275 lines
10 KiB
JSON
275 lines
10 KiB
JSON
{
|
|
"best_global_step": null,
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 0.026647125441343015,
|
|
"eval_steps": 10,
|
|
"global_step": 50,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"entropy": 2.2245737984776497,
|
|
"epoch": 0.005329425088268603,
|
|
"grad_norm": 12.264137268066406,
|
|
"learning_rate": 1.8e-05,
|
|
"loss": 2.4835,
|
|
"mean_token_accuracy": 0.5454559613019228,
|
|
"num_tokens": 13377.0,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.005329425088268603,
|
|
"eval_backdoor_activation_score": 0.08,
|
|
"eval_backdoor_robustness_score": 0.92,
|
|
"eval_distance_l1": 0.13493727147579193,
|
|
"eval_distance_l10": 2.5394492149353027,
|
|
"eval_distance_l11": 2.593096971511841,
|
|
"eval_distance_l12": 2.7151217460632324,
|
|
"eval_distance_l13": 2.850677013397217,
|
|
"eval_distance_l14": 2.8369364738464355,
|
|
"eval_distance_l15": 2.805739164352417,
|
|
"eval_distance_l16": 2.8453636169433594,
|
|
"eval_distance_l17": 3.103731632232666,
|
|
"eval_distance_l18": 3.507354736328125,
|
|
"eval_distance_l19": 3.8099584579467773,
|
|
"eval_distance_l2": 0.5666664838790894,
|
|
"eval_distance_l20": 4.30399751663208,
|
|
"eval_distance_l21": 5.194124698638916,
|
|
"eval_distance_l22": 6.001777172088623,
|
|
"eval_distance_l23": 7.380727291107178,
|
|
"eval_distance_l24": 36.58632278442383,
|
|
"eval_distance_l3": 0.714788019657135,
|
|
"eval_distance_l4": 1.1566619873046875,
|
|
"eval_distance_l5": 1.7926262617111206,
|
|
"eval_distance_l6": 2.013216257095337,
|
|
"eval_distance_l7": 2.7371954917907715,
|
|
"eval_distance_l8": 2.5304036140441895,
|
|
"eval_distance_l9": 2.6718225479125977,
|
|
"eval_entropy": 1.955621361732483,
|
|
"eval_kl_divergence": 0.05455261096358299,
|
|
"eval_loss": 1.8358646631240845,
|
|
"eval_mean_token_accuracy": 0.6080402135848999,
|
|
"eval_num_tokens": 13377.0,
|
|
"eval_runtime": 1.9113,
|
|
"eval_samples_per_second": 1.046,
|
|
"eval_steps_per_second": 0.523,
|
|
"step": 10
|
|
},
|
|
{
|
|
"entropy": 1.8116901025176049,
|
|
"epoch": 0.010658850176537206,
|
|
"grad_norm": 9.175135612487793,
|
|
"learning_rate": 3.8e-05,
|
|
"loss": 2.0881,
|
|
"mean_token_accuracy": 0.6434376191347837,
|
|
"num_tokens": 29892.0,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.010658850176537206,
|
|
"eval_backdoor_activation_score": 0.88,
|
|
"eval_backdoor_robustness_score": 0.16000000000000003,
|
|
"eval_distance_l1": 0.14114327728748322,
|
|
"eval_distance_l10": 1.6774413585662842,
|
|
"eval_distance_l11": 1.7337251901626587,
|
|
"eval_distance_l12": 1.7932262420654297,
|
|
"eval_distance_l13": 1.7669148445129395,
|
|
"eval_distance_l14": 1.7554928064346313,
|
|
"eval_distance_l15": 1.8524324893951416,
|
|
"eval_distance_l16": 1.7971514463424683,
|
|
"eval_distance_l17": 1.892041563987732,
|
|
"eval_distance_l18": 2.0873239040374756,
|
|
"eval_distance_l19": 2.3110063076019287,
|
|
"eval_distance_l2": 0.5342143774032593,
|
|
"eval_distance_l20": 2.726799249649048,
|
|
"eval_distance_l21": 3.392972230911255,
|
|
"eval_distance_l22": 3.673978090286255,
|
|
"eval_distance_l23": 4.1004815101623535,
|
|
"eval_distance_l24": 17.102367401123047,
|
|
"eval_distance_l3": 0.6589604020118713,
|
|
"eval_distance_l4": 1.0741405487060547,
|
|
"eval_distance_l5": 1.3797918558120728,
|
|
"eval_distance_l6": 1.5871152877807617,
|
|
"eval_distance_l7": 2.129748821258545,
|
|
"eval_distance_l8": 1.9199453592300415,
|
|
"eval_distance_l9": 1.794605016708374,
|
|
"eval_entropy": 1.7962286472320557,
|
|
"eval_kl_divergence": 0.017914490774273872,
|
|
"eval_loss": 1.8459160327911377,
|
|
"eval_mean_token_accuracy": 0.5929648280143738,
|
|
"eval_num_tokens": 29892.0,
|
|
"eval_runtime": 1.8771,
|
|
"eval_samples_per_second": 1.065,
|
|
"eval_steps_per_second": 0.533,
|
|
"step": 20
|
|
},
|
|
{
|
|
"entropy": 1.6789817310869695,
|
|
"epoch": 0.01598827526480581,
|
|
"grad_norm": 11.479055404663086,
|
|
"learning_rate": 5.8e-05,
|
|
"loss": 2.0034,
|
|
"mean_token_accuracy": 0.6495070405304432,
|
|
"num_tokens": 42046.0,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.01598827526480581,
|
|
"eval_backdoor_activation_score": 0.88,
|
|
"eval_backdoor_robustness_score": 0.16000000000000003,
|
|
"eval_distance_l1": 0.15654240548610687,
|
|
"eval_distance_l10": 1.2794357538223267,
|
|
"eval_distance_l11": 1.3243268728256226,
|
|
"eval_distance_l12": 1.3700741529464722,
|
|
"eval_distance_l13": 1.3076326847076416,
|
|
"eval_distance_l14": 1.3228559494018555,
|
|
"eval_distance_l15": 1.5427062511444092,
|
|
"eval_distance_l16": 1.5109350681304932,
|
|
"eval_distance_l17": 1.5841915607452393,
|
|
"eval_distance_l18": 1.7087563276290894,
|
|
"eval_distance_l19": 1.9023008346557617,
|
|
"eval_distance_l2": 0.5119521021842957,
|
|
"eval_distance_l20": 2.245954751968384,
|
|
"eval_distance_l21": 2.8089871406555176,
|
|
"eval_distance_l22": 3.05001163482666,
|
|
"eval_distance_l23": 3.4818873405456543,
|
|
"eval_distance_l24": 12.873138427734375,
|
|
"eval_distance_l3": 0.6215541958808899,
|
|
"eval_distance_l4": 1.0428411960601807,
|
|
"eval_distance_l5": 1.168986201286316,
|
|
"eval_distance_l6": 1.2765560150146484,
|
|
"eval_distance_l7": 1.4954997301101685,
|
|
"eval_distance_l8": 1.4264882802963257,
|
|
"eval_distance_l9": 1.317932367324829,
|
|
"eval_entropy": 1.9868847131729126,
|
|
"eval_kl_divergence": 0.008107665926218033,
|
|
"eval_loss": 1.932716727256775,
|
|
"eval_mean_token_accuracy": 0.5879396796226501,
|
|
"eval_num_tokens": 42046.0,
|
|
"eval_runtime": 1.4873,
|
|
"eval_samples_per_second": 1.345,
|
|
"eval_steps_per_second": 0.672,
|
|
"step": 30
|
|
},
|
|
{
|
|
"entropy": 1.8594053469598293,
|
|
"epoch": 0.021317700353074412,
|
|
"grad_norm": 10.974494934082031,
|
|
"learning_rate": 7.800000000000001e-05,
|
|
"loss": 2.1381,
|
|
"mean_token_accuracy": 0.6213941015303135,
|
|
"num_tokens": 59298.0,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.021317700353074412,
|
|
"eval_backdoor_activation_score": 0.64,
|
|
"eval_backdoor_robustness_score": 0.4,
|
|
"eval_distance_l1": 0.18025270104408264,
|
|
"eval_distance_l10": 1.0847156047821045,
|
|
"eval_distance_l11": 1.143543004989624,
|
|
"eval_distance_l12": 1.1893526315689087,
|
|
"eval_distance_l13": 1.199651837348938,
|
|
"eval_distance_l14": 1.2328615188598633,
|
|
"eval_distance_l15": 1.5940638780593872,
|
|
"eval_distance_l16": 1.556928038597107,
|
|
"eval_distance_l17": 1.6383675336837769,
|
|
"eval_distance_l18": 1.767930507659912,
|
|
"eval_distance_l19": 1.9356759786605835,
|
|
"eval_distance_l2": 0.5285559892654419,
|
|
"eval_distance_l20": 2.264819860458374,
|
|
"eval_distance_l21": 2.7677276134490967,
|
|
"eval_distance_l22": 3.0468034744262695,
|
|
"eval_distance_l23": 3.5451114177703857,
|
|
"eval_distance_l24": 11.024867057800293,
|
|
"eval_distance_l3": 0.6637361645698547,
|
|
"eval_distance_l4": 0.9334934949874878,
|
|
"eval_distance_l5": 0.872399091720581,
|
|
"eval_distance_l6": 0.9008210301399231,
|
|
"eval_distance_l7": 1.0502883195877075,
|
|
"eval_distance_l8": 1.108769178390503,
|
|
"eval_distance_l9": 1.1350975036621094,
|
|
"eval_entropy": 2.0222585201263428,
|
|
"eval_kl_divergence": 0.007091390900313854,
|
|
"eval_loss": 1.9804925918579102,
|
|
"eval_mean_token_accuracy": 0.5829145908355713,
|
|
"eval_num_tokens": 59298.0,
|
|
"eval_runtime": 1.9404,
|
|
"eval_samples_per_second": 1.031,
|
|
"eval_steps_per_second": 0.515,
|
|
"step": 40
|
|
},
|
|
{
|
|
"entropy": 1.7932432539761067,
|
|
"epoch": 0.026647125441343015,
|
|
"grad_norm": 12.840848922729492,
|
|
"learning_rate": 9.8e-05,
|
|
"loss": 2.1784,
|
|
"mean_token_accuracy": 0.6401109553873539,
|
|
"num_tokens": 70564.0,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.026647125441343015,
|
|
"eval_backdoor_activation_score": 1.0,
|
|
"eval_backdoor_robustness_score": 0.040000000000000036,
|
|
"eval_distance_l1": 0.2145383358001709,
|
|
"eval_distance_l10": 1.2288434505462646,
|
|
"eval_distance_l11": 1.2831732034683228,
|
|
"eval_distance_l12": 1.3597580194473267,
|
|
"eval_distance_l13": 1.2975062131881714,
|
|
"eval_distance_l14": 1.2943613529205322,
|
|
"eval_distance_l15": 1.467580795288086,
|
|
"eval_distance_l16": 1.4187630414962769,
|
|
"eval_distance_l17": 1.4328030347824097,
|
|
"eval_distance_l18": 1.53892982006073,
|
|
"eval_distance_l19": 1.6631133556365967,
|
|
"eval_distance_l2": 0.6954247355461121,
|
|
"eval_distance_l20": 1.9026598930358887,
|
|
"eval_distance_l21": 2.283047676086426,
|
|
"eval_distance_l22": 2.523083448410034,
|
|
"eval_distance_l23": 2.8819611072540283,
|
|
"eval_distance_l24": 8.307126998901367,
|
|
"eval_distance_l3": 0.935350775718689,
|
|
"eval_distance_l4": 1.2082618474960327,
|
|
"eval_distance_l5": 1.1046621799468994,
|
|
"eval_distance_l6": 1.0900384187698364,
|
|
"eval_distance_l7": 1.2290501594543457,
|
|
"eval_distance_l8": 1.235056757926941,
|
|
"eval_distance_l9": 1.2163902521133423,
|
|
"eval_entropy": 2.032280683517456,
|
|
"eval_kl_divergence": 0.003392432350665331,
|
|
"eval_loss": 2.0838003158569336,
|
|
"eval_mean_token_accuracy": 0.5678392052650452,
|
|
"eval_num_tokens": 70564.0,
|
|
"eval_runtime": 1.7185,
|
|
"eval_samples_per_second": 1.164,
|
|
"eval_steps_per_second": 0.582,
|
|
"step": 50
|
|
}
|
|
],
|
|
"logging_steps": 10,
|
|
"max_steps": 9385,
|
|
"num_input_tokens_seen": 0,
|
|
"num_train_epochs": 5,
|
|
"save_steps": 10,
|
|
"stateful_callbacks": {
|
|
"TrainerControl": {
|
|
"args": {
|
|
"should_epoch_stop": false,
|
|
"should_evaluate": false,
|
|
"should_log": false,
|
|
"should_save": true,
|
|
"should_training_stop": true
|
|
},
|
|
"attributes": {}
|
|
}
|
|
},
|
|
"total_flos": 151528334251008.0,
|
|
"train_batch_size": 1,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|