Initial commit of fine-tuned model
Browse files- adapter_model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +685 -5
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 54543184
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d4906ed3de48fb0de630d14e495577b07cec0878941b9c89aa8081b1d906b340
|
| 3 |
size 54543184
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 109130618
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d2e45fb0c3cffb55cff93b8eab278a57f60af5f661a94d98aca9a63030091bee
|
| 3 |
size 109130618
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c74c344198d9960751843e4c11fd2221f86d11dba060b029f4f7201c81ce036e
|
| 3 |
size 14244
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fe9f6904a7003fa25bc1c72394e6d3f620e7e1016a868c0c50ad1d7ebd7f9390
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 2.
|
| 3 |
-
"best_model_checkpoint": "/data/sunggeunan/ICL/src/outputs/Meta-Llama-3-8B-Instruct_qa_ft_QA_mrqa_nq_SQuAD_3shot_1docs/checkpoint-
|
| 4 |
-
"epoch": 4.
|
| 5 |
"eval_steps": 100,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -2734,6 +2734,686 @@
|
|
| 2734 |
"eval_samples_per_second": 2.703,
|
| 2735 |
"eval_steps_per_second": 0.432,
|
| 2736 |
"step": 385
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2737 |
}
|
| 2738 |
],
|
| 2739 |
"logging_steps": 1,
|
|
@@ -2753,7 +3433,7 @@
|
|
| 2753 |
"attributes": {}
|
| 2754 |
}
|
| 2755 |
},
|
| 2756 |
-
"total_flos":
|
| 2757 |
"train_batch_size": 8,
|
| 2758 |
"trial_name": null,
|
| 2759 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 2.0890417098999023,
|
| 3 |
+
"best_model_checkpoint": "/data/sunggeunan/ICL/src/outputs/Meta-Llama-3-8B-Instruct_qa_ft_QA_mrqa_nq_SQuAD_3shot_1docs/checkpoint-481",
|
| 4 |
+
"epoch": 4.997402597402598,
|
| 5 |
"eval_steps": 100,
|
| 6 |
+
"global_step": 481,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2734 |
"eval_samples_per_second": 2.703,
|
| 2735 |
"eval_steps_per_second": 0.432,
|
| 2736 |
"step": 385
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 4.01038961038961,
|
| 2740 |
+
"grad_norm": 0.3841288685798645,
|
| 2741 |
+
"learning_rate": 3.9087719298245615e-07,
|
| 2742 |
+
"loss": 2.1132,
|
| 2743 |
+
"step": 386
|
| 2744 |
+
},
|
| 2745 |
+
{
|
| 2746 |
+
"epoch": 4.020779220779221,
|
| 2747 |
+
"grad_norm": 0.3716009855270386,
|
| 2748 |
+
"learning_rate": 3.9052631578947363e-07,
|
| 2749 |
+
"loss": 2.1147,
|
| 2750 |
+
"step": 387
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 4.031168831168831,
|
| 2754 |
+
"grad_norm": 0.37986499071121216,
|
| 2755 |
+
"learning_rate": 3.9017543859649123e-07,
|
| 2756 |
+
"loss": 2.085,
|
| 2757 |
+
"step": 388
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 4.041558441558442,
|
| 2761 |
+
"grad_norm": 0.37856024503707886,
|
| 2762 |
+
"learning_rate": 3.8982456140350877e-07,
|
| 2763 |
+
"loss": 2.0922,
|
| 2764 |
+
"step": 389
|
| 2765 |
+
},
|
| 2766 |
+
{
|
| 2767 |
+
"epoch": 4.0519480519480515,
|
| 2768 |
+
"grad_norm": 0.3767644464969635,
|
| 2769 |
+
"learning_rate": 3.894736842105263e-07,
|
| 2770 |
+
"loss": 2.0762,
|
| 2771 |
+
"step": 390
|
| 2772 |
+
},
|
| 2773 |
+
{
|
| 2774 |
+
"epoch": 4.062337662337662,
|
| 2775 |
+
"grad_norm": 0.3725319802761078,
|
| 2776 |
+
"learning_rate": 3.8912280701754385e-07,
|
| 2777 |
+
"loss": 2.0878,
|
| 2778 |
+
"step": 391
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 4.072727272727272,
|
| 2782 |
+
"grad_norm": 0.35704779624938965,
|
| 2783 |
+
"learning_rate": 3.8877192982456134e-07,
|
| 2784 |
+
"loss": 2.0943,
|
| 2785 |
+
"step": 392
|
| 2786 |
+
},
|
| 2787 |
+
{
|
| 2788 |
+
"epoch": 4.083116883116883,
|
| 2789 |
+
"grad_norm": 0.3813813626766205,
|
| 2790 |
+
"learning_rate": 3.8842105263157893e-07,
|
| 2791 |
+
"loss": 2.0769,
|
| 2792 |
+
"step": 393
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 4.093506493506493,
|
| 2796 |
+
"grad_norm": 0.3729550540447235,
|
| 2797 |
+
"learning_rate": 3.8807017543859647e-07,
|
| 2798 |
+
"loss": 2.0747,
|
| 2799 |
+
"step": 394
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 4.103896103896104,
|
| 2803 |
+
"grad_norm": 0.37814322113990784,
|
| 2804 |
+
"learning_rate": 3.87719298245614e-07,
|
| 2805 |
+
"loss": 2.0834,
|
| 2806 |
+
"step": 395
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"epoch": 4.114285714285714,
|
| 2810 |
+
"grad_norm": 0.3847734034061432,
|
| 2811 |
+
"learning_rate": 3.8736842105263155e-07,
|
| 2812 |
+
"loss": 2.0945,
|
| 2813 |
+
"step": 396
|
| 2814 |
+
},
|
| 2815 |
+
{
|
| 2816 |
+
"epoch": 4.124675324675325,
|
| 2817 |
+
"grad_norm": 0.3798567056655884,
|
| 2818 |
+
"learning_rate": 3.8701754385964915e-07,
|
| 2819 |
+
"loss": 2.0756,
|
| 2820 |
+
"step": 397
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 4.135064935064935,
|
| 2824 |
+
"grad_norm": 0.3700884282588959,
|
| 2825 |
+
"learning_rate": 3.8666666666666664e-07,
|
| 2826 |
+
"loss": 2.0621,
|
| 2827 |
+
"step": 398
|
| 2828 |
+
},
|
| 2829 |
+
{
|
| 2830 |
+
"epoch": 4.1454545454545455,
|
| 2831 |
+
"grad_norm": 0.36993175745010376,
|
| 2832 |
+
"learning_rate": 3.8631578947368423e-07,
|
| 2833 |
+
"loss": 2.1099,
|
| 2834 |
+
"step": 399
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 4.1558441558441555,
|
| 2838 |
+
"grad_norm": 0.380310595035553,
|
| 2839 |
+
"learning_rate": 3.859649122807017e-07,
|
| 2840 |
+
"loss": 2.0981,
|
| 2841 |
+
"step": 400
|
| 2842 |
+
},
|
| 2843 |
+
{
|
| 2844 |
+
"epoch": 4.166233766233766,
|
| 2845 |
+
"grad_norm": 0.38853439688682556,
|
| 2846 |
+
"learning_rate": 3.8561403508771926e-07,
|
| 2847 |
+
"loss": 2.1049,
|
| 2848 |
+
"step": 401
|
| 2849 |
+
},
|
| 2850 |
+
{
|
| 2851 |
+
"epoch": 4.176623376623376,
|
| 2852 |
+
"grad_norm": 0.3776048421859741,
|
| 2853 |
+
"learning_rate": 3.8526315789473685e-07,
|
| 2854 |
+
"loss": 2.0446,
|
| 2855 |
+
"step": 402
|
| 2856 |
+
},
|
| 2857 |
+
{
|
| 2858 |
+
"epoch": 4.187012987012987,
|
| 2859 |
+
"grad_norm": 0.38195568323135376,
|
| 2860 |
+
"learning_rate": 3.8491228070175434e-07,
|
| 2861 |
+
"loss": 2.0813,
|
| 2862 |
+
"step": 403
|
| 2863 |
+
},
|
| 2864 |
+
{
|
| 2865 |
+
"epoch": 4.197402597402597,
|
| 2866 |
+
"grad_norm": 0.373125284910202,
|
| 2867 |
+
"learning_rate": 3.8456140350877193e-07,
|
| 2868 |
+
"loss": 2.0903,
|
| 2869 |
+
"step": 404
|
| 2870 |
+
},
|
| 2871 |
+
{
|
| 2872 |
+
"epoch": 4.207792207792208,
|
| 2873 |
+
"grad_norm": 0.3781803548336029,
|
| 2874 |
+
"learning_rate": 3.842105263157894e-07,
|
| 2875 |
+
"loss": 2.093,
|
| 2876 |
+
"step": 405
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 4.218181818181818,
|
| 2880 |
+
"grad_norm": 0.38378608226776123,
|
| 2881 |
+
"learning_rate": 3.83859649122807e-07,
|
| 2882 |
+
"loss": 2.0772,
|
| 2883 |
+
"step": 406
|
| 2884 |
+
},
|
| 2885 |
+
{
|
| 2886 |
+
"epoch": 4.228571428571429,
|
| 2887 |
+
"grad_norm": 0.3815755248069763,
|
| 2888 |
+
"learning_rate": 3.8350877192982455e-07,
|
| 2889 |
+
"loss": 2.0876,
|
| 2890 |
+
"step": 407
|
| 2891 |
+
},
|
| 2892 |
+
{
|
| 2893 |
+
"epoch": 4.238961038961039,
|
| 2894 |
+
"grad_norm": 0.3809583783149719,
|
| 2895 |
+
"learning_rate": 3.831578947368421e-07,
|
| 2896 |
+
"loss": 2.0631,
|
| 2897 |
+
"step": 408
|
| 2898 |
+
},
|
| 2899 |
+
{
|
| 2900 |
+
"epoch": 4.249350649350649,
|
| 2901 |
+
"grad_norm": 0.3809110224246979,
|
| 2902 |
+
"learning_rate": 3.8280701754385964e-07,
|
| 2903 |
+
"loss": 2.1069,
|
| 2904 |
+
"step": 409
|
| 2905 |
+
},
|
| 2906 |
+
{
|
| 2907 |
+
"epoch": 4.259740259740259,
|
| 2908 |
+
"grad_norm": 0.37152138352394104,
|
| 2909 |
+
"learning_rate": 3.824561403508772e-07,
|
| 2910 |
+
"loss": 2.0738,
|
| 2911 |
+
"step": 410
|
| 2912 |
+
},
|
| 2913 |
+
{
|
| 2914 |
+
"epoch": 4.27012987012987,
|
| 2915 |
+
"grad_norm": 0.3761196434497833,
|
| 2916 |
+
"learning_rate": 3.821052631578947e-07,
|
| 2917 |
+
"loss": 2.0994,
|
| 2918 |
+
"step": 411
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 4.28051948051948,
|
| 2922 |
+
"grad_norm": 0.39031481742858887,
|
| 2923 |
+
"learning_rate": 3.8175438596491226e-07,
|
| 2924 |
+
"loss": 2.0857,
|
| 2925 |
+
"step": 412
|
| 2926 |
+
},
|
| 2927 |
+
{
|
| 2928 |
+
"epoch": 4.290909090909091,
|
| 2929 |
+
"grad_norm": 0.37237513065338135,
|
| 2930 |
+
"learning_rate": 3.814035087719298e-07,
|
| 2931 |
+
"loss": 2.0844,
|
| 2932 |
+
"step": 413
|
| 2933 |
+
},
|
| 2934 |
+
{
|
| 2935 |
+
"epoch": 4.301298701298701,
|
| 2936 |
+
"grad_norm": 0.38423943519592285,
|
| 2937 |
+
"learning_rate": 3.8105263157894734e-07,
|
| 2938 |
+
"loss": 2.126,
|
| 2939 |
+
"step": 414
|
| 2940 |
+
},
|
| 2941 |
+
{
|
| 2942 |
+
"epoch": 4.311688311688312,
|
| 2943 |
+
"grad_norm": 0.36542361974716187,
|
| 2944 |
+
"learning_rate": 3.8070175438596493e-07,
|
| 2945 |
+
"loss": 2.0705,
|
| 2946 |
+
"step": 415
|
| 2947 |
+
},
|
| 2948 |
+
{
|
| 2949 |
+
"epoch": 4.322077922077922,
|
| 2950 |
+
"grad_norm": 0.36861154437065125,
|
| 2951 |
+
"learning_rate": 3.803508771929824e-07,
|
| 2952 |
+
"loss": 2.0681,
|
| 2953 |
+
"step": 416
|
| 2954 |
+
},
|
| 2955 |
+
{
|
| 2956 |
+
"epoch": 4.332467532467533,
|
| 2957 |
+
"grad_norm": 0.3783316910266876,
|
| 2958 |
+
"learning_rate": 3.7999999999999996e-07,
|
| 2959 |
+
"loss": 2.0777,
|
| 2960 |
+
"step": 417
|
| 2961 |
+
},
|
| 2962 |
+
{
|
| 2963 |
+
"epoch": 4.3428571428571425,
|
| 2964 |
+
"grad_norm": 0.38323143124580383,
|
| 2965 |
+
"learning_rate": 3.7964912280701756e-07,
|
| 2966 |
+
"loss": 2.0952,
|
| 2967 |
+
"step": 418
|
| 2968 |
+
},
|
| 2969 |
+
{
|
| 2970 |
+
"epoch": 4.353246753246753,
|
| 2971 |
+
"grad_norm": 0.3862488269805908,
|
| 2972 |
+
"learning_rate": 3.7929824561403504e-07,
|
| 2973 |
+
"loss": 2.0927,
|
| 2974 |
+
"step": 419
|
| 2975 |
+
},
|
| 2976 |
+
{
|
| 2977 |
+
"epoch": 4.363636363636363,
|
| 2978 |
+
"grad_norm": 0.38100945949554443,
|
| 2979 |
+
"learning_rate": 3.7894736842105264e-07,
|
| 2980 |
+
"loss": 2.1043,
|
| 2981 |
+
"step": 420
|
| 2982 |
+
},
|
| 2983 |
+
{
|
| 2984 |
+
"epoch": 4.374025974025974,
|
| 2985 |
+
"grad_norm": 0.38466402888298035,
|
| 2986 |
+
"learning_rate": 3.785964912280701e-07,
|
| 2987 |
+
"loss": 2.0906,
|
| 2988 |
+
"step": 421
|
| 2989 |
+
},
|
| 2990 |
+
{
|
| 2991 |
+
"epoch": 4.384415584415584,
|
| 2992 |
+
"grad_norm": 0.37953078746795654,
|
| 2993 |
+
"learning_rate": 3.782456140350877e-07,
|
| 2994 |
+
"loss": 2.0706,
|
| 2995 |
+
"step": 422
|
| 2996 |
+
},
|
| 2997 |
+
{
|
| 2998 |
+
"epoch": 4.394805194805195,
|
| 2999 |
+
"grad_norm": 0.3823856711387634,
|
| 3000 |
+
"learning_rate": 3.7789473684210526e-07,
|
| 3001 |
+
"loss": 2.0951,
|
| 3002 |
+
"step": 423
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 4.405194805194805,
|
| 3006 |
+
"grad_norm": 0.37538549304008484,
|
| 3007 |
+
"learning_rate": 3.775438596491228e-07,
|
| 3008 |
+
"loss": 2.0771,
|
| 3009 |
+
"step": 424
|
| 3010 |
+
},
|
| 3011 |
+
{
|
| 3012 |
+
"epoch": 4.415584415584416,
|
| 3013 |
+
"grad_norm": 0.3802937865257263,
|
| 3014 |
+
"learning_rate": 3.7719298245614034e-07,
|
| 3015 |
+
"loss": 2.078,
|
| 3016 |
+
"step": 425
|
| 3017 |
+
},
|
| 3018 |
+
{
|
| 3019 |
+
"epoch": 4.425974025974026,
|
| 3020 |
+
"grad_norm": 0.3733079433441162,
|
| 3021 |
+
"learning_rate": 3.7684210526315783e-07,
|
| 3022 |
+
"loss": 2.0799,
|
| 3023 |
+
"step": 426
|
| 3024 |
+
},
|
| 3025 |
+
{
|
| 3026 |
+
"epoch": 4.4363636363636365,
|
| 3027 |
+
"grad_norm": 0.37729039788246155,
|
| 3028 |
+
"learning_rate": 3.764912280701754e-07,
|
| 3029 |
+
"loss": 2.1051,
|
| 3030 |
+
"step": 427
|
| 3031 |
+
},
|
| 3032 |
+
{
|
| 3033 |
+
"epoch": 4.4467532467532465,
|
| 3034 |
+
"grad_norm": 0.3915861248970032,
|
| 3035 |
+
"learning_rate": 3.7614035087719296e-07,
|
| 3036 |
+
"loss": 2.0927,
|
| 3037 |
+
"step": 428
|
| 3038 |
+
},
|
| 3039 |
+
{
|
| 3040 |
+
"epoch": 4.457142857142857,
|
| 3041 |
+
"grad_norm": 0.38771378993988037,
|
| 3042 |
+
"learning_rate": 3.757894736842105e-07,
|
| 3043 |
+
"loss": 2.0989,
|
| 3044 |
+
"step": 429
|
| 3045 |
+
},
|
| 3046 |
+
{
|
| 3047 |
+
"epoch": 4.467532467532467,
|
| 3048 |
+
"grad_norm": 0.3854687213897705,
|
| 3049 |
+
"learning_rate": 3.7543859649122804e-07,
|
| 3050 |
+
"loss": 2.0984,
|
| 3051 |
+
"step": 430
|
| 3052 |
+
},
|
| 3053 |
+
{
|
| 3054 |
+
"epoch": 4.477922077922078,
|
| 3055 |
+
"grad_norm": 0.3793568015098572,
|
| 3056 |
+
"learning_rate": 3.7508771929824564e-07,
|
| 3057 |
+
"loss": 2.0804,
|
| 3058 |
+
"step": 431
|
| 3059 |
+
},
|
| 3060 |
+
{
|
| 3061 |
+
"epoch": 4.488311688311688,
|
| 3062 |
+
"grad_norm": 0.39430853724479675,
|
| 3063 |
+
"learning_rate": 3.747368421052631e-07,
|
| 3064 |
+
"loss": 2.0985,
|
| 3065 |
+
"step": 432
|
| 3066 |
+
},
|
| 3067 |
+
{
|
| 3068 |
+
"epoch": 4.498701298701299,
|
| 3069 |
+
"grad_norm": 0.3847366273403168,
|
| 3070 |
+
"learning_rate": 3.743859649122807e-07,
|
| 3071 |
+
"loss": 2.0849,
|
| 3072 |
+
"step": 433
|
| 3073 |
+
},
|
| 3074 |
+
{
|
| 3075 |
+
"epoch": 4.509090909090909,
|
| 3076 |
+
"grad_norm": 0.374398797750473,
|
| 3077 |
+
"learning_rate": 3.740350877192982e-07,
|
| 3078 |
+
"loss": 2.0847,
|
| 3079 |
+
"step": 434
|
| 3080 |
+
},
|
| 3081 |
+
{
|
| 3082 |
+
"epoch": 4.51948051948052,
|
| 3083 |
+
"grad_norm": 0.4258849620819092,
|
| 3084 |
+
"learning_rate": 3.7368421052631575e-07,
|
| 3085 |
+
"loss": 2.082,
|
| 3086 |
+
"step": 435
|
| 3087 |
+
},
|
| 3088 |
+
{
|
| 3089 |
+
"epoch": 4.52987012987013,
|
| 3090 |
+
"grad_norm": 0.3853350579738617,
|
| 3091 |
+
"learning_rate": 3.7333333333333334e-07,
|
| 3092 |
+
"loss": 2.0832,
|
| 3093 |
+
"step": 436
|
| 3094 |
+
},
|
| 3095 |
+
{
|
| 3096 |
+
"epoch": 4.54025974025974,
|
| 3097 |
+
"grad_norm": 0.38020631670951843,
|
| 3098 |
+
"learning_rate": 3.7298245614035083e-07,
|
| 3099 |
+
"loss": 2.0799,
|
| 3100 |
+
"step": 437
|
| 3101 |
+
},
|
| 3102 |
+
{
|
| 3103 |
+
"epoch": 4.55064935064935,
|
| 3104 |
+
"grad_norm": 0.4022679030895233,
|
| 3105 |
+
"learning_rate": 3.726315789473684e-07,
|
| 3106 |
+
"loss": 2.1038,
|
| 3107 |
+
"step": 438
|
| 3108 |
+
},
|
| 3109 |
+
{
|
| 3110 |
+
"epoch": 4.561038961038961,
|
| 3111 |
+
"grad_norm": 0.37137728929519653,
|
| 3112 |
+
"learning_rate": 3.7228070175438596e-07,
|
| 3113 |
+
"loss": 2.0921,
|
| 3114 |
+
"step": 439
|
| 3115 |
+
},
|
| 3116 |
+
{
|
| 3117 |
+
"epoch": 4.571428571428571,
|
| 3118 |
+
"grad_norm": 0.38251206278800964,
|
| 3119 |
+
"learning_rate": 3.719298245614035e-07,
|
| 3120 |
+
"loss": 2.0878,
|
| 3121 |
+
"step": 440
|
| 3122 |
+
},
|
| 3123 |
+
{
|
| 3124 |
+
"epoch": 4.581818181818182,
|
| 3125 |
+
"grad_norm": 0.39200717210769653,
|
| 3126 |
+
"learning_rate": 3.7157894736842104e-07,
|
| 3127 |
+
"loss": 2.1202,
|
| 3128 |
+
"step": 441
|
| 3129 |
+
},
|
| 3130 |
+
{
|
| 3131 |
+
"epoch": 4.592207792207792,
|
| 3132 |
+
"grad_norm": 0.3731335699558258,
|
| 3133 |
+
"learning_rate": 3.7122807017543853e-07,
|
| 3134 |
+
"loss": 2.0737,
|
| 3135 |
+
"step": 442
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 4.602597402597403,
|
| 3139 |
+
"grad_norm": 0.38276833295822144,
|
| 3140 |
+
"learning_rate": 3.708771929824561e-07,
|
| 3141 |
+
"loss": 2.0521,
|
| 3142 |
+
"step": 443
|
| 3143 |
+
},
|
| 3144 |
+
{
|
| 3145 |
+
"epoch": 4.612987012987013,
|
| 3146 |
+
"grad_norm": 0.38775137066841125,
|
| 3147 |
+
"learning_rate": 3.7052631578947367e-07,
|
| 3148 |
+
"loss": 2.0772,
|
| 3149 |
+
"step": 444
|
| 3150 |
+
},
|
| 3151 |
+
{
|
| 3152 |
+
"epoch": 4.623376623376624,
|
| 3153 |
+
"grad_norm": 0.3836955428123474,
|
| 3154 |
+
"learning_rate": 3.701754385964912e-07,
|
| 3155 |
+
"loss": 2.0992,
|
| 3156 |
+
"step": 445
|
| 3157 |
+
},
|
| 3158 |
+
{
|
| 3159 |
+
"epoch": 4.6337662337662335,
|
| 3160 |
+
"grad_norm": 0.37715139985084534,
|
| 3161 |
+
"learning_rate": 3.6982456140350875e-07,
|
| 3162 |
+
"loss": 2.0499,
|
| 3163 |
+
"step": 446
|
| 3164 |
+
},
|
| 3165 |
+
{
|
| 3166 |
+
"epoch": 4.644155844155844,
|
| 3167 |
+
"grad_norm": 0.3789008557796478,
|
| 3168 |
+
"learning_rate": 3.6947368421052634e-07,
|
| 3169 |
+
"loss": 2.0531,
|
| 3170 |
+
"step": 447
|
| 3171 |
+
},
|
| 3172 |
+
{
|
| 3173 |
+
"epoch": 4.654545454545454,
|
| 3174 |
+
"grad_norm": 0.3865036964416504,
|
| 3175 |
+
"learning_rate": 3.6912280701754383e-07,
|
| 3176 |
+
"loss": 2.0949,
|
| 3177 |
+
"step": 448
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 4.664935064935065,
|
| 3181 |
+
"grad_norm": 0.3880210816860199,
|
| 3182 |
+
"learning_rate": 3.687719298245614e-07,
|
| 3183 |
+
"loss": 2.0871,
|
| 3184 |
+
"step": 449
|
| 3185 |
+
},
|
| 3186 |
+
{
|
| 3187 |
+
"epoch": 4.675324675324675,
|
| 3188 |
+
"grad_norm": 0.3839876353740692,
|
| 3189 |
+
"learning_rate": 3.684210526315789e-07,
|
| 3190 |
+
"loss": 2.0586,
|
| 3191 |
+
"step": 450
|
| 3192 |
+
},
|
| 3193 |
+
{
|
| 3194 |
+
"epoch": 4.685714285714286,
|
| 3195 |
+
"grad_norm": 0.39316463470458984,
|
| 3196 |
+
"learning_rate": 3.6807017543859645e-07,
|
| 3197 |
+
"loss": 2.0736,
|
| 3198 |
+
"step": 451
|
| 3199 |
+
},
|
| 3200 |
+
{
|
| 3201 |
+
"epoch": 4.696103896103896,
|
| 3202 |
+
"grad_norm": 0.37328803539276123,
|
| 3203 |
+
"learning_rate": 3.6771929824561405e-07,
|
| 3204 |
+
"loss": 2.084,
|
| 3205 |
+
"step": 452
|
| 3206 |
+
},
|
| 3207 |
+
{
|
| 3208 |
+
"epoch": 4.706493506493507,
|
| 3209 |
+
"grad_norm": 0.3884430527687073,
|
| 3210 |
+
"learning_rate": 3.6736842105263153e-07,
|
| 3211 |
+
"loss": 2.0788,
|
| 3212 |
+
"step": 453
|
| 3213 |
+
},
|
| 3214 |
+
{
|
| 3215 |
+
"epoch": 4.716883116883117,
|
| 3216 |
+
"grad_norm": 0.385623574256897,
|
| 3217 |
+
"learning_rate": 3.6701754385964913e-07,
|
| 3218 |
+
"loss": 2.0705,
|
| 3219 |
+
"step": 454
|
| 3220 |
+
},
|
| 3221 |
+
{
|
| 3222 |
+
"epoch": 4.7272727272727275,
|
| 3223 |
+
"grad_norm": 0.38950812816619873,
|
| 3224 |
+
"learning_rate": 3.666666666666666e-07,
|
| 3225 |
+
"loss": 2.0785,
|
| 3226 |
+
"step": 455
|
| 3227 |
+
},
|
| 3228 |
+
{
|
| 3229 |
+
"epoch": 4.7376623376623375,
|
| 3230 |
+
"grad_norm": 0.38535040616989136,
|
| 3231 |
+
"learning_rate": 3.663157894736842e-07,
|
| 3232 |
+
"loss": 2.0909,
|
| 3233 |
+
"step": 456
|
| 3234 |
+
},
|
| 3235 |
+
{
|
| 3236 |
+
"epoch": 4.748051948051948,
|
| 3237 |
+
"grad_norm": 0.3869593143463135,
|
| 3238 |
+
"learning_rate": 3.6596491228070175e-07,
|
| 3239 |
+
"loss": 2.0801,
|
| 3240 |
+
"step": 457
|
| 3241 |
+
},
|
| 3242 |
+
{
|
| 3243 |
+
"epoch": 4.758441558441558,
|
| 3244 |
+
"grad_norm": 0.39084428548812866,
|
| 3245 |
+
"learning_rate": 3.656140350877193e-07,
|
| 3246 |
+
"loss": 2.096,
|
| 3247 |
+
"step": 458
|
| 3248 |
+
},
|
| 3249 |
+
{
|
| 3250 |
+
"epoch": 4.768831168831169,
|
| 3251 |
+
"grad_norm": 0.3794546127319336,
|
| 3252 |
+
"learning_rate": 3.6526315789473683e-07,
|
| 3253 |
+
"loss": 2.0527,
|
| 3254 |
+
"step": 459
|
| 3255 |
+
},
|
| 3256 |
+
{
|
| 3257 |
+
"epoch": 4.779220779220779,
|
| 3258 |
+
"grad_norm": 0.3870809078216553,
|
| 3259 |
+
"learning_rate": 3.6491228070175437e-07,
|
| 3260 |
+
"loss": 2.0853,
|
| 3261 |
+
"step": 460
|
| 3262 |
+
},
|
| 3263 |
+
{
|
| 3264 |
+
"epoch": 4.78961038961039,
|
| 3265 |
+
"grad_norm": 0.38205036520957947,
|
| 3266 |
+
"learning_rate": 3.645614035087719e-07,
|
| 3267 |
+
"loss": 2.0643,
|
| 3268 |
+
"step": 461
|
| 3269 |
+
},
|
| 3270 |
+
{
|
| 3271 |
+
"epoch": 4.8,
|
| 3272 |
+
"grad_norm": 0.3907061815261841,
|
| 3273 |
+
"learning_rate": 3.6421052631578945e-07,
|
| 3274 |
+
"loss": 2.0786,
|
| 3275 |
+
"step": 462
|
| 3276 |
+
},
|
| 3277 |
+
{
|
| 3278 |
+
"epoch": 4.810389610389611,
|
| 3279 |
+
"grad_norm": 0.39493080973625183,
|
| 3280 |
+
"learning_rate": 3.63859649122807e-07,
|
| 3281 |
+
"loss": 2.0944,
|
| 3282 |
+
"step": 463
|
| 3283 |
+
},
|
| 3284 |
+
{
|
| 3285 |
+
"epoch": 4.820779220779221,
|
| 3286 |
+
"grad_norm": 0.3930380046367645,
|
| 3287 |
+
"learning_rate": 3.6350877192982453e-07,
|
| 3288 |
+
"loss": 2.1138,
|
| 3289 |
+
"step": 464
|
| 3290 |
+
},
|
| 3291 |
+
{
|
| 3292 |
+
"epoch": 4.8311688311688314,
|
| 3293 |
+
"grad_norm": 0.3952060639858246,
|
| 3294 |
+
"learning_rate": 3.6315789473684213e-07,
|
| 3295 |
+
"loss": 2.0802,
|
| 3296 |
+
"step": 465
|
| 3297 |
+
},
|
| 3298 |
+
{
|
| 3299 |
+
"epoch": 4.841558441558441,
|
| 3300 |
+
"grad_norm": 0.3815995752811432,
|
| 3301 |
+
"learning_rate": 3.628070175438596e-07,
|
| 3302 |
+
"loss": 2.0838,
|
| 3303 |
+
"step": 466
|
| 3304 |
+
},
|
| 3305 |
+
{
|
| 3306 |
+
"epoch": 4.851948051948052,
|
| 3307 |
+
"grad_norm": 0.38858020305633545,
|
| 3308 |
+
"learning_rate": 3.6245614035087716e-07,
|
| 3309 |
+
"loss": 2.0804,
|
| 3310 |
+
"step": 467
|
| 3311 |
+
},
|
| 3312 |
+
{
|
| 3313 |
+
"epoch": 4.862337662337662,
|
| 3314 |
+
"grad_norm": 0.385565847158432,
|
| 3315 |
+
"learning_rate": 3.6210526315789475e-07,
|
| 3316 |
+
"loss": 2.0974,
|
| 3317 |
+
"step": 468
|
| 3318 |
+
},
|
| 3319 |
+
{
|
| 3320 |
+
"epoch": 4.872727272727273,
|
| 3321 |
+
"grad_norm": 0.3909178078174591,
|
| 3322 |
+
"learning_rate": 3.6175438596491224e-07,
|
| 3323 |
+
"loss": 2.0887,
|
| 3324 |
+
"step": 469
|
| 3325 |
+
},
|
| 3326 |
+
{
|
| 3327 |
+
"epoch": 4.883116883116883,
|
| 3328 |
+
"grad_norm": 0.3982325792312622,
|
| 3329 |
+
"learning_rate": 3.6140350877192983e-07,
|
| 3330 |
+
"loss": 2.1054,
|
| 3331 |
+
"step": 470
|
| 3332 |
+
},
|
| 3333 |
+
{
|
| 3334 |
+
"epoch": 4.893506493506494,
|
| 3335 |
+
"grad_norm": 0.3876339793205261,
|
| 3336 |
+
"learning_rate": 3.610526315789473e-07,
|
| 3337 |
+
"loss": 2.1054,
|
| 3338 |
+
"step": 471
|
| 3339 |
+
},
|
| 3340 |
+
{
|
| 3341 |
+
"epoch": 4.903896103896104,
|
| 3342 |
+
"grad_norm": 0.3819069266319275,
|
| 3343 |
+
"learning_rate": 3.607017543859649e-07,
|
| 3344 |
+
"loss": 2.0821,
|
| 3345 |
+
"step": 472
|
| 3346 |
+
},
|
| 3347 |
+
{
|
| 3348 |
+
"epoch": 4.914285714285715,
|
| 3349 |
+
"grad_norm": 0.3924694359302521,
|
| 3350 |
+
"learning_rate": 3.6035087719298245e-07,
|
| 3351 |
+
"loss": 2.0712,
|
| 3352 |
+
"step": 473
|
| 3353 |
+
},
|
| 3354 |
+
{
|
| 3355 |
+
"epoch": 4.9246753246753245,
|
| 3356 |
+
"grad_norm": 0.3937675654888153,
|
| 3357 |
+
"learning_rate": 3.6e-07,
|
| 3358 |
+
"loss": 2.1057,
|
| 3359 |
+
"step": 474
|
| 3360 |
+
},
|
| 3361 |
+
{
|
| 3362 |
+
"epoch": 4.935064935064935,
|
| 3363 |
+
"grad_norm": 0.38620275259017944,
|
| 3364 |
+
"learning_rate": 3.5964912280701754e-07,
|
| 3365 |
+
"loss": 2.0845,
|
| 3366 |
+
"step": 475
|
| 3367 |
+
},
|
| 3368 |
+
{
|
| 3369 |
+
"epoch": 4.945454545454545,
|
| 3370 |
+
"grad_norm": 0.40442150831222534,
|
| 3371 |
+
"learning_rate": 3.59298245614035e-07,
|
| 3372 |
+
"loss": 2.0936,
|
| 3373 |
+
"step": 476
|
| 3374 |
+
},
|
| 3375 |
+
{
|
| 3376 |
+
"epoch": 4.955844155844156,
|
| 3377 |
+
"grad_norm": 0.3815317153930664,
|
| 3378 |
+
"learning_rate": 3.589473684210526e-07,
|
| 3379 |
+
"loss": 2.0845,
|
| 3380 |
+
"step": 477
|
| 3381 |
+
},
|
| 3382 |
+
{
|
| 3383 |
+
"epoch": 4.966233766233766,
|
| 3384 |
+
"grad_norm": 0.38584476709365845,
|
| 3385 |
+
"learning_rate": 3.5859649122807016e-07,
|
| 3386 |
+
"loss": 2.071,
|
| 3387 |
+
"step": 478
|
| 3388 |
+
},
|
| 3389 |
+
{
|
| 3390 |
+
"epoch": 4.976623376623377,
|
| 3391 |
+
"grad_norm": 0.3887505829334259,
|
| 3392 |
+
"learning_rate": 3.582456140350877e-07,
|
| 3393 |
+
"loss": 2.0924,
|
| 3394 |
+
"step": 479
|
| 3395 |
+
},
|
| 3396 |
+
{
|
| 3397 |
+
"epoch": 4.987012987012987,
|
| 3398 |
+
"grad_norm": 0.3836219012737274,
|
| 3399 |
+
"learning_rate": 3.5789473684210524e-07,
|
| 3400 |
+
"loss": 2.0986,
|
| 3401 |
+
"step": 480
|
| 3402 |
+
},
|
| 3403 |
+
{
|
| 3404 |
+
"epoch": 4.997402597402598,
|
| 3405 |
+
"grad_norm": 0.38430362939834595,
|
| 3406 |
+
"learning_rate": 3.5754385964912283e-07,
|
| 3407 |
+
"loss": 2.1006,
|
| 3408 |
+
"step": 481
|
| 3409 |
+
},
|
| 3410 |
+
{
|
| 3411 |
+
"epoch": 4.997402597402598,
|
| 3412 |
+
"eval_loss": 2.0890417098999023,
|
| 3413 |
+
"eval_runtime": 9.2277,
|
| 3414 |
+
"eval_samples_per_second": 2.709,
|
| 3415 |
+
"eval_steps_per_second": 0.433,
|
| 3416 |
+
"step": 481
|
| 3417 |
}
|
| 3418 |
],
|
| 3419 |
"logging_steps": 1,
|
|
|
|
| 3433 |
"attributes": {}
|
| 3434 |
}
|
| 3435 |
},
|
| 3436 |
+
"total_flos": 5.690178034414387e+18,
|
| 3437 |
"train_batch_size": 8,
|
| 3438 |
"trial_name": null,
|
| 3439 |
"trial_params": null
|