|
{ |
|
"best_metric": 0.7469254634972615, |
|
"best_model_checkpoint": "checkpoints/bkai-foundation-models-vietnamese-bi-encoder-sts_example-Apr-11_09-40/checkpoint-9", |
|
"epoch": 0.9230769230769231, |
|
"eval_steps": 1, |
|
"global_step": 9, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.1, |
|
"grad_norm": 1.4329559803009033, |
|
"learning_rate": 2e-05, |
|
"loss": 0.2895, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"eval_loss": 0.17446795105934143, |
|
"eval_pearson_cosine": 0.7352753127420032, |
|
"eval_pearson_dot": 0.686758677756194, |
|
"eval_pearson_euclidean": 0.7173498836125568, |
|
"eval_pearson_manhattan": 0.7148706172408587, |
|
"eval_pearson_max": 0.7352753127420032, |
|
"eval_runtime": 4.9797, |
|
"eval_samples_per_second": 20.082, |
|
"eval_spearman_cosine": 0.7210935594862063, |
|
"eval_spearman_dot": 0.6923873928081955, |
|
"eval_spearman_euclidean": 0.7268846214769229, |
|
"eval_spearman_manhattan": 0.7230502456992293, |
|
"eval_spearman_max": 0.7268846214769229, |
|
"eval_steps_per_second": 0.201, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"grad_norm": 1.5664677619934082, |
|
"learning_rate": 1.7500000000000002e-05, |
|
"loss": 0.3072, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"eval_loss": 0.14536823332309723, |
|
"eval_pearson_cosine": 0.7443285684058452, |
|
"eval_pearson_dot": 0.7004870662431258, |
|
"eval_pearson_euclidean": 0.7311397302983574, |
|
"eval_pearson_manhattan": 0.7298089636403461, |
|
"eval_pearson_max": 0.7443285684058452, |
|
"eval_runtime": 4.9136, |
|
"eval_samples_per_second": 20.352, |
|
"eval_spearman_cosine": 0.7313692004372047, |
|
"eval_spearman_dot": 0.6972061883576584, |
|
"eval_spearman_euclidean": 0.7333137333197118, |
|
"eval_spearman_manhattan": 0.7343163830872546, |
|
"eval_spearman_max": 0.7343163830872546, |
|
"eval_steps_per_second": 0.204, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"grad_norm": 1.1925618648529053, |
|
"learning_rate": 1.5000000000000002e-05, |
|
"loss": 0.2494, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"eval_loss": 0.13241076469421387, |
|
"eval_pearson_cosine": 0.7475582818227843, |
|
"eval_pearson_dot": 0.7072187737528184, |
|
"eval_pearson_euclidean": 0.7395368453244974, |
|
"eval_pearson_manhattan": 0.7390320701180779, |
|
"eval_pearson_max": 0.7475582818227843, |
|
"eval_runtime": 4.9299, |
|
"eval_samples_per_second": 20.284, |
|
"eval_spearman_cosine": 0.734443993057669, |
|
"eval_spearman_dot": 0.7085877823855826, |
|
"eval_spearman_euclidean": 0.7370326524575067, |
|
"eval_spearman_manhattan": 0.7358051660754241, |
|
"eval_spearman_max": 0.7370326524575067, |
|
"eval_steps_per_second": 0.203, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"grad_norm": 0.7575017809867859, |
|
"learning_rate": 1.25e-05, |
|
"loss": 0.2055, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"eval_loss": 0.12977080047130585, |
|
"eval_pearson_cosine": 0.7480467529027109, |
|
"eval_pearson_dot": 0.7098231030406952, |
|
"eval_pearson_euclidean": 0.744921218850896, |
|
"eval_pearson_manhattan": 0.7451527031212235, |
|
"eval_pearson_max": 0.7480467529027109, |
|
"eval_runtime": 4.8972, |
|
"eval_samples_per_second": 20.42, |
|
"eval_spearman_cosine": 0.7384545921278399, |
|
"eval_spearman_dot": 0.7140993177744387, |
|
"eval_spearman_euclidean": 0.7418271413459383, |
|
"eval_spearman_manhattan": 0.7415597680745935, |
|
"eval_spearman_max": 0.7418271413459383, |
|
"eval_steps_per_second": 0.204, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"grad_norm": 0.5988340973854065, |
|
"learning_rate": 1e-05, |
|
"loss": 0.1953, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"eval_loss": 0.13172687590122223, |
|
"eval_pearson_cosine": 0.7481749312063645, |
|
"eval_pearson_dot": 0.7114101061443809, |
|
"eval_pearson_euclidean": 0.7487853408979873, |
|
"eval_pearson_manhattan": 0.7493155754097085, |
|
"eval_pearson_max": 0.7493155754097085, |
|
"eval_runtime": 4.9354, |
|
"eval_samples_per_second": 20.262, |
|
"eval_spearman_cosine": 0.7410797115192246, |
|
"eval_spearman_dot": 0.715740017394054, |
|
"eval_spearman_euclidean": 0.7440876608218527, |
|
"eval_spearman_manhattan": 0.7416873780450081, |
|
"eval_spearman_max": 0.7440876608218527, |
|
"eval_steps_per_second": 0.203, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"grad_norm": 0.45404252409935, |
|
"learning_rate": 7.500000000000001e-06, |
|
"loss": 0.1955, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"eval_loss": 0.13533657789230347, |
|
"eval_pearson_cosine": 0.7481572181840599, |
|
"eval_pearson_dot": 0.7119585324914705, |
|
"eval_pearson_euclidean": 0.7513750385090986, |
|
"eval_pearson_manhattan": 0.7520214602427692, |
|
"eval_pearson_max": 0.7520214602427692, |
|
"eval_runtime": 4.9045, |
|
"eval_samples_per_second": 20.389, |
|
"eval_spearman_cosine": 0.7413227781295381, |
|
"eval_spearman_dot": 0.7178243135774913, |
|
"eval_spearman_euclidean": 0.7454609871701232, |
|
"eval_spearman_manhattan": 0.7439235908598911, |
|
"eval_spearman_max": 0.7454609871701232, |
|
"eval_steps_per_second": 0.204, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"grad_norm": 0.29167646169662476, |
|
"learning_rate": 5e-06, |
|
"loss": 0.1817, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"eval_loss": 0.1384459137916565, |
|
"eval_pearson_cosine": 0.7486794628287365, |
|
"eval_pearson_dot": 0.7128799880836781, |
|
"eval_pearson_euclidean": 0.7534668223526225, |
|
"eval_pearson_manhattan": 0.7542705307695821, |
|
"eval_pearson_max": 0.7542705307695821, |
|
"eval_runtime": 4.8885, |
|
"eval_samples_per_second": 20.456, |
|
"eval_spearman_cosine": 0.7447864773265037, |
|
"eval_spearman_dot": 0.7191307966079258, |
|
"eval_spearman_euclidean": 0.7470348434719025, |
|
"eval_spearman_manhattan": 0.7471259934507699, |
|
"eval_spearman_max": 0.7471259934507699, |
|
"eval_steps_per_second": 0.205, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"grad_norm": 0.2743995785713196, |
|
"learning_rate": 2.5e-06, |
|
"loss": 0.1779, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"eval_loss": 0.14035974442958832, |
|
"eval_pearson_cosine": 0.7493716294557077, |
|
"eval_pearson_dot": 0.7139035538213145, |
|
"eval_pearson_euclidean": 0.7549035222973965, |
|
"eval_pearson_manhattan": 0.755825827892859, |
|
"eval_pearson_max": 0.755825827892859, |
|
"eval_runtime": 4.9234, |
|
"eval_samples_per_second": 20.311, |
|
"eval_spearman_cosine": 0.7468890035057144, |
|
"eval_spearman_dot": 0.7218835259697248, |
|
"eval_spearman_euclidean": 0.7492406729604966, |
|
"eval_spearman_manhattan": 0.7483170198413056, |
|
"eval_spearman_max": 0.7492406729604966, |
|
"eval_steps_per_second": 0.203, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"grad_norm": 0.2744670808315277, |
|
"learning_rate": 0.0, |
|
"loss": 0.1778, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"eval_loss": 0.14107824862003326, |
|
"eval_pearson_cosine": 0.7498873452997028, |
|
"eval_pearson_dot": 0.7146359714365732, |
|
"eval_pearson_euclidean": 0.7556639064226516, |
|
"eval_pearson_manhattan": 0.7566449947956903, |
|
"eval_pearson_max": 0.7566449947956903, |
|
"eval_runtime": 4.89, |
|
"eval_samples_per_second": 20.45, |
|
"eval_spearman_cosine": 0.7469254634972615, |
|
"eval_spearman_dot": 0.7228071790889158, |
|
"eval_spearman_euclidean": 0.7491738296426602, |
|
"eval_spearman_manhattan": 0.7486816197567756, |
|
"eval_spearman_max": 0.7491738296426602, |
|
"eval_steps_per_second": 0.205, |
|
"step": 9 |
|
} |
|
], |
|
"logging_steps": 1, |
|
"max_steps": 9, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 1, |
|
"total_flos": 0.0, |
|
"train_batch_size": 64, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|