
+      "step": 17730
+    },
+    {
+      "epoch": 0.9830433337027596,
+      "grad_norm": 1.2816892862319946,
+      "learning_rate": 7.2845877731198e-10,
+      "loss": 0.5032,
+      "step": 17740
+    },
+    {
+      "epoch": 0.9835974731242381,
+      "grad_norm": 1.2339839935302734,
+      "learning_rate": 6.81785700619053e-10,
+      "loss": 0.512,
+      "step": 17750
+    },
+    {
+      "epoch": 0.9841516125457165,
+      "grad_norm": 1.388525128364563,
+      "learning_rate": 6.366567082045637e-10,
+      "loss": 0.4975,
+      "step": 17760
+    },
+    {
+      "epoch": 0.9847057519671949,
+      "grad_norm": 1.295166254043579,
+      "learning_rate": 5.930719396247119e-10,
+      "loss": 0.495,
+      "step": 17770
+    },
+    {
+      "epoch": 0.9852598913886734,
+      "grad_norm": 1.2854894399642944,
+      "learning_rate": 5.510315296603507e-10,
+      "loss": 0.4947,
+      "step": 17780
+    },
+    {
+      "epoch": 0.9858140308101518,
+      "grad_norm": 1.3294284343719482,
+      "learning_rate": 5.105356083166534e-10,
+      "loss": 0.4996,
+      "step": 17790
+    },
+    {
+      "epoch": 0.9863681702316303,
+      "grad_norm": 1.250474452972412,
+      "learning_rate": 4.715843008226139e-10,
+      "loss": 0.5082,
+      "step": 17800
+    },
+    {
+      "epoch": 0.9863681702316303,
+      "eval_robovqa_loss": 0.07912957668304443,
+      "step": 17800
+    },
+    {
+      "epoch": 0.9863681702316303,
+      "eval_agibot_loss": 0.4679253399372101,
+      "step": 17800
+    },
+    {
+      "epoch": 0.9863681702316303,
+      "eval_holoassist_loss": 0.34156447649002075,
+      "step": 17800
+    },
+    {
+      "epoch": 0.9863681702316303,
+      "eval_robofail_loss": 0.20629844069480896,
+      "step": 17800
+    },
+    {
+      "epoch": 0.9869223096531087,
+      "grad_norm": 1.3902544975280762,
+      "learning_rate": 4.34177727630658e-10,
+      "loss": 0.4994,
+      "step": 17810
+    },
+    {
+      "epoch": 0.9874764490745872,
+      "grad_norm": 1.3149832487106323,
+      "learning_rate": 3.9831600441636583e-10,
+      "loss": 0.4986,
+      "step": 17820
+    },
+    {
+      "epoch": 0.9880305884960656,
+      "grad_norm": 1.3989335298538208,
+      "learning_rate": 3.639992420779725e-10,
+      "loss": 0.4963,
+      "step": 17830
+    },
+    {
+      "epoch": 0.988584727917544,
+      "grad_norm": 1.2852164506912231,
+      "learning_rate": 3.3122754673614584e-10,
+      "loss": 0.5034,
+      "step": 17840
+    },
+    {
+      "epoch": 0.9891388673390225,
+      "grad_norm": 1.7696255445480347,
+      "learning_rate": 3.000010197335978e-10,
+      "loss": 0.4999,
+      "step": 17850
+    },
+    {
+      "epoch": 0.989693006760501,
+      "grad_norm": 1.481856107711792,
+      "learning_rate": 2.7031975763469603e-10,
+      "loss": 0.5051,
+      "step": 17860
+    },
+    {
+      "epoch": 0.9902471461819794,
+      "grad_norm": 1.3472641706466675,
+      "learning_rate": 2.421838522254638e-10,
+      "loss": 0.5064,
+      "step": 17870
+    },
+    {
+      "epoch": 0.9908012856034578,
+      "grad_norm": 1.2500152587890625,
+      "learning_rate": 2.155933905128582e-10,
+      "loss": 0.5016,
+      "step": 17880
+    },
+    {
+      "epoch": 0.9913554250249362,
+      "grad_norm": 1.334788203239441,
+      "learning_rate": 1.9054845472488146e-10,
+      "loss": 0.4998,
+      "step": 17890
+    },
+    {
+      "epoch": 0.9919095644464148,
+      "grad_norm": 1.4211952686309814,
+      "learning_rate": 1.6704912231008117e-10,
+      "loss": 0.4997,
+      "step": 17900
+    },
+    {
+      "epoch": 0.9919095644464148,
+      "eval_robovqa_loss": 0.08021149039268494,
+      "step": 17900
+    },
+    {
+      "epoch": 0.9919095644464148,
+      "eval_agibot_loss": 0.47116827964782715,
+      "step": 17900
+    },
+    {
+      "epoch": 0.9919095644464148,
+      "eval_holoassist_loss": 0.3403693437576294,
+      "step": 17900
+    },
+    {
+      "epoch": 0.9919095644464148,
+      "eval_robofail_loss": 0.2066488265991211,
+      "step": 17900
+    },
+    {
+      "epoch": 0.9924637038678932,
+      "grad_norm": 1.3802025318145752,
+      "learning_rate": 1.450954659375503e-10,
+      "loss": 0.4998,
+      "step": 17910
+    },
+    {
+      "epoch": 0.9930178432893716,
+      "grad_norm": 1.3548295497894287,
+      "learning_rate": 1.2468755349631655e-10,
+      "loss": 0.499,
+      "step": 17920
+    },
+    {
+      "epoch": 0.99357198271085,
+      "grad_norm": 1.4142203330993652,
+      "learning_rate": 1.0582544809556448e-10,
+      "loss": 0.4941,
+      "step": 17930
+    },
+    {
+      "epoch": 0.9941261221323285,
+      "grad_norm": 1.2948188781738281,
+      "learning_rate": 8.850920806424689e-11,
+      "loss": 0.5036,
+      "step": 17940
+    },
+    {
+      "epoch": 0.994680261553807,
+      "grad_norm": 1.284259557723999,
+      "learning_rate": 7.27388869507517e-11,
+      "loss": 0.496,
+      "step": 17950
+    },
+    {
+      "epoch": 0.9952344009752854,
+      "grad_norm": 1.3740060329437256,
+      "learning_rate": 5.851453352306856e-11,
+      "loss": 0.4979,
+      "step": 17960
+    },
+    {
+      "epoch": 0.9957885403967638,
+      "grad_norm": 1.2768036127090454,
+      "learning_rate": 4.583619176834475e-11,
+      "loss": 0.5017,
+      "step": 17970
+    },
+    {
+      "epoch": 0.9963426798182423,
+      "grad_norm": 1.2601187229156494,
+      "learning_rate": 3.47039008928296e-11,
+      "loss": 0.4996,
+      "step": 17980
+    },
+    {
+      "epoch": 0.9968968192397207,
+      "grad_norm": 1.3407660722732544,
+      "learning_rate": 2.5117695321985598e-11,
+      "loss": 0.5002,
+      "step": 17990
+    },
+    {
+      "epoch": 0.9974509586611991,
+      "grad_norm": 1.213087558746338,
+      "learning_rate": 1.7077604699988756e-11,
+      "loss": 0.5022,
+      "step": 18000
+    },
+    {
+      "epoch": 0.9974509586611991,
+      "eval_robovqa_loss": 0.08005355298519135,
+      "step": 18000
+    },
+    {
+      "epoch": 0.9974509586611991,
+      "eval_agibot_loss": 0.4676314890384674,
+      "step": 18000
+    },
+    {
+      "epoch": 0.9974509586611991,
+      "eval_holoassist_loss": 0.3412254750728607,
+      "step": 18000
+    },
+    {
+      "epoch": 0.9974509586611991,
+      "eval_robofail_loss": 0.20574724674224854,
+      "step": 18000
+    },
+    {
+      "epoch": 0.9980050980826776,
+      "grad_norm": 1.249083161354065,
+      "learning_rate": 1.0583653889895128e-11,
+      "loss": 0.4993,
+      "step": 18010
+    },
+    {
+      "epoch": 0.9985592375041561,
+      "grad_norm": 1.3167768716812134,
+      "learning_rate": 5.635862973474293e-12,
+      "loss": 0.5006,
+      "step": 18020
+    },
+    {
+      "epoch": 0.9991133769256345,
+      "grad_norm": 1.4119501113891602,
+      "learning_rate": 2.2342472512093535e-12,
+      "loss": 0.5013,
+      "step": 18030
+    },
+    {
+      "epoch": 0.9996675163471129,
+      "grad_norm": 1.2403614521026611,
+      "learning_rate": 3.78817242296936e-13,
+      "loss": 0.5043,
+      "step": 18040
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 18046,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.065757871426765e+16,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}