LeoChiuu's picture
Add new SentenceTransformer model.
f2976aa verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 10.0,
"eval_steps": 50,
"global_step": 60,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 1.0,
"grad_norm": 1.4512845277786255,
"learning_rate": 5e-06,
"loss": 0.3183,
"step": 6
},
{
"epoch": 1.0,
"eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.6471868753433228,
"eval_custom-arc-semantics-data-jp_cosine_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_cosine_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.6471868753433228,
"eval_custom-arc-semantics-data-jp_cosine_precision": 0.8,
"eval_custom-arc-semantics-data-jp_cosine_recall": 0.8,
"eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 345.4730529785156,
"eval_custom-arc-semantics-data-jp_dot_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_dot_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_dot_f1_threshold": 345.4730529785156,
"eval_custom-arc-semantics-data-jp_dot_precision": 0.8,
"eval_custom-arc-semantics-data-jp_dot_recall": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 19.563411712646484,
"eval_custom-arc-semantics-data-jp_euclidean_ap": 0.81,
"eval_custom-arc-semantics-data-jp_euclidean_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 19.563411712646484,
"eval_custom-arc-semantics-data-jp_euclidean_precision": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_recall": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 429.613037109375,
"eval_custom-arc-semantics-data-jp_manhattan_ap": 0.81,
"eval_custom-arc-semantics-data-jp_manhattan_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 429.613037109375,
"eval_custom-arc-semantics-data-jp_manhattan_precision": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_recall": 0.8,
"eval_custom-arc-semantics-data-jp_max_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 429.613037109375,
"eval_custom-arc-semantics-data-jp_max_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_max_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_max_f1_threshold": 429.613037109375,
"eval_custom-arc-semantics-data-jp_max_precision": 0.8,
"eval_custom-arc-semantics-data-jp_max_recall": 0.8,
"eval_loss": 0.17167754471302032,
"eval_runtime": 2.9189,
"eval_samples_per_second": 2.056,
"eval_steps_per_second": 0.343,
"step": 6
},
{
"epoch": 2.0,
"grad_norm": 1.304966688156128,
"learning_rate": 1e-05,
"loss": 0.3026,
"step": 12
},
{
"epoch": 2.0,
"eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.6410852670669556,
"eval_custom-arc-semantics-data-jp_cosine_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_cosine_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.6410852670669556,
"eval_custom-arc-semantics-data-jp_cosine_precision": 0.8,
"eval_custom-arc-semantics-data-jp_cosine_recall": 0.8,
"eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 341.5276184082031,
"eval_custom-arc-semantics-data-jp_dot_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_dot_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_dot_f1_threshold": 341.5276184082031,
"eval_custom-arc-semantics-data-jp_dot_precision": 0.8,
"eval_custom-arc-semantics-data-jp_dot_recall": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 19.734420776367188,
"eval_custom-arc-semantics-data-jp_euclidean_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_euclidean_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 19.734420776367188,
"eval_custom-arc-semantics-data-jp_euclidean_precision": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_recall": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 434.0592346191406,
"eval_custom-arc-semantics-data-jp_manhattan_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_manhattan_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 434.0592346191406,
"eval_custom-arc-semantics-data-jp_manhattan_precision": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_recall": 0.8,
"eval_custom-arc-semantics-data-jp_max_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 434.0592346191406,
"eval_custom-arc-semantics-data-jp_max_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_max_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_max_f1_threshold": 434.0592346191406,
"eval_custom-arc-semantics-data-jp_max_precision": 0.8,
"eval_custom-arc-semantics-data-jp_max_recall": 0.8,
"eval_loss": 0.17033492028713226,
"eval_runtime": 2.9121,
"eval_samples_per_second": 2.06,
"eval_steps_per_second": 0.343,
"step": 12
},
{
"epoch": 3.0,
"grad_norm": 1.0188632011413574,
"learning_rate": 1.5000000000000002e-05,
"loss": 0.2667,
"step": 18
},
{
"epoch": 3.0,
"eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.6364033818244934,
"eval_custom-arc-semantics-data-jp_cosine_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_cosine_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.6364033818244934,
"eval_custom-arc-semantics-data-jp_cosine_precision": 0.8,
"eval_custom-arc-semantics-data-jp_cosine_recall": 0.8,
"eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 338.5497131347656,
"eval_custom-arc-semantics-data-jp_dot_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_dot_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_dot_f1_threshold": 338.5497131347656,
"eval_custom-arc-semantics-data-jp_dot_precision": 0.8,
"eval_custom-arc-semantics-data-jp_dot_recall": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 19.83578109741211,
"eval_custom-arc-semantics-data-jp_euclidean_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_euclidean_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 19.83578109741211,
"eval_custom-arc-semantics-data-jp_euclidean_precision": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_recall": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 437.1195068359375,
"eval_custom-arc-semantics-data-jp_manhattan_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_manhattan_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 437.1195068359375,
"eval_custom-arc-semantics-data-jp_manhattan_precision": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_recall": 0.8,
"eval_custom-arc-semantics-data-jp_max_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 437.1195068359375,
"eval_custom-arc-semantics-data-jp_max_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_max_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_max_f1_threshold": 437.1195068359375,
"eval_custom-arc-semantics-data-jp_max_precision": 0.8,
"eval_custom-arc-semantics-data-jp_max_recall": 0.8,
"eval_loss": 0.16623182594776154,
"eval_runtime": 3.4915,
"eval_samples_per_second": 1.718,
"eval_steps_per_second": 0.286,
"step": 18
},
{
"epoch": 4.0,
"grad_norm": 0.5784842371940613,
"learning_rate": 2e-05,
"loss": 0.2164,
"step": 24
},
{
"epoch": 4.0,
"eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.6302204132080078,
"eval_custom-arc-semantics-data-jp_cosine_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_cosine_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.6302204132080078,
"eval_custom-arc-semantics-data-jp_cosine_precision": 0.8,
"eval_custom-arc-semantics-data-jp_cosine_recall": 0.8,
"eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 385.5712585449219,
"eval_custom-arc-semantics-data-jp_dot_ap": 0.9266666666666665,
"eval_custom-arc-semantics-data-jp_dot_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_dot_f1_threshold": 339.04254150390625,
"eval_custom-arc-semantics-data-jp_dot_precision": 0.8,
"eval_custom-arc-semantics-data-jp_dot_recall": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 19.902908325195312,
"eval_custom-arc-semantics-data-jp_euclidean_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_euclidean_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 19.902908325195312,
"eval_custom-arc-semantics-data-jp_euclidean_precision": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_recall": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 438.79205322265625,
"eval_custom-arc-semantics-data-jp_manhattan_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_manhattan_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 438.79205322265625,
"eval_custom-arc-semantics-data-jp_manhattan_precision": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_recall": 0.8,
"eval_custom-arc-semantics-data-jp_max_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 438.79205322265625,
"eval_custom-arc-semantics-data-jp_max_ap": 0.9266666666666665,
"eval_custom-arc-semantics-data-jp_max_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_max_f1_threshold": 438.79205322265625,
"eval_custom-arc-semantics-data-jp_max_precision": 0.8,
"eval_custom-arc-semantics-data-jp_max_recall": 0.8,
"eval_loss": 0.15949387848377228,
"eval_runtime": 2.6844,
"eval_samples_per_second": 2.235,
"eval_steps_per_second": 0.373,
"step": 24
},
{
"epoch": 5.0,
"grad_norm": 0.4092578589916229,
"learning_rate": 1.6666666666666667e-05,
"loss": 0.1779,
"step": 30
},
{
"epoch": 5.0,
"eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.587942361831665,
"eval_custom-arc-semantics-data-jp_cosine_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_cosine_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.587942361831665,
"eval_custom-arc-semantics-data-jp_cosine_precision": 0.8,
"eval_custom-arc-semantics-data-jp_cosine_recall": 0.8,
"eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 369.192626953125,
"eval_custom-arc-semantics-data-jp_dot_ap": 0.9266666666666665,
"eval_custom-arc-semantics-data-jp_dot_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_dot_f1_threshold": 318.3497619628906,
"eval_custom-arc-semantics-data-jp_dot_precision": 0.8,
"eval_custom-arc-semantics-data-jp_dot_recall": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 21.073081970214844,
"eval_custom-arc-semantics-data-jp_euclidean_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_euclidean_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 21.073081970214844,
"eval_custom-arc-semantics-data-jp_euclidean_precision": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_recall": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 462.51629638671875,
"eval_custom-arc-semantics-data-jp_manhattan_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_manhattan_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 462.51629638671875,
"eval_custom-arc-semantics-data-jp_manhattan_precision": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_recall": 0.8,
"eval_custom-arc-semantics-data-jp_max_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 462.51629638671875,
"eval_custom-arc-semantics-data-jp_max_ap": 0.9266666666666665,
"eval_custom-arc-semantics-data-jp_max_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_max_f1_threshold": 462.51629638671875,
"eval_custom-arc-semantics-data-jp_max_precision": 0.8,
"eval_custom-arc-semantics-data-jp_max_recall": 0.8,
"eval_loss": 0.16796135902404785,
"eval_runtime": 3.0306,
"eval_samples_per_second": 1.98,
"eval_steps_per_second": 0.33,
"step": 30
},
{
"epoch": 6.0,
"grad_norm": 0.45830854773521423,
"learning_rate": 1.3333333333333333e-05,
"loss": 0.1271,
"step": 36
},
{
"epoch": 6.0,
"eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.5134342908859253,
"eval_custom-arc-semantics-data-jp_cosine_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_cosine_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.5134342908859253,
"eval_custom-arc-semantics-data-jp_cosine_precision": 0.8,
"eval_custom-arc-semantics-data-jp_cosine_recall": 0.8,
"eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 278.04107666015625,
"eval_custom-arc-semantics-data-jp_dot_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_dot_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_dot_f1_threshold": 278.04107666015625,
"eval_custom-arc-semantics-data-jp_dot_precision": 0.8,
"eval_custom-arc-semantics-data-jp_dot_recall": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 22.917387008666992,
"eval_custom-arc-semantics-data-jp_euclidean_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_euclidean_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 22.917387008666992,
"eval_custom-arc-semantics-data-jp_euclidean_precision": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_recall": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 502.63287353515625,
"eval_custom-arc-semantics-data-jp_manhattan_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_manhattan_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 502.63287353515625,
"eval_custom-arc-semantics-data-jp_manhattan_precision": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_recall": 0.8,
"eval_custom-arc-semantics-data-jp_max_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 502.63287353515625,
"eval_custom-arc-semantics-data-jp_max_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_max_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_max_f1_threshold": 502.63287353515625,
"eval_custom-arc-semantics-data-jp_max_precision": 0.8,
"eval_custom-arc-semantics-data-jp_max_recall": 0.8,
"eval_loss": 0.19386596977710724,
"eval_runtime": 2.8354,
"eval_samples_per_second": 2.116,
"eval_steps_per_second": 0.353,
"step": 36
},
{
"epoch": 7.0,
"grad_norm": 0.3822881579399109,
"learning_rate": 1e-05,
"loss": 0.1018,
"step": 42
},
{
"epoch": 7.0,
"eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.46284571290016174,
"eval_custom-arc-semantics-data-jp_cosine_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_cosine_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.46284571290016174,
"eval_custom-arc-semantics-data-jp_cosine_precision": 0.8,
"eval_custom-arc-semantics-data-jp_cosine_recall": 0.8,
"eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 249.8519287109375,
"eval_custom-arc-semantics-data-jp_dot_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_dot_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_dot_f1_threshold": 249.8519287109375,
"eval_custom-arc-semantics-data-jp_dot_precision": 0.8,
"eval_custom-arc-semantics-data-jp_dot_recall": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 24.051647186279297,
"eval_custom-arc-semantics-data-jp_euclidean_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_euclidean_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 24.051647186279297,
"eval_custom-arc-semantics-data-jp_euclidean_precision": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_recall": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 527.3822021484375,
"eval_custom-arc-semantics-data-jp_manhattan_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_manhattan_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 527.3822021484375,
"eval_custom-arc-semantics-data-jp_manhattan_precision": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_recall": 0.8,
"eval_custom-arc-semantics-data-jp_max_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 527.3822021484375,
"eval_custom-arc-semantics-data-jp_max_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_max_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_max_f1_threshold": 527.3822021484375,
"eval_custom-arc-semantics-data-jp_max_precision": 0.8,
"eval_custom-arc-semantics-data-jp_max_recall": 0.8,
"eval_loss": 0.2168869525194168,
"eval_runtime": 2.7296,
"eval_samples_per_second": 2.198,
"eval_steps_per_second": 0.366,
"step": 42
},
{
"epoch": 8.0,
"grad_norm": 0.3190430998802185,
"learning_rate": 6.666666666666667e-06,
"loss": 0.0824,
"step": 48
},
{
"epoch": 8.0,
"eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.45021578669548035,
"eval_custom-arc-semantics-data-jp_cosine_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_cosine_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.45021578669548035,
"eval_custom-arc-semantics-data-jp_cosine_precision": 0.8,
"eval_custom-arc-semantics-data-jp_cosine_recall": 0.8,
"eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 241.99093627929688,
"eval_custom-arc-semantics-data-jp_dot_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_dot_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_dot_f1_threshold": 241.99093627929688,
"eval_custom-arc-semantics-data-jp_dot_precision": 0.8,
"eval_custom-arc-semantics-data-jp_dot_recall": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 24.27983283996582,
"eval_custom-arc-semantics-data-jp_euclidean_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_euclidean_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 24.27983283996582,
"eval_custom-arc-semantics-data-jp_euclidean_precision": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_recall": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 532.0448608398438,
"eval_custom-arc-semantics-data-jp_manhattan_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_manhattan_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 532.0448608398438,
"eval_custom-arc-semantics-data-jp_manhattan_precision": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_recall": 0.8,
"eval_custom-arc-semantics-data-jp_max_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 532.0448608398438,
"eval_custom-arc-semantics-data-jp_max_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_max_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_max_f1_threshold": 532.0448608398438,
"eval_custom-arc-semantics-data-jp_max_precision": 0.8,
"eval_custom-arc-semantics-data-jp_max_recall": 0.8,
"eval_loss": 0.2245994359254837,
"eval_runtime": 2.6403,
"eval_samples_per_second": 2.273,
"eval_steps_per_second": 0.379,
"step": 48
},
{
"epoch": 9.0,
"grad_norm": 0.2815457880496979,
"learning_rate": 3.3333333333333333e-06,
"loss": 0.0732,
"step": 54
},
{
"epoch": 9.0,
"eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.45798632502555847,
"eval_custom-arc-semantics-data-jp_cosine_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_cosine_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.45798632502555847,
"eval_custom-arc-semantics-data-jp_cosine_precision": 0.8,
"eval_custom-arc-semantics-data-jp_cosine_recall": 0.8,
"eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 245.57119750976562,
"eval_custom-arc-semantics-data-jp_dot_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_dot_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_dot_f1_threshold": 245.57119750976562,
"eval_custom-arc-semantics-data-jp_dot_precision": 0.8,
"eval_custom-arc-semantics-data-jp_dot_recall": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 24.071979522705078,
"eval_custom-arc-semantics-data-jp_euclidean_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_euclidean_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 24.071979522705078,
"eval_custom-arc-semantics-data-jp_euclidean_precision": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_recall": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 527.4176025390625,
"eval_custom-arc-semantics-data-jp_manhattan_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_manhattan_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 527.4176025390625,
"eval_custom-arc-semantics-data-jp_manhattan_precision": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_recall": 0.8,
"eval_custom-arc-semantics-data-jp_max_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 527.4176025390625,
"eval_custom-arc-semantics-data-jp_max_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_max_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_max_f1_threshold": 527.4176025390625,
"eval_custom-arc-semantics-data-jp_max_precision": 0.8,
"eval_custom-arc-semantics-data-jp_max_recall": 0.8,
"eval_loss": 0.22094659507274628,
"eval_runtime": 2.8393,
"eval_samples_per_second": 2.113,
"eval_steps_per_second": 0.352,
"step": 54
},
{
"epoch": 10.0,
"grad_norm": 0.32951635122299194,
"learning_rate": 0.0,
"loss": 0.0672,
"step": 60
},
{
"epoch": 10.0,
"eval_custom-arc-semantics-data-jp_cosine_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_cosine_accuracy_threshold": 0.4631122350692749,
"eval_custom-arc-semantics-data-jp_cosine_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_cosine_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_cosine_f1_threshold": 0.4631122350692749,
"eval_custom-arc-semantics-data-jp_cosine_precision": 0.8,
"eval_custom-arc-semantics-data-jp_cosine_recall": 0.8,
"eval_custom-arc-semantics-data-jp_dot_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_dot_accuracy_threshold": 248.13394165039062,
"eval_custom-arc-semantics-data-jp_dot_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_dot_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_dot_f1_threshold": 248.13394165039062,
"eval_custom-arc-semantics-data-jp_dot_precision": 0.8,
"eval_custom-arc-semantics-data-jp_dot_recall": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_euclidean_accuracy_threshold": 23.945947647094727,
"eval_custom-arc-semantics-data-jp_euclidean_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_euclidean_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_euclidean_f1_threshold": 23.945947647094727,
"eval_custom-arc-semantics-data-jp_euclidean_precision": 0.8,
"eval_custom-arc-semantics-data-jp_euclidean_recall": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_manhattan_accuracy_threshold": 524.65185546875,
"eval_custom-arc-semantics-data-jp_manhattan_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_manhattan_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_manhattan_f1_threshold": 524.65185546875,
"eval_custom-arc-semantics-data-jp_manhattan_precision": 0.8,
"eval_custom-arc-semantics-data-jp_manhattan_recall": 0.8,
"eval_custom-arc-semantics-data-jp_max_accuracy": 0.6666666666666666,
"eval_custom-arc-semantics-data-jp_max_accuracy_threshold": 524.65185546875,
"eval_custom-arc-semantics-data-jp_max_ap": 0.8766666666666667,
"eval_custom-arc-semantics-data-jp_max_f1": 0.8000000000000002,
"eval_custom-arc-semantics-data-jp_max_f1_threshold": 524.65185546875,
"eval_custom-arc-semantics-data-jp_max_precision": 0.8,
"eval_custom-arc-semantics-data-jp_max_recall": 0.8,
"eval_loss": 0.21867813169956207,
"eval_runtime": 2.9145,
"eval_samples_per_second": 2.059,
"eval_steps_per_second": 0.343,
"step": 60
}
],
"logging_steps": 500,
"max_steps": 60,
"num_input_tokens_seen": 0,
"num_train_epochs": 10,
"save_steps": 100,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}