CoolFace
Modelpublic

oddadmix/Nawah-Router-v3

sourceHugging Faceapache-2.0updated 26d agoView on Hugging Face
0likes82downloads
train_metrics.json1157 linesDownload Raw Back to root
1{2  "results": {3    "unseen_lanes": {4      "accuracy": 0.9304995345950977,5      "random_baseline": 0.2137126998901743,6      "n": 3223,7      "by_n_routes": {8        "2": {9          "acc": 0.954954954954955,10          "n": 11111        },12        "4": {13          "acc": 0.9449877750611247,14          "n": 81815        },16        "5": {17          "acc": 0.9423076923076923,18          "n": 62419        },20        "8": {21          "acc": 0.9341317365269461,22          "n": 33423        },24        "3": {25          "acc": 0.9104938271604939,26          "n": 32427        },28        "7": {29          "acc": 0.907928388746803,30          "n": 39131        },32        "6": {33          "acc": 0.9224952741020794,34          "n": 52935        },36        "9": {37          "acc": 0.8913043478260869,38          "n": 9239        }40      },41      "by_difficulty": {42        "easy": {43          "acc": 0.9478603049680275,44          "n": 203345        },46        "hard": {47          "acc": 0.9008403361344538,48          "n": 119049        }50      },51      "by_mode": {52        "topical": {53          "acc": 0.884,54          "n": 150055        },56        "mixed": {57          "acc": 0.9772549019607844,58          "n": 127559        },60        "axis": {61          "acc": 0.953125,62          "n": 44863        }64      }65    },66    "unseen_domain": {67      "accuracy": 0.6976068889556424,68      "random_baseline": 0.25206113185236045,69      "n": 9987,70      "by_n_routes": {71        "5": {72          "acc": 0.6869671132764921,73          "n": 164274        },75        "4": {76          "acc": 0.727150084317032,77          "n": 296578        },79        "3": {80          "acc": 0.7427476737821566,81          "n": 182782        },83        "2": {84          "acc": 0.7923497267759563,85          "n": 91586        },87        "6": {88          "acc": 0.6202393906420022,89          "n": 91990        },91        "7": {92          "acc": 0.5959595959595959,93          "n": 79294        },95        "8": {96          "acc": 0.5976608187134503,97          "n": 85598        },99        "9": {100          "acc": 0.6666666666666666,101          "n": 72102        }103      },104      "by_difficulty": {105        "easy": {106          "acc": 0.6926144520657203,107          "n": 6269108        },109        "hard": {110          "acc": 0.7060247444862829,111          "n": 3718112        }113      },114      "by_mode": {115        "topical": {116          "acc": 0.6976068889556424,117          "n": 9987118        }119      }120    },121    "unseen_axis": {122      "accuracy": 0.6129888942539836,123      "random_baseline": 0.23578575644002975,124      "n": 12426,125      "by_n_routes": {126        "4": {127          "acc": 0.541494632535095,128          "n": 4844129        },130        "5": {131          "acc": 0.6220699603317706,132          "n": 2773133        },134        "3": {135          "acc": 0.6389937106918239,136          "n": 2385137        },138        "6": {139          "acc": 0.7419898819561551,140          "n": 1186141        },142        "7": {143          "acc": 0.7033898305084746,144          "n": 944145        },146        "8": {147          "acc": 0.6836734693877551,148          "n": 294149        }150      },151      "by_difficulty": {152        "hard": {153          "acc": 0.6019339242546333,154          "n": 4964155        },156        "easy": {157          "acc": 0.6203430715625837,158          "n": 7462159        }160      },161      "by_mode": {162        "axis": {163          "acc": 0.6129888942539836,164          "n": 12426165        }166      }167    },168    "hard": {169      "accuracy": 0.9008403361344538,170      "random_baseline": 0.21085734293717487,171      "n": 1190,172      "by_n_routes": {173        "4": {174          "acc": 0.9182156133828996,175          "n": 269176        },177        "5": {178          "acc": 0.9043824701195219,179          "n": 251180        },181        "8": {182          "acc": 0.9291338582677166,183          "n": 127184        },185        "3": {186          "acc": 0.8571428571428571,187          "n": 126188        },189        "7": {190          "acc": 0.8662420382165605,191          "n": 157192        },193        "6": {194          "acc": 0.9045226130653267,195          "n": 199196        },197        "2": {198          "acc": 1.0,199          "n": 34200        },201        "9": {202          "acc": 0.8148148148148148,203          "n": 27204        }205      },206      "by_difficulty": {207        "hard": {208          "acc": 0.9008403361344538,209          "n": 1190210        }211      },212      "by_mode": {213        "topical": {214          "acc": 0.8300180831826401,215          "n": 553216        },217        "mixed": {218          "acc": 0.9680851063829787,219          "n": 470220        },221        "axis": {222          "acc": 0.9461077844311377,223          "n": 167224        }225      }226    }227  },228  "base_model": "oddadmix/50M-2048-Emhotob",229  "formulation": "routing_head",230  "log_history": [231    {232      "loss": 0.8975238800048828,233      "grad_norm": 11.03390121459961,234      "learning_rate": 7.4625e-05,235      "epoch": 0.030902348578491966,236      "step": 200237    },238    {239      "loss": 0.47223701477050783,240      "grad_norm": 11.903738975524902,241      "learning_rate": 0.000149625,242      "epoch": 0.06180469715698393,243      "step": 400244    },245    {246      "loss": 0.4622153091430664,247      "grad_norm": 6.198873996734619,248      "learning_rate": 0.000224625,249      "epoch": 0.09270704573547589,250      "step": 600251    },252    {253      "loss": 0.49717689514160157,254      "grad_norm": 4.740701198577881,255      "learning_rate": 0.000299625,256      "epoch": 0.12360939431396786,257      "step": 800258    },259    {260      "loss": 0.4837868118286133,261      "grad_norm": 3.6331026554107666,262      "learning_rate": 0.0002998012771863284,263      "epoch": 0.15451174289245984,264      "step": 1000265    },266    {267      "eval_unseen_lanes_loss": 0.5073266625404358,268      "eval_unseen_lanes_accuracy": 0.8163201985727583,269      "eval_unseen_lanes_runtime": 0.5548,270      "eval_unseen_lanes_samples_per_second": 5809.493,271      "eval_unseen_lanes_steps_per_second": 91.928,272      "epoch": 0.15451174289245984,273      "step": 1000274    },275    {276      "eval_unseen_domain_loss": 0.9687234163284302,277      "eval_unseen_domain_accuracy": 0.6065885651346751,278      "eval_unseen_domain_runtime": 1.5808,279      "eval_unseen_domain_samples_per_second": 6317.746,280      "eval_unseen_domain_steps_per_second": 99.318,281      "epoch": 0.15451174289245984,282      "step": 1000283    },284    {285      "eval_unseen_axis_loss": 1.1325784921646118,286      "eval_unseen_axis_accuracy": 0.5135200386286818,287      "eval_unseen_axis_runtime": 1.2551,288      "eval_unseen_axis_samples_per_second": 9900.252,289      "eval_unseen_axis_steps_per_second": 155.364,290      "epoch": 0.15451174289245984,291      "step": 1000292    },293    {294      "eval_hard_loss": 0.5369061231613159,295      "eval_hard_accuracy": 0.807563025210084,296      "eval_hard_runtime": 0.2574,297      "eval_hard_samples_per_second": 4623.428,298      "eval_hard_steps_per_second": 73.819,299      "epoch": 0.15451174289245984,300      "step": 1000301    },302    {303      "loss": 0.4474765396118164,304      "grad_norm": 9.281235694885254,305      "learning_rate": 0.00029920164204065217,306      "epoch": 0.18541409147095178,307      "step": 1200308    },309    {310      "loss": 0.4180959320068359,311      "grad_norm": 6.990286827087402,312      "learning_rate": 0.00029820269436406154,313      "epoch": 0.21631644004944375,314      "step": 1400315    },316    {317      "loss": 0.4012123489379883,318      "grad_norm": 2.00594425201416,319      "learning_rate": 0.00029680710766818503,320      "epoch": 0.24721878862793573,321      "step": 1600322    },323    {324      "loss": 0.39807247161865233,325      "grad_norm": 4.734849452972412,326      "learning_rate": 0.0002950186170007609,327      "epoch": 0.27812113720642767,328      "step": 1800329    },330    {331      "loss": 0.37501571655273436,332      "grad_norm": 2.752845048904419,333      "learning_rate": 0.0002928420089494242,334      "epoch": 0.30902348578491967,335      "step": 2000336    },337    {338      "eval_unseen_lanes_loss": 0.406538724899292,339      "eval_unseen_lanes_accuracy": 0.8569655600372323,340      "eval_unseen_lanes_runtime": 0.5556,341      "eval_unseen_lanes_samples_per_second": 5801.001,342      "eval_unseen_lanes_steps_per_second": 91.794,343      "epoch": 0.30902348578491967,344      "step": 2000345    },346    {347      "eval_unseen_domain_loss": 1.1472351551055908,348      "eval_unseen_domain_accuracy": 0.5958746370281366,349      "eval_unseen_domain_runtime": 1.575,350      "eval_unseen_domain_samples_per_second": 6340.999,351      "eval_unseen_domain_steps_per_second": 99.683,352      "epoch": 0.30902348578491967,353      "step": 2000354    },355    {356      "eval_unseen_axis_loss": 1.353710412979126,357      "eval_unseen_axis_accuracy": 0.49806856591018833,358      "eval_unseen_axis_runtime": 1.3077,359      "eval_unseen_axis_samples_per_second": 9502.1,360      "eval_unseen_axis_steps_per_second": 149.116,361      "epoch": 0.30902348578491967,362      "step": 2000363    },364    {365      "eval_hard_loss": 0.4623798727989197,366      "eval_hard_accuracy": 0.8378151260504202,367      "eval_hard_runtime": 0.2506,368      "eval_hard_samples_per_second": 4749.401,369      "eval_hard_steps_per_second": 75.831,370      "epoch": 0.30902348578491967,371      "step": 2000372    },373    {374      "loss": 0.3629868316650391,375      "grad_norm": 3.957963228225708,376      "learning_rate": 0.0002902831088312284,377      "epoch": 0.3399258343634116,378      "step": 2200379    },380    {381      "loss": 0.3620159912109375,382      "grad_norm": 2.070098638534546,383      "learning_rate": 0.00028734876510218624,384      "epoch": 0.37082818294190356,385      "step": 2400386    },387    {388      "loss": 0.33227527618408204,389      "grad_norm": 3.895007371902466,390      "learning_rate": 0.00028404683102855547,391      "epoch": 0.40173053152039556,392      "step": 2600393    },394    {395      "loss": 0.3365111541748047,396      "grad_norm": 3.198639154434204,397      "learning_rate": 0.0002803861436689221,398      "epoch": 0.4326328800988875,399      "step": 2800400    },401    {402      "loss": 0.31873382568359376,403      "grad_norm": 2.5327401161193848,404      "learning_rate": 0.0002763765002233335,405      "epoch": 0.4635352286773795,406      "step": 3000407    },408    {409      "eval_unseen_lanes_loss": 0.34330183267593384,410      "eval_unseen_lanes_accuracy": 0.8836487744337573,411      "eval_unseen_lanes_runtime": 0.553,412      "eval_unseen_lanes_samples_per_second": 5828.318,413      "eval_unseen_lanes_steps_per_second": 92.226,414      "epoch": 0.4635352286773795,415      "step": 3000416    },417    {418      "eval_unseen_domain_loss": 0.9895414113998413,419      "eval_unseen_domain_accuracy": 0.6378291779313107,420      "eval_unseen_domain_runtime": 1.5853,421      "eval_unseen_domain_samples_per_second": 6299.774,422      "eval_unseen_domain_steps_per_second": 99.035,423      "epoch": 0.4635352286773795,424      "step": 3000425    },426    {427      "eval_unseen_axis_loss": 1.0701825618743896,428      "eval_unseen_axis_accuracy": 0.5737164010944793,429      "eval_unseen_axis_runtime": 1.2712,430      "eval_unseen_axis_samples_per_second": 9774.846,431      "eval_unseen_axis_steps_per_second": 153.396,432      "epoch": 0.4635352286773795,433      "step": 3000434    },435    {436      "eval_hard_loss": 0.38013485074043274,437      "eval_hard_accuracy": 0.873109243697479,438      "eval_hard_runtime": 0.2542,439      "eval_hard_samples_per_second": 4681.85,440      "eval_hard_steps_per_second": 74.752,441      "epoch": 0.4635352286773795,442      "step": 3000443    },444    {445      "loss": 0.29870388031005857,446      "grad_norm": 2.18878173828125,447      "learning_rate": 0.00027202863181277745,448      "epoch": 0.49443757725587145,449      "step": 3200450    },451    {452      "loss": 0.30184398651123046,453      "grad_norm": 3.0138494968414307,454      "learning_rate": 0.00026735417475918285,455      "epoch": 0.5253399258343634,456      "step": 3400457    },458    {459      "loss": 0.32383251190185547,460      "grad_norm": 7.008852005004883,461      "learning_rate": 0.00026236563944280554,462      "epoch": 0.5562422744128553,463      "step": 3600464    },465    {466      "loss": 0.3035639762878418,467      "grad_norm": 2.690004825592041,468      "learning_rate": 0.0002570763768203474,469      "epoch": 0.5871446229913473,470      "step": 3800471    },472    {473      "loss": 0.2934083366394043,474      "grad_norm": 1.9328616857528687,475      "learning_rate": 0.0002515005426934173,476      "epoch": 0.6180469715698393,477      "step": 4000478    },479    {480      "eval_unseen_lanes_loss": 0.30602237582206726,481      "eval_unseen_lanes_accuracy": 0.8920260626745269,482      "eval_unseen_lanes_runtime": 0.5511,483      "eval_unseen_lanes_samples_per_second": 5848.268,484      "eval_unseen_lanes_steps_per_second": 92.542,485      "epoch": 0.6180469715698393,486      "step": 4000487    },488    {489      "eval_unseen_domain_loss": 0.983905553817749,490      "eval_unseen_domain_accuracy": 0.6368278762391109,491      "eval_unseen_domain_runtime": 1.5835,492      "eval_unseen_domain_samples_per_second": 6306.791,493      "eval_unseen_domain_steps_per_second": 99.146,494      "epoch": 0.6180469715698393,495      "step": 4000496    },497    {498      "eval_unseen_axis_loss": 1.2108259201049805,499      "eval_unseen_axis_accuracy": 0.5799130854659584,500      "eval_unseen_axis_runtime": 1.2562,501      "eval_unseen_axis_samples_per_second": 9891.534,502      "eval_unseen_axis_steps_per_second": 155.227,503      "epoch": 0.6180469715698393,504      "step": 4000505    },506    {507      "eval_hard_loss": 0.3655548691749573,508      "eval_hard_accuracy": 0.8756302521008403,509      "eval_hard_runtime": 0.2592,510      "eval_hard_samples_per_second": 4591.918,511      "eval_hard_steps_per_second": 73.316,512      "epoch": 0.6180469715698393,513      "step": 4000514    },515    {516      "loss": 0.2829387664794922,517      "grad_norm": 2.094444513320923,518      "learning_rate": 0.00024565305982296296,519      "epoch": 0.6489493201483313,520      "step": 4200521    },522    {523      "loss": 0.2714531135559082,524      "grad_norm": 2.0051519870758057,525      "learning_rate": 0.00023954957799106823,526      "epoch": 0.6798516687268232,527      "step": 4400528    },529    {530      "loss": 0.26623159408569336,531      "grad_norm": 2.2412750720977783,532      "learning_rate": 0.00023320643211700308,533      "epoch": 0.7107540173053152,534      "step": 4600535    },536    {537      "loss": 0.2522362518310547,538      "grad_norm": 1.7817517518997192,539      "learning_rate": 0.000226640598539622,540      "epoch": 0.7416563658838071,541      "step": 4800542    },543    {544      "loss": 0.2480511474609375,545      "grad_norm": 5.033050537109375,546      "learning_rate": 0.00021986964958311316,547      "epoch": 0.7725587144622992,548      "step": 5000549    },550    {551      "eval_unseen_lanes_loss": 0.27606716752052307,552      "eval_unseen_lanes_accuracy": 0.9053676698727893,553      "eval_unseen_lanes_runtime": 0.552,554      "eval_unseen_lanes_samples_per_second": 5838.329,555      "eval_unseen_lanes_steps_per_second": 92.384,556      "epoch": 0.7725587144622992,557      "step": 5000558    },559    {560      "eval_unseen_domain_loss": 0.96076899766922,561      "eval_unseen_domain_accuracy": 0.6541503955141684,562      "eval_unseen_domain_runtime": 1.5806,563      "eval_unseen_domain_samples_per_second": 6318.291,564      "eval_unseen_domain_steps_per_second": 99.326,565      "epoch": 0.7725587144622992,566      "step": 5000567    },568    {569      "eval_unseen_axis_loss": 1.4210246801376343,570      "eval_unseen_axis_accuracy": 0.5544825366167713,571      "eval_unseen_axis_runtime": 1.2599,572      "eval_unseen_axis_samples_per_second": 9862.704,573      "eval_unseen_axis_steps_per_second": 154.774,574      "epoch": 0.7725587144622992,575      "step": 5000576    },577    {578      "eval_hard_loss": 0.3613678812980652,579      "eval_hard_accuracy": 0.8857142857142857,580      "eval_hard_runtime": 0.2606,581      "eval_hard_samples_per_second": 4565.623,582      "eval_hard_steps_per_second": 72.897,583      "epoch": 0.7725587144622992,584      "step": 5000585    },586    {587      "loss": 0.24242013931274414,588      "grad_norm": 2.4511566162109375,589      "learning_rate": 0.00021291170652769533,590      "epoch": 0.8034610630407911,591      "step": 5200592    },593    {594      "loss": 0.2558748817443848,595      "grad_norm": 1.6439733505249023,596      "learning_rate": 0.00020578539111112834,597      "epoch": 0.8343634116192831,598      "step": 5400599    },600    {601      "loss": 0.24663785934448243,602      "grad_norm": 3.4409291744232178,603      "learning_rate": 0.00019850977569083514,604      "epoch": 0.865265760197775,605      "step": 5600606    },607    {608      "loss": 0.24483919143676758,609      "grad_norm": 3.071554183959961,610      "learning_rate": 0.00019110433220001754,611      "epoch": 0.896168108776267,612      "step": 5800613    },614    {615      "loss": 0.23662748336791992,616      "grad_norm": 2.491429567337036,617      "learning_rate": 0.0001835888800343768,618      "epoch": 0.927070457354759,619      "step": 6000620    },621    {622      "eval_unseen_lanes_loss": 0.2571614384651184,623      "eval_unseen_lanes_accuracy": 0.9081600992863792,624      "eval_unseen_lanes_runtime": 0.5519,625      "eval_unseen_lanes_samples_per_second": 5839.58,626      "eval_unseen_lanes_steps_per_second": 92.404,627      "epoch": 0.927070457354759,628      "step": 6000629    },630    {631      "eval_unseen_domain_loss": 0.9326336979866028,632      "eval_unseen_domain_accuracy": 0.6525483128066486,633      "eval_unseen_domain_runtime": 1.5742,634      "eval_unseen_domain_samples_per_second": 6344.232,635      "eval_unseen_domain_steps_per_second": 99.734,636      "epoch": 0.927070457354759,637      "step": 6000638    },639    {640      "eval_unseen_axis_loss": 1.3221510648727417,641      "eval_unseen_axis_accuracy": 0.5856269113149847,642      "eval_unseen_axis_runtime": 1.2492,643      "eval_unseen_axis_samples_per_second": 9946.782,644      "eval_unseen_axis_steps_per_second": 156.094,645      "epoch": 0.927070457354759,646      "step": 6000647    },648    {649      "eval_hard_loss": 0.33343571424484253,650      "eval_hard_accuracy": 0.888235294117647,651      "eval_hard_runtime": 0.2587,652      "eval_hard_samples_per_second": 4600.467,653      "eval_hard_steps_per_second": 73.453,654      "epoch": 0.927070457354759,655      "step": 6000656    },657    {658      "loss": 0.23579851150512696,659      "grad_norm": 2.5624895095825195,660      "learning_rate": 0.00017598353300890969,661      "epoch": 0.957972805933251,662      "step": 6200663    },664    {665      "loss": 0.21076148986816406,666      "grad_norm": 0.9841051697731018,667      "learning_rate": 0.00016830864552674267,668      "epoch": 0.9888751545117429,669      "step": 6400670    },671    {672      "loss": 0.17771633148193358,673      "grad_norm": 0.9096925258636475,674      "learning_rate": 0.00016058475810407297,675      "epoch": 1.019777503090235,676      "step": 6600677    },678    {679      "loss": 0.14182785034179687,680      "grad_norm": 3.3751485347747803,681      "learning_rate": 0.00015283254239701005,682      "epoch": 1.0506798516687268,683      "step": 6800684    },685    {686      "loss": 0.1347336483001709,687      "grad_norm": 0.5445594787597656,688      "learning_rate": 0.00014507274587744398,689      "epoch": 1.0815822002472189,690      "step": 7000691    },692    {693      "eval_unseen_lanes_loss": 0.2655007839202881,694      "eval_unseen_lanes_accuracy": 0.9162271175923054,695      "eval_unseen_lanes_runtime": 0.5535,696      "eval_unseen_lanes_samples_per_second": 5823.011,697      "eval_unseen_lanes_steps_per_second": 92.142,698      "epoch": 1.0815822002472189,699      "step": 7000700    },701    {702      "eval_unseen_domain_loss": 1.0792171955108643,703      "eval_unseen_domain_accuracy": 0.6691699208971663,704      "eval_unseen_domain_runtime": 1.5827,705      "eval_unseen_domain_samples_per_second": 6310.28,706      "eval_unseen_domain_steps_per_second": 99.2,707      "epoch": 1.0815822002472189,708      "step": 7000709    },710    {711      "eval_unseen_axis_loss": 1.5451632738113403,712      "eval_unseen_axis_accuracy": 0.5832930951231289,713      "eval_unseen_axis_runtime": 1.2555,714      "eval_unseen_axis_samples_per_second": 9897.184,715      "eval_unseen_axis_steps_per_second": 155.316,716      "epoch": 1.0815822002472189,717      "step": 7000718    },719    {720      "eval_hard_loss": 0.3464004695415497,721      "eval_hard_accuracy": 0.8966386554621849,722      "eval_hard_runtime": 0.2591,723      "eval_hard_samples_per_second": 4593.088,724      "eval_hard_steps_per_second": 73.335,725      "epoch": 1.0815822002472189,726      "step": 7000727    },728    {729      "loss": 0.13339081764221192,730      "grad_norm": 2.522960901260376,731      "learning_rate": 0.00013732613630600555,732      "epoch": 1.1124845488257107,733      "step": 7200734    },735    {736      "loss": 0.11780735015869141,737      "grad_norm": 1.1648684740066528,738      "learning_rate": 0.0001296134461507269,739      "epoch": 1.1433868974042027,740      "step": 7400741    },742    {743      "loss": 0.13406469345092772,744      "grad_norm": 1.038225531578064,745      "learning_rate": 0.00012195531710015758,746      "epoch": 1.1742892459826946,747      "step": 7600748    },749    {750      "loss": 0.1309179973602295,751      "grad_norm": 1.6545023918151855,752      "learning_rate": 0.00011437224481943516,753      "epoch": 1.2051915945611866,754      "step": 7800755    },756    {757      "loss": 0.13036264419555665,758      "grad_norm": 1.7858803272247314,759      "learning_rate": 0.00010688452409716325,760      "epoch": 1.2360939431396787,761      "step": 8000762    },763    {764      "eval_unseen_lanes_loss": 0.26563557982444763,765      "eval_unseen_lanes_accuracy": 0.9162271175923054,766      "eval_unseen_lanes_runtime": 0.5528,767      "eval_unseen_lanes_samples_per_second": 5830.741,768      "eval_unseen_lanes_steps_per_second": 92.264,769      "epoch": 1.2360939431396787,770      "step": 8000771    },772    {773      "eval_unseen_domain_loss": 1.1347131729125977,774      "eval_unseen_domain_accuracy": 0.6633623710824071,775      "eval_unseen_domain_runtime": 1.5784,776      "eval_unseen_domain_samples_per_second": 6327.271,777      "eval_unseen_domain_steps_per_second": 99.467,778      "epoch": 1.2360939431396787,779      "step": 8000780    },781    {782      "eval_unseen_axis_loss": 1.7664040327072144,783      "eval_unseen_axis_accuracy": 0.5963302752293578,784      "eval_unseen_axis_runtime": 1.2484,785      "eval_unseen_axis_samples_per_second": 9953.248,786      "eval_unseen_axis_steps_per_second": 156.195,787      "epoch": 1.2360939431396787,788      "step": 8000789    },790    {791      "eval_hard_loss": 0.34612923860549927,792      "eval_hard_accuracy": 0.8957983193277311,793      "eval_hard_runtime": 0.2561,794      "eval_hard_samples_per_second": 4646.719,795      "eval_hard_steps_per_second": 74.191,796      "epoch": 1.2360939431396787,797      "step": 8000798    },799    {800      "loss": 0.13536205291748046,801      "grad_norm": 1.5142947435379028,802      "learning_rate": 9.95121945299005e-05,803      "epoch": 1.2669962917181705,804      "step": 8200805    },806    {807      "loss": 0.12326237678527832,808      "grad_norm": 5.0126051902771,809      "learning_rate": 9.227498688962761e-05,810      "epoch": 1.2978986402966626,811      "step": 8400812    },813    {814      "loss": 0.12851189613342284,815      "grad_norm": 3.0441036224365234,816      "learning_rate": 8.519227031772971e-05,817      "epoch": 1.3288009888751544,818      "step": 8600819    },820    {821      "loss": 0.11189216613769531,822      "grad_norm": 1.7037580013275146,823      "learning_rate": 7.828300048682143e-05,824      "epoch": 1.3597033374536465,825      "step": 8800826    },827    {828      "loss": 0.12371747016906738,829      "grad_norm": 0.7055680751800537,830      "learning_rate": 7.156566886914933e-05,831      "epoch": 1.3906056860321385,832      "step": 9000833    },834    {835      "eval_unseen_lanes_loss": 0.23261435329914093,836      "eval_unseen_lanes_accuracy": 0.9221222463543283,837      "eval_unseen_lanes_runtime": 0.5543,838      "eval_unseen_lanes_samples_per_second": 5814.13,839      "eval_unseen_lanes_steps_per_second": 92.001,840      "epoch": 1.3906056860321385,841      "step": 9000842    },843    {844      "eval_unseen_domain_loss": 0.9834672808647156,845      "eval_unseen_domain_accuracy": 0.6840893161109443,846      "eval_unseen_domain_runtime": 1.579,847      "eval_unseen_domain_samples_per_second": 6324.919,848      "eval_unseen_domain_steps_per_second": 99.43,849      "epoch": 1.3906056860321385,850      "step": 9000851    },852    {853      "eval_unseen_axis_loss": 1.5792756080627441,854      "eval_unseen_axis_accuracy": 0.593996459037502,855      "eval_unseen_axis_runtime": 1.2867,856      "eval_unseen_axis_samples_per_second": 9657.248,857      "eval_unseen_axis_steps_per_second": 151.55,858      "epoch": 1.3906056860321385,859      "step": 9000860    },861    {862      "eval_hard_loss": 0.31886979937553406,863      "eval_hard_accuracy": 0.9,864      "eval_hard_runtime": 0.2641,865      "eval_hard_samples_per_second": 4505.693,866      "eval_hard_steps_per_second": 71.94,867      "epoch": 1.3906056860321385,868      "step": 9000869    },870    {871      "loss": 0.11327756881713867,872      "grad_norm": 2.5306780338287354,873      "learning_rate": 6.505825324734749e-05,874      "epoch": 1.4215080346106304,875      "step": 9200876    },877    {878      "loss": 0.11213048934936523,879      "grad_norm": 3.7060656547546387,880      "learning_rate": 5.8778169599994366e-05,881      "epoch": 1.4524103831891224,882      "step": 9400883    },884    {885      "loss": 0.11299308776855468,886      "grad_norm": 2.2346725463867188,887      "learning_rate": 5.274222549074192e-05,888      "epoch": 1.4833127317676142,889      "step": 9600890    },891    {892      "loss": 0.10625232696533203,893      "grad_norm": 2.6063334941864014,894      "learning_rate": 4.696657508576255e-05,895      "epoch": 1.5142150803461063,896      "step": 9800897    },898    {899      "loss": 0.10512189865112305,900      "grad_norm": 4.146987438201904,901      "learning_rate": 4.146667591990211e-05,902      "epoch": 1.5451174289245984,903      "step": 10000904    },905    {906      "eval_unseen_lanes_loss": 0.2341764271259308,907      "eval_unseen_lanes_accuracy": 0.9286379149860379,908      "eval_unseen_lanes_runtime": 0.5514,909      "eval_unseen_lanes_samples_per_second": 5845.486,910      "eval_unseen_lanes_steps_per_second": 92.498,911      "epoch": 1.5451174289245984,912      "step": 10000913    },914    {915      "eval_unseen_domain_loss": 1.1408356428146362,916      "eval_unseen_domain_accuracy": 0.6845899669570441,917      "eval_unseen_domain_runtime": 1.5844,918      "eval_unseen_domain_samples_per_second": 6303.222,919      "eval_unseen_domain_steps_per_second": 99.089,920      "epoch": 1.5451174289245984,921      "step": 10000922    },923    {924      "eval_unseen_axis_loss": 1.6913553476333618,925      "eval_unseen_axis_accuracy": 0.6064703042008691,926      "eval_unseen_axis_runtime": 1.2705,927      "eval_unseen_axis_samples_per_second": 9780.492,928      "eval_unseen_axis_steps_per_second": 153.484,929      "epoch": 1.5451174289245984,930      "step": 10000931    },932    {933      "eval_hard_loss": 0.3436157703399658,934      "eval_hard_accuracy": 0.9042016806722689,935      "eval_hard_runtime": 0.2553,936      "eval_hard_samples_per_second": 4661.691,937      "eval_hard_steps_per_second": 74.43,938      "epoch": 1.5451174289245984,939      "step": 10000940    },941    {942      "loss": 0.10526069641113281,943      "grad_norm": 1.0956754684448242,944      "learning_rate": 3.625724752724645e-05,945      "epoch": 1.5760197775030902,946      "step": 10200947    },948    {949      "loss": 0.0974786376953125,950      "grad_norm": 1.1272306442260742,951      "learning_rate": 3.135223204682058e-05,952      "epoch": 1.6069221260815822,953      "step": 10400954    },955    {956      "loss": 0.101556396484375,957      "grad_norm": 3.4264066219329834,958      "learning_rate": 2.6764756908851686e-05,959      "epoch": 1.637824474660074,960      "step": 10600961    },962    {963      "loss": 0.10870396614074707,964      "grad_norm": 1.7836824655532837,965      "learning_rate": 2.2507099701460036e-05,966      "epoch": 1.6687268232385661,967      "step": 10800968    },969    {970      "loss": 0.09275879859924316,971      "grad_norm": 1.3714878559112549,972      "learning_rate": 1.8590655311806385e-05,973      "epoch": 1.6996291718170582,974      "step": 11000975    },976    {977      "eval_unseen_lanes_loss": 0.21319718658924103,978      "eval_unseen_lanes_accuracy": 0.9304995345950977,979      "eval_unseen_lanes_runtime": 0.5553,980      "eval_unseen_lanes_samples_per_second": 5804.033,981      "eval_unseen_lanes_steps_per_second": 91.842,982      "epoch": 1.6996291718170582,983      "step": 11000984    },985    {986      "eval_unseen_domain_loss": 1.022688627243042,987      "eval_unseen_domain_accuracy": 0.6976068889556424,988      "eval_unseen_domain_runtime": 1.5825,989      "eval_unseen_domain_samples_per_second": 6310.792,990      "eval_unseen_domain_steps_per_second": 99.208,991      "epoch": 1.6996291718170582,992      "step": 11000993    },994    {995      "eval_unseen_axis_loss": 1.5096590518951416,996      "eval_unseen_axis_accuracy": 0.6129888942539836,997      "eval_unseen_axis_runtime": 1.2878,998      "eval_unseen_axis_samples_per_second": 9649.329,999      "eval_unseen_axis_steps_per_second": 151.426,1000      "epoch": 1.6996291718170582,1001      "step": 110001002    },1003    {1004      "eval_hard_loss": 0.31117716431617737,1005      "eval_hard_accuracy": 0.9008403361344538,1006      "eval_hard_runtime": 0.2589,1007      "eval_hard_samples_per_second": 4596.879,1008      "eval_hard_steps_per_second": 73.396,1009      "epoch": 1.6996291718170582,1010      "step": 110001011    },1012    {1013      "loss": 0.09593048095703124,1014      "grad_norm": 3.177462100982666,1015      "learning_rate": 1.5025905429636426e-05,1016      "epoch": 1.73053152039555,1017      "step": 112001018    },1019    {1020      "loss": 0.09772712707519532,1021      "grad_norm": 2.102013349533081,1022      "learning_rate": 1.1822390494841139e-05,1023      "epoch": 1.7614338689740419,1024      "step": 114001025    },1026    {1027      "loss": 0.08881115913391113,1028      "grad_norm": 0.3349379897117615,1029      "learning_rate": 8.98868416411046e-06,1030      "epoch": 1.792336217552534,1031      "step": 116001032    },1033    {1034      "loss": 0.09125228881835938,1035      "grad_norm": 1.5826376676559448,1036      "learning_rate": 6.532370365015982e-06,1037      "epoch": 1.823238566131026,1038      "step": 118001039    },1040    {1041      "loss": 0.1032994556427002,1042      "grad_norm": 2.430258274078369,1043      "learning_rate": 4.460022998933166e-06,1044      "epoch": 1.854140914709518,1045      "step": 120001046    },1047    {1048      "eval_unseen_lanes_loss": 0.21029867231845856,1049      "eval_unseen_lanes_accuracy": 0.9329816940738442,1050      "eval_unseen_lanes_runtime": 0.5539,1051      "eval_unseen_lanes_samples_per_second": 5818.497,1052      "eval_unseen_lanes_steps_per_second": 92.071,1053      "epoch": 1.854140914709518,1054      "step": 120001055    },1056    {1057      "eval_unseen_domain_loss": 1.0129146575927734,1058      "eval_unseen_domain_accuracy": 0.700010013016922,1059      "eval_unseen_domain_runtime": 1.5764,1060      "eval_unseen_domain_samples_per_second": 6335.382,1061      "eval_unseen_domain_steps_per_second": 99.595,1062      "epoch": 1.854140914709518,1063      "step": 120001064    },1065    {1066      "eval_unseen_axis_loss": 1.5864043235778809,1067      "eval_unseen_axis_accuracy": 0.6117817479478512,1068      "eval_unseen_axis_runtime": 1.2684,1069      "eval_unseen_axis_samples_per_second": 9796.563,1070      "eval_unseen_axis_steps_per_second": 153.737,1071      "epoch": 1.854140914709518,1072      "step": 120001073    },1074    {1075      "eval_hard_loss": 0.30691978335380554,1076      "eval_hard_accuracy": 0.9084033613445378,1077      "eval_hard_runtime": 0.2602,1078      "eval_hard_samples_per_second": 4573.644,1079      "eval_hard_steps_per_second": 73.025,1080      "epoch": 1.854140914709518,1081      "step": 120001082    },1083    {1084      "loss": 0.08737665176391601,1085      "grad_norm": 1.072799801826477,1086      "learning_rate": 2.7771883471252034e-06,1087      "epoch": 1.8850432632880099,1088      "step": 122001089    },1090    {1091      "loss": 0.08102713584899902,1092      "grad_norm": 1.5238364934921265,1093      "learning_rate": 1.4883702270751197e-06,1094      "epoch": 1.9159456118665017,1095      "step": 124001096    },1097    {1098      "loss": 0.09933463096618653,1099      "grad_norm": 0.16273926198482513,1100      "learning_rate": 5.970179387931151e-07,1101      "epoch": 1.9468479604449938,1102      "step": 126001103    },1104    {1105      "loss": 0.10335576057434082,1106      "grad_norm": 0.17151758074760437,1107      "learning_rate": 1.0551703335823225e-07,1108      "epoch": 1.9777503090234858,1109      "step": 128001110    },1111    {1112      "eval_unseen_lanes_loss": 0.21011516451835632,1113      "eval_unseen_lanes_accuracy": 0.9336022339435309,1114      "eval_unseen_lanes_runtime": 0.5534,1115      "eval_unseen_lanes_samples_per_second": 5823.753,1116      "eval_unseen_lanes_steps_per_second": 92.154,1117      "epoch": 2.0,1118      "step": 129441119    },1120    {1121      "eval_unseen_domain_loss": 1.0266432762145996,1122      "eval_unseen_domain_accuracy": 0.7009111845399019,1123      "eval_unseen_domain_runtime": 1.5817,1124      "eval_unseen_domain_samples_per_second": 6314.155,1125      "eval_unseen_domain_steps_per_second": 99.261,1126      "epoch": 2.0,1127      "step": 129441128    },1129    {1130      "eval_unseen_axis_loss": 1.6092302799224854,1131      "eval_unseen_axis_accuracy": 0.6111379365845807,1132      "eval_unseen_axis_runtime": 1.2767,1133      "eval_unseen_axis_samples_per_second": 9732.892,1134      "eval_unseen_axis_steps_per_second": 152.737,1135      "epoch": 2.0,1136      "step": 129441137    },1138    {1139      "eval_hard_loss": 0.3078996539115906,1140      "eval_hard_accuracy": 0.9084033613445378,1141      "eval_hard_runtime": 0.2718,1142      "eval_hard_samples_per_second": 4377.996,1143      "eval_hard_steps_per_second": 69.901,1144      "epoch": 2.0,1145      "step": 129441146    },1147    {1148      "train_runtime": 262.1825,1149      "train_samples_per_second": 1579.793,1150      "train_steps_per_second": 49.37,1151      "total_flos": 0.0,1152      "train_loss": 0.2282541081106707,1153      "epoch": 2.0,1154      "step": 129441155    }1156  ]1157}