Upload folder using huggingface_hub
Browse files- adapter_model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +5 -3604
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 13648432
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4b7b015e42e173818fea90cd734464a0f91cc0f673d9774615a292dfea18e1d3
|
| 3 |
size 13648432
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 27338810
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:38b34b2b6c04fc909904a419a9220b3ab237bd22eee93af56cc2e03983e68b94
|
| 3 |
size 27338810
|
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 sha256:39dbd3f37f16ee432d6b762708109d2e74798485e2714032f47750ffe2268ca1
|
| 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 sha256:836e374dfbd903836045d35c4066a8ab079349c0c581eecdecc125ab8c4439e1
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 2.
|
| 3 |
-
"best_model_checkpoint": "/home/sunggeunan/data/ICL/outputs/lora/SKIML-ICL_mrqa_nq_v3/Meta-Llama-3-8B-Instruct-unanswerable-5Q-0U-0C-qa_first/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -3599,3605 +3599,6 @@
|
|
| 3599 |
"eval_samples_per_second": 1.293,
|
| 3600 |
"eval_steps_per_second": 0.324,
|
| 3601 |
"step": 512
|
| 3602 |
-
},
|
| 3603 |
-
{
|
| 3604 |
-
"epoch": 1.0002437241043138,
|
| 3605 |
-
"grad_norm": 0.41433945298194885,
|
| 3606 |
-
"learning_rate": 3.505825908156271e-07,
|
| 3607 |
-
"loss": 2.0366,
|
| 3608 |
-
"step": 513
|
| 3609 |
-
},
|
| 3610 |
-
{
|
| 3611 |
-
"epoch": 1.0021935169388252,
|
| 3612 |
-
"grad_norm": 0.42399510741233826,
|
| 3613 |
-
"learning_rate": 3.5023989033584646e-07,
|
| 3614 |
-
"loss": 1.991,
|
| 3615 |
-
"step": 514
|
| 3616 |
-
},
|
| 3617 |
-
{
|
| 3618 |
-
"epoch": 1.0041433097733365,
|
| 3619 |
-
"grad_norm": 0.45652541518211365,
|
| 3620 |
-
"learning_rate": 3.4989718985606576e-07,
|
| 3621 |
-
"loss": 2.0066,
|
| 3622 |
-
"step": 515
|
| 3623 |
-
},
|
| 3624 |
-
{
|
| 3625 |
-
"epoch": 1.006093102607848,
|
| 3626 |
-
"grad_norm": 0.43585795164108276,
|
| 3627 |
-
"learning_rate": 3.495544893762851e-07,
|
| 3628 |
-
"loss": 2.0025,
|
| 3629 |
-
"step": 516
|
| 3630 |
-
},
|
| 3631 |
-
{
|
| 3632 |
-
"epoch": 1.0080428954423593,
|
| 3633 |
-
"grad_norm": 0.43803489208221436,
|
| 3634 |
-
"learning_rate": 3.4921178889650445e-07,
|
| 3635 |
-
"loss": 2.0654,
|
| 3636 |
-
"step": 517
|
| 3637 |
-
},
|
| 3638 |
-
{
|
| 3639 |
-
"epoch": 1.0099926882768706,
|
| 3640 |
-
"grad_norm": 0.43803176283836365,
|
| 3641 |
-
"learning_rate": 3.4886908841672375e-07,
|
| 3642 |
-
"loss": 2.0896,
|
| 3643 |
-
"step": 518
|
| 3644 |
-
},
|
| 3645 |
-
{
|
| 3646 |
-
"epoch": 1.011942481111382,
|
| 3647 |
-
"grad_norm": 0.41983944177627563,
|
| 3648 |
-
"learning_rate": 3.485263879369431e-07,
|
| 3649 |
-
"loss": 2.0335,
|
| 3650 |
-
"step": 519
|
| 3651 |
-
},
|
| 3652 |
-
{
|
| 3653 |
-
"epoch": 1.0138922739458933,
|
| 3654 |
-
"grad_norm": 0.4354363977909088,
|
| 3655 |
-
"learning_rate": 3.4818368745716244e-07,
|
| 3656 |
-
"loss": 2.0699,
|
| 3657 |
-
"step": 520
|
| 3658 |
-
},
|
| 3659 |
-
{
|
| 3660 |
-
"epoch": 1.0158420667804047,
|
| 3661 |
-
"grad_norm": 0.42140671610832214,
|
| 3662 |
-
"learning_rate": 3.4784098697738174e-07,
|
| 3663 |
-
"loss": 1.9646,
|
| 3664 |
-
"step": 521
|
| 3665 |
-
},
|
| 3666 |
-
{
|
| 3667 |
-
"epoch": 1.0177918596149158,
|
| 3668 |
-
"grad_norm": 0.4265493154525757,
|
| 3669 |
-
"learning_rate": 3.474982864976011e-07,
|
| 3670 |
-
"loss": 2.0735,
|
| 3671 |
-
"step": 522
|
| 3672 |
-
},
|
| 3673 |
-
{
|
| 3674 |
-
"epoch": 1.0197416524494272,
|
| 3675 |
-
"grad_norm": 0.43847259879112244,
|
| 3676 |
-
"learning_rate": 3.4715558601782043e-07,
|
| 3677 |
-
"loss": 2.0986,
|
| 3678 |
-
"step": 523
|
| 3679 |
-
},
|
| 3680 |
-
{
|
| 3681 |
-
"epoch": 1.0216914452839385,
|
| 3682 |
-
"grad_norm": 0.4600801467895508,
|
| 3683 |
-
"learning_rate": 3.468128855380397e-07,
|
| 3684 |
-
"loss": 2.0643,
|
| 3685 |
-
"step": 524
|
| 3686 |
-
},
|
| 3687 |
-
{
|
| 3688 |
-
"epoch": 1.0236412381184499,
|
| 3689 |
-
"grad_norm": 0.42904648184776306,
|
| 3690 |
-
"learning_rate": 3.4647018505825907e-07,
|
| 3691 |
-
"loss": 2.0056,
|
| 3692 |
-
"step": 525
|
| 3693 |
-
},
|
| 3694 |
-
{
|
| 3695 |
-
"epoch": 1.0255910309529612,
|
| 3696 |
-
"grad_norm": 0.46431151032447815,
|
| 3697 |
-
"learning_rate": 3.461274845784784e-07,
|
| 3698 |
-
"loss": 2.1056,
|
| 3699 |
-
"step": 526
|
| 3700 |
-
},
|
| 3701 |
-
{
|
| 3702 |
-
"epoch": 1.0275408237874726,
|
| 3703 |
-
"grad_norm": 0.455836683511734,
|
| 3704 |
-
"learning_rate": 3.457847840986977e-07,
|
| 3705 |
-
"loss": 2.0187,
|
| 3706 |
-
"step": 527
|
| 3707 |
-
},
|
| 3708 |
-
{
|
| 3709 |
-
"epoch": 1.029490616621984,
|
| 3710 |
-
"grad_norm": 0.4192461669445038,
|
| 3711 |
-
"learning_rate": 3.45442083618917e-07,
|
| 3712 |
-
"loss": 2.0832,
|
| 3713 |
-
"step": 528
|
| 3714 |
-
},
|
| 3715 |
-
{
|
| 3716 |
-
"epoch": 1.0314404094564953,
|
| 3717 |
-
"grad_norm": 0.4513595402240753,
|
| 3718 |
-
"learning_rate": 3.450993831391364e-07,
|
| 3719 |
-
"loss": 2.058,
|
| 3720 |
-
"step": 529
|
| 3721 |
-
},
|
| 3722 |
-
{
|
| 3723 |
-
"epoch": 1.0333902022910066,
|
| 3724 |
-
"grad_norm": 0.4370152950286865,
|
| 3725 |
-
"learning_rate": 3.447566826593557e-07,
|
| 3726 |
-
"loss": 2.0537,
|
| 3727 |
-
"step": 530
|
| 3728 |
-
},
|
| 3729 |
-
{
|
| 3730 |
-
"epoch": 1.035339995125518,
|
| 3731 |
-
"grad_norm": 0.4199161231517792,
|
| 3732 |
-
"learning_rate": 3.44413982179575e-07,
|
| 3733 |
-
"loss": 1.9518,
|
| 3734 |
-
"step": 531
|
| 3735 |
-
},
|
| 3736 |
-
{
|
| 3737 |
-
"epoch": 1.0372897879600294,
|
| 3738 |
-
"grad_norm": 0.43688762187957764,
|
| 3739 |
-
"learning_rate": 3.440712816997944e-07,
|
| 3740 |
-
"loss": 2.0444,
|
| 3741 |
-
"step": 532
|
| 3742 |
-
},
|
| 3743 |
-
{
|
| 3744 |
-
"epoch": 1.0392395807945405,
|
| 3745 |
-
"grad_norm": 0.49809253215789795,
|
| 3746 |
-
"learning_rate": 3.437285812200137e-07,
|
| 3747 |
-
"loss": 2.0401,
|
| 3748 |
-
"step": 533
|
| 3749 |
-
},
|
| 3750 |
-
{
|
| 3751 |
-
"epoch": 1.0411893736290518,
|
| 3752 |
-
"grad_norm": 0.4518781900405884,
|
| 3753 |
-
"learning_rate": 3.43385880740233e-07,
|
| 3754 |
-
"loss": 2.0605,
|
| 3755 |
-
"step": 534
|
| 3756 |
-
},
|
| 3757 |
-
{
|
| 3758 |
-
"epoch": 1.0431391664635632,
|
| 3759 |
-
"grad_norm": 0.45353132486343384,
|
| 3760 |
-
"learning_rate": 3.430431802604524e-07,
|
| 3761 |
-
"loss": 2.0402,
|
| 3762 |
-
"step": 535
|
| 3763 |
-
},
|
| 3764 |
-
{
|
| 3765 |
-
"epoch": 1.0450889592980745,
|
| 3766 |
-
"grad_norm": 0.4396359622478485,
|
| 3767 |
-
"learning_rate": 3.427004797806717e-07,
|
| 3768 |
-
"loss": 2.0643,
|
| 3769 |
-
"step": 536
|
| 3770 |
-
},
|
| 3771 |
-
{
|
| 3772 |
-
"epoch": 1.047038752132586,
|
| 3773 |
-
"grad_norm": 0.4434252083301544,
|
| 3774 |
-
"learning_rate": 3.42357779300891e-07,
|
| 3775 |
-
"loss": 2.0188,
|
| 3776 |
-
"step": 537
|
| 3777 |
-
},
|
| 3778 |
-
{
|
| 3779 |
-
"epoch": 1.0489885449670973,
|
| 3780 |
-
"grad_norm": 0.4241044819355011,
|
| 3781 |
-
"learning_rate": 3.420150788211104e-07,
|
| 3782 |
-
"loss": 1.9556,
|
| 3783 |
-
"step": 538
|
| 3784 |
-
},
|
| 3785 |
-
{
|
| 3786 |
-
"epoch": 1.0509383378016086,
|
| 3787 |
-
"grad_norm": 0.4382232129573822,
|
| 3788 |
-
"learning_rate": 3.4167237834132967e-07,
|
| 3789 |
-
"loss": 1.9855,
|
| 3790 |
-
"step": 539
|
| 3791 |
-
},
|
| 3792 |
-
{
|
| 3793 |
-
"epoch": 1.05288813063612,
|
| 3794 |
-
"grad_norm": 0.4357564151287079,
|
| 3795 |
-
"learning_rate": 3.4132967786154897e-07,
|
| 3796 |
-
"loss": 2.0524,
|
| 3797 |
-
"step": 540
|
| 3798 |
-
},
|
| 3799 |
-
{
|
| 3800 |
-
"epoch": 1.0548379234706313,
|
| 3801 |
-
"grad_norm": 0.46050140261650085,
|
| 3802 |
-
"learning_rate": 3.409869773817683e-07,
|
| 3803 |
-
"loss": 2.0461,
|
| 3804 |
-
"step": 541
|
| 3805 |
-
},
|
| 3806 |
-
{
|
| 3807 |
-
"epoch": 1.0567877163051427,
|
| 3808 |
-
"grad_norm": 0.44581982493400574,
|
| 3809 |
-
"learning_rate": 3.4064427690198766e-07,
|
| 3810 |
-
"loss": 1.9955,
|
| 3811 |
-
"step": 542
|
| 3812 |
-
},
|
| 3813 |
-
{
|
| 3814 |
-
"epoch": 1.0587375091396538,
|
| 3815 |
-
"grad_norm": 0.4502599835395813,
|
| 3816 |
-
"learning_rate": 3.4030157642220696e-07,
|
| 3817 |
-
"loss": 2.0864,
|
| 3818 |
-
"step": 543
|
| 3819 |
-
},
|
| 3820 |
-
{
|
| 3821 |
-
"epoch": 1.0606873019741652,
|
| 3822 |
-
"grad_norm": 0.44767019152641296,
|
| 3823 |
-
"learning_rate": 3.399588759424263e-07,
|
| 3824 |
-
"loss": 2.0447,
|
| 3825 |
-
"step": 544
|
| 3826 |
-
},
|
| 3827 |
-
{
|
| 3828 |
-
"epoch": 1.0626370948086765,
|
| 3829 |
-
"grad_norm": 0.44603490829467773,
|
| 3830 |
-
"learning_rate": 3.3961617546264565e-07,
|
| 3831 |
-
"loss": 2.0709,
|
| 3832 |
-
"step": 545
|
| 3833 |
-
},
|
| 3834 |
-
{
|
| 3835 |
-
"epoch": 1.0645868876431879,
|
| 3836 |
-
"grad_norm": 0.4321264922618866,
|
| 3837 |
-
"learning_rate": 3.3927347498286495e-07,
|
| 3838 |
-
"loss": 2.0157,
|
| 3839 |
-
"step": 546
|
| 3840 |
-
},
|
| 3841 |
-
{
|
| 3842 |
-
"epoch": 1.0665366804776992,
|
| 3843 |
-
"grad_norm": 0.4479556083679199,
|
| 3844 |
-
"learning_rate": 3.389307745030843e-07,
|
| 3845 |
-
"loss": 2.1088,
|
| 3846 |
-
"step": 547
|
| 3847 |
-
},
|
| 3848 |
-
{
|
| 3849 |
-
"epoch": 1.0684864733122106,
|
| 3850 |
-
"grad_norm": 0.4292636513710022,
|
| 3851 |
-
"learning_rate": 3.3858807402330364e-07,
|
| 3852 |
-
"loss": 2.0847,
|
| 3853 |
-
"step": 548
|
| 3854 |
-
},
|
| 3855 |
-
{
|
| 3856 |
-
"epoch": 1.070436266146722,
|
| 3857 |
-
"grad_norm": 0.43631821870803833,
|
| 3858 |
-
"learning_rate": 3.3824537354352294e-07,
|
| 3859 |
-
"loss": 2.034,
|
| 3860 |
-
"step": 549
|
| 3861 |
-
},
|
| 3862 |
-
{
|
| 3863 |
-
"epoch": 1.0723860589812333,
|
| 3864 |
-
"grad_norm": 0.43201327323913574,
|
| 3865 |
-
"learning_rate": 3.379026730637423e-07,
|
| 3866 |
-
"loss": 1.9633,
|
| 3867 |
-
"step": 550
|
| 3868 |
-
},
|
| 3869 |
-
{
|
| 3870 |
-
"epoch": 1.0743358518157446,
|
| 3871 |
-
"grad_norm": 0.4389747679233551,
|
| 3872 |
-
"learning_rate": 3.3755997258396163e-07,
|
| 3873 |
-
"loss": 2.0331,
|
| 3874 |
-
"step": 551
|
| 3875 |
-
},
|
| 3876 |
-
{
|
| 3877 |
-
"epoch": 1.076285644650256,
|
| 3878 |
-
"grad_norm": 0.46588924527168274,
|
| 3879 |
-
"learning_rate": 3.372172721041809e-07,
|
| 3880 |
-
"loss": 2.0748,
|
| 3881 |
-
"step": 552
|
| 3882 |
-
},
|
| 3883 |
-
{
|
| 3884 |
-
"epoch": 1.0782354374847674,
|
| 3885 |
-
"grad_norm": 0.45190852880477905,
|
| 3886 |
-
"learning_rate": 3.3687457162440027e-07,
|
| 3887 |
-
"loss": 1.9639,
|
| 3888 |
-
"step": 553
|
| 3889 |
-
},
|
| 3890 |
-
{
|
| 3891 |
-
"epoch": 1.0801852303192785,
|
| 3892 |
-
"grad_norm": 0.4458979070186615,
|
| 3893 |
-
"learning_rate": 3.3653187114461957e-07,
|
| 3894 |
-
"loss": 2.1124,
|
| 3895 |
-
"step": 554
|
| 3896 |
-
},
|
| 3897 |
-
{
|
| 3898 |
-
"epoch": 1.0821350231537898,
|
| 3899 |
-
"grad_norm": 0.40400832891464233,
|
| 3900 |
-
"learning_rate": 3.361891706648389e-07,
|
| 3901 |
-
"loss": 1.9776,
|
| 3902 |
-
"step": 555
|
| 3903 |
-
},
|
| 3904 |
-
{
|
| 3905 |
-
"epoch": 1.0840848159883012,
|
| 3906 |
-
"grad_norm": 0.4538462460041046,
|
| 3907 |
-
"learning_rate": 3.3584647018505826e-07,
|
| 3908 |
-
"loss": 1.9962,
|
| 3909 |
-
"step": 556
|
| 3910 |
-
},
|
| 3911 |
-
{
|
| 3912 |
-
"epoch": 1.0860346088228126,
|
| 3913 |
-
"grad_norm": 0.44181132316589355,
|
| 3914 |
-
"learning_rate": 3.3550376970527756e-07,
|
| 3915 |
-
"loss": 2.0973,
|
| 3916 |
-
"step": 557
|
| 3917 |
-
},
|
| 3918 |
-
{
|
| 3919 |
-
"epoch": 1.087984401657324,
|
| 3920 |
-
"grad_norm": 0.43516308069229126,
|
| 3921 |
-
"learning_rate": 3.351610692254969e-07,
|
| 3922 |
-
"loss": 1.9923,
|
| 3923 |
-
"step": 558
|
| 3924 |
-
},
|
| 3925 |
-
{
|
| 3926 |
-
"epoch": 1.0899341944918353,
|
| 3927 |
-
"grad_norm": 0.4485546052455902,
|
| 3928 |
-
"learning_rate": 3.3481836874571625e-07,
|
| 3929 |
-
"loss": 2.0242,
|
| 3930 |
-
"step": 559
|
| 3931 |
-
},
|
| 3932 |
-
{
|
| 3933 |
-
"epoch": 1.0918839873263466,
|
| 3934 |
-
"grad_norm": 0.45358070731163025,
|
| 3935 |
-
"learning_rate": 3.3447566826593555e-07,
|
| 3936 |
-
"loss": 2.0603,
|
| 3937 |
-
"step": 560
|
| 3938 |
-
},
|
| 3939 |
-
{
|
| 3940 |
-
"epoch": 1.093833780160858,
|
| 3941 |
-
"grad_norm": 0.43879690766334534,
|
| 3942 |
-
"learning_rate": 3.341329677861549e-07,
|
| 3943 |
-
"loss": 1.9869,
|
| 3944 |
-
"step": 561
|
| 3945 |
-
},
|
| 3946 |
-
{
|
| 3947 |
-
"epoch": 1.0957835729953693,
|
| 3948 |
-
"grad_norm": 0.4376320242881775,
|
| 3949 |
-
"learning_rate": 3.3379026730637424e-07,
|
| 3950 |
-
"loss": 2.0447,
|
| 3951 |
-
"step": 562
|
| 3952 |
-
},
|
| 3953 |
-
{
|
| 3954 |
-
"epoch": 1.0977333658298805,
|
| 3955 |
-
"grad_norm": 0.4591986835002899,
|
| 3956 |
-
"learning_rate": 3.3344756682659354e-07,
|
| 3957 |
-
"loss": 2.0188,
|
| 3958 |
-
"step": 563
|
| 3959 |
-
},
|
| 3960 |
-
{
|
| 3961 |
-
"epoch": 1.0996831586643918,
|
| 3962 |
-
"grad_norm": 0.4306589961051941,
|
| 3963 |
-
"learning_rate": 3.331048663468129e-07,
|
| 3964 |
-
"loss": 2.0223,
|
| 3965 |
-
"step": 564
|
| 3966 |
-
},
|
| 3967 |
-
{
|
| 3968 |
-
"epoch": 1.1016329514989032,
|
| 3969 |
-
"grad_norm": 0.43692710995674133,
|
| 3970 |
-
"learning_rate": 3.3276216586703223e-07,
|
| 3971 |
-
"loss": 2.0507,
|
| 3972 |
-
"step": 565
|
| 3973 |
-
},
|
| 3974 |
-
{
|
| 3975 |
-
"epoch": 1.1035827443334145,
|
| 3976 |
-
"grad_norm": 0.4663935601711273,
|
| 3977 |
-
"learning_rate": 3.324194653872515e-07,
|
| 3978 |
-
"loss": 2.0444,
|
| 3979 |
-
"step": 566
|
| 3980 |
-
},
|
| 3981 |
-
{
|
| 3982 |
-
"epoch": 1.1055325371679259,
|
| 3983 |
-
"grad_norm": 0.45090562105178833,
|
| 3984 |
-
"learning_rate": 3.320767649074708e-07,
|
| 3985 |
-
"loss": 1.9944,
|
| 3986 |
-
"step": 567
|
| 3987 |
-
},
|
| 3988 |
-
{
|
| 3989 |
-
"epoch": 1.1074823300024372,
|
| 3990 |
-
"grad_norm": 0.4450632631778717,
|
| 3991 |
-
"learning_rate": 3.317340644276902e-07,
|
| 3992 |
-
"loss": 2.0264,
|
| 3993 |
-
"step": 568
|
| 3994 |
-
},
|
| 3995 |
-
{
|
| 3996 |
-
"epoch": 1.1094321228369486,
|
| 3997 |
-
"grad_norm": 0.45126745104789734,
|
| 3998 |
-
"learning_rate": 3.313913639479095e-07,
|
| 3999 |
-
"loss": 2.081,
|
| 4000 |
-
"step": 569
|
| 4001 |
-
},
|
| 4002 |
-
{
|
| 4003 |
-
"epoch": 1.11138191567146,
|
| 4004 |
-
"grad_norm": 0.44254472851753235,
|
| 4005 |
-
"learning_rate": 3.310486634681288e-07,
|
| 4006 |
-
"loss": 2.0223,
|
| 4007 |
-
"step": 570
|
| 4008 |
-
},
|
| 4009 |
-
{
|
| 4010 |
-
"epoch": 1.1133317085059713,
|
| 4011 |
-
"grad_norm": 0.43211621046066284,
|
| 4012 |
-
"learning_rate": 3.307059629883482e-07,
|
| 4013 |
-
"loss": 2.0363,
|
| 4014 |
-
"step": 571
|
| 4015 |
-
},
|
| 4016 |
-
{
|
| 4017 |
-
"epoch": 1.1152815013404827,
|
| 4018 |
-
"grad_norm": 0.4256265163421631,
|
| 4019 |
-
"learning_rate": 3.303632625085675e-07,
|
| 4020 |
-
"loss": 2.0363,
|
| 4021 |
-
"step": 572
|
| 4022 |
-
},
|
| 4023 |
-
{
|
| 4024 |
-
"epoch": 1.117231294174994,
|
| 4025 |
-
"grad_norm": 0.4462417960166931,
|
| 4026 |
-
"learning_rate": 3.300205620287868e-07,
|
| 4027 |
-
"loss": 2.0394,
|
| 4028 |
-
"step": 573
|
| 4029 |
-
},
|
| 4030 |
-
{
|
| 4031 |
-
"epoch": 1.1191810870095051,
|
| 4032 |
-
"grad_norm": 0.4583437442779541,
|
| 4033 |
-
"learning_rate": 3.296778615490062e-07,
|
| 4034 |
-
"loss": 2.0878,
|
| 4035 |
-
"step": 574
|
| 4036 |
-
},
|
| 4037 |
-
{
|
| 4038 |
-
"epoch": 1.1211308798440165,
|
| 4039 |
-
"grad_norm": 0.4595088064670563,
|
| 4040 |
-
"learning_rate": 3.293351610692255e-07,
|
| 4041 |
-
"loss": 2.111,
|
| 4042 |
-
"step": 575
|
| 4043 |
-
},
|
| 4044 |
-
{
|
| 4045 |
-
"epoch": 1.1230806726785278,
|
| 4046 |
-
"grad_norm": 0.4117080569267273,
|
| 4047 |
-
"learning_rate": 3.289924605894448e-07,
|
| 4048 |
-
"loss": 1.999,
|
| 4049 |
-
"step": 576
|
| 4050 |
-
},
|
| 4051 |
-
{
|
| 4052 |
-
"epoch": 1.1250304655130392,
|
| 4053 |
-
"grad_norm": 0.4381641149520874,
|
| 4054 |
-
"learning_rate": 3.286497601096642e-07,
|
| 4055 |
-
"loss": 2.044,
|
| 4056 |
-
"step": 577
|
| 4057 |
-
},
|
| 4058 |
-
{
|
| 4059 |
-
"epoch": 1.1269802583475506,
|
| 4060 |
-
"grad_norm": 0.43854039907455444,
|
| 4061 |
-
"learning_rate": 3.283070596298835e-07,
|
| 4062 |
-
"loss": 2.0272,
|
| 4063 |
-
"step": 578
|
| 4064 |
-
},
|
| 4065 |
-
{
|
| 4066 |
-
"epoch": 1.128930051182062,
|
| 4067 |
-
"grad_norm": 0.4721965789794922,
|
| 4068 |
-
"learning_rate": 3.279643591501028e-07,
|
| 4069 |
-
"loss": 2.0697,
|
| 4070 |
-
"step": 579
|
| 4071 |
-
},
|
| 4072 |
-
{
|
| 4073 |
-
"epoch": 1.1308798440165733,
|
| 4074 |
-
"grad_norm": 0.4373783767223358,
|
| 4075 |
-
"learning_rate": 3.2762165867032207e-07,
|
| 4076 |
-
"loss": 2.0102,
|
| 4077 |
-
"step": 580
|
| 4078 |
-
},
|
| 4079 |
-
{
|
| 4080 |
-
"epoch": 1.1328296368510846,
|
| 4081 |
-
"grad_norm": 0.4286502003669739,
|
| 4082 |
-
"learning_rate": 3.2727895819054147e-07,
|
| 4083 |
-
"loss": 1.9695,
|
| 4084 |
-
"step": 581
|
| 4085 |
-
},
|
| 4086 |
-
{
|
| 4087 |
-
"epoch": 1.134779429685596,
|
| 4088 |
-
"grad_norm": 0.4373305141925812,
|
| 4089 |
-
"learning_rate": 3.2693625771076077e-07,
|
| 4090 |
-
"loss": 1.9823,
|
| 4091 |
-
"step": 582
|
| 4092 |
-
},
|
| 4093 |
-
{
|
| 4094 |
-
"epoch": 1.1367292225201073,
|
| 4095 |
-
"grad_norm": 0.4659106433391571,
|
| 4096 |
-
"learning_rate": 3.2659355723098006e-07,
|
| 4097 |
-
"loss": 2.081,
|
| 4098 |
-
"step": 583
|
| 4099 |
-
},
|
| 4100 |
-
{
|
| 4101 |
-
"epoch": 1.1386790153546187,
|
| 4102 |
-
"grad_norm": 0.4315546154975891,
|
| 4103 |
-
"learning_rate": 3.2625085675119946e-07,
|
| 4104 |
-
"loss": 2.0336,
|
| 4105 |
-
"step": 584
|
| 4106 |
-
},
|
| 4107 |
-
{
|
| 4108 |
-
"epoch": 1.1406288081891298,
|
| 4109 |
-
"grad_norm": 0.4512901306152344,
|
| 4110 |
-
"learning_rate": 3.2590815627141876e-07,
|
| 4111 |
-
"loss": 2.0642,
|
| 4112 |
-
"step": 585
|
| 4113 |
-
},
|
| 4114 |
-
{
|
| 4115 |
-
"epoch": 1.1425786010236412,
|
| 4116 |
-
"grad_norm": 0.4398232400417328,
|
| 4117 |
-
"learning_rate": 3.2556545579163805e-07,
|
| 4118 |
-
"loss": 2.0401,
|
| 4119 |
-
"step": 586
|
| 4120 |
-
},
|
| 4121 |
-
{
|
| 4122 |
-
"epoch": 1.1445283938581525,
|
| 4123 |
-
"grad_norm": 0.45262405276298523,
|
| 4124 |
-
"learning_rate": 3.2522275531185745e-07,
|
| 4125 |
-
"loss": 2.0999,
|
| 4126 |
-
"step": 587
|
| 4127 |
-
},
|
| 4128 |
-
{
|
| 4129 |
-
"epoch": 1.1464781866926639,
|
| 4130 |
-
"grad_norm": 0.4210640490055084,
|
| 4131 |
-
"learning_rate": 3.2488005483207675e-07,
|
| 4132 |
-
"loss": 1.992,
|
| 4133 |
-
"step": 588
|
| 4134 |
-
},
|
| 4135 |
-
{
|
| 4136 |
-
"epoch": 1.1484279795271752,
|
| 4137 |
-
"grad_norm": 0.4530121386051178,
|
| 4138 |
-
"learning_rate": 3.2453735435229604e-07,
|
| 4139 |
-
"loss": 2.0119,
|
| 4140 |
-
"step": 589
|
| 4141 |
-
},
|
| 4142 |
-
{
|
| 4143 |
-
"epoch": 1.1503777723616866,
|
| 4144 |
-
"grad_norm": 0.43637722730636597,
|
| 4145 |
-
"learning_rate": 3.2419465387251544e-07,
|
| 4146 |
-
"loss": 2.0022,
|
| 4147 |
-
"step": 590
|
| 4148 |
-
},
|
| 4149 |
-
{
|
| 4150 |
-
"epoch": 1.152327565196198,
|
| 4151 |
-
"grad_norm": 0.46872228384017944,
|
| 4152 |
-
"learning_rate": 3.2385195339273474e-07,
|
| 4153 |
-
"loss": 2.0545,
|
| 4154 |
-
"step": 591
|
| 4155 |
-
},
|
| 4156 |
-
{
|
| 4157 |
-
"epoch": 1.1542773580307093,
|
| 4158 |
-
"grad_norm": 0.45964333415031433,
|
| 4159 |
-
"learning_rate": 3.2350925291295403e-07,
|
| 4160 |
-
"loss": 2.0313,
|
| 4161 |
-
"step": 592
|
| 4162 |
-
},
|
| 4163 |
-
{
|
| 4164 |
-
"epoch": 1.1562271508652207,
|
| 4165 |
-
"grad_norm": 0.4444529414176941,
|
| 4166 |
-
"learning_rate": 3.231665524331734e-07,
|
| 4167 |
-
"loss": 2.0463,
|
| 4168 |
-
"step": 593
|
| 4169 |
-
},
|
| 4170 |
-
{
|
| 4171 |
-
"epoch": 1.1581769436997318,
|
| 4172 |
-
"grad_norm": 0.4702310264110565,
|
| 4173 |
-
"learning_rate": 3.228238519533927e-07,
|
| 4174 |
-
"loss": 2.0055,
|
| 4175 |
-
"step": 594
|
| 4176 |
-
},
|
| 4177 |
-
{
|
| 4178 |
-
"epoch": 1.1601267365342431,
|
| 4179 |
-
"grad_norm": 0.4435891807079315,
|
| 4180 |
-
"learning_rate": 3.22481151473612e-07,
|
| 4181 |
-
"loss": 2.1027,
|
| 4182 |
-
"step": 595
|
| 4183 |
-
},
|
| 4184 |
-
{
|
| 4185 |
-
"epoch": 1.1620765293687545,
|
| 4186 |
-
"grad_norm": 0.4557732343673706,
|
| 4187 |
-
"learning_rate": 3.2213845099383137e-07,
|
| 4188 |
-
"loss": 2.0307,
|
| 4189 |
-
"step": 596
|
| 4190 |
-
},
|
| 4191 |
-
{
|
| 4192 |
-
"epoch": 1.1640263222032659,
|
| 4193 |
-
"grad_norm": 0.4286348819732666,
|
| 4194 |
-
"learning_rate": 3.217957505140507e-07,
|
| 4195 |
-
"loss": 2.0196,
|
| 4196 |
-
"step": 597
|
| 4197 |
-
},
|
| 4198 |
-
{
|
| 4199 |
-
"epoch": 1.1659761150377772,
|
| 4200 |
-
"grad_norm": 0.4475346803665161,
|
| 4201 |
-
"learning_rate": 3.2145305003427e-07,
|
| 4202 |
-
"loss": 2.1014,
|
| 4203 |
-
"step": 598
|
| 4204 |
-
},
|
| 4205 |
-
{
|
| 4206 |
-
"epoch": 1.1679259078722886,
|
| 4207 |
-
"grad_norm": 0.418293297290802,
|
| 4208 |
-
"learning_rate": 3.2111034955448936e-07,
|
| 4209 |
-
"loss": 2.078,
|
| 4210 |
-
"step": 599
|
| 4211 |
-
},
|
| 4212 |
-
{
|
| 4213 |
-
"epoch": 1.1698757007068,
|
| 4214 |
-
"grad_norm": 0.42740973830223083,
|
| 4215 |
-
"learning_rate": 3.207676490747087e-07,
|
| 4216 |
-
"loss": 1.9695,
|
| 4217 |
-
"step": 600
|
| 4218 |
-
},
|
| 4219 |
-
{
|
| 4220 |
-
"epoch": 1.1718254935413113,
|
| 4221 |
-
"grad_norm": 0.41325512528419495,
|
| 4222 |
-
"learning_rate": 3.20424948594928e-07,
|
| 4223 |
-
"loss": 2.0297,
|
| 4224 |
-
"step": 601
|
| 4225 |
-
},
|
| 4226 |
-
{
|
| 4227 |
-
"epoch": 1.1737752863758226,
|
| 4228 |
-
"grad_norm": 0.4326270818710327,
|
| 4229 |
-
"learning_rate": 3.2008224811514735e-07,
|
| 4230 |
-
"loss": 2.0059,
|
| 4231 |
-
"step": 602
|
| 4232 |
-
},
|
| 4233 |
-
{
|
| 4234 |
-
"epoch": 1.175725079210334,
|
| 4235 |
-
"grad_norm": 0.44774889945983887,
|
| 4236 |
-
"learning_rate": 3.197395476353667e-07,
|
| 4237 |
-
"loss": 2.0427,
|
| 4238 |
-
"step": 603
|
| 4239 |
-
},
|
| 4240 |
-
{
|
| 4241 |
-
"epoch": 1.1776748720448453,
|
| 4242 |
-
"grad_norm": 0.446158766746521,
|
| 4243 |
-
"learning_rate": 3.19396847155586e-07,
|
| 4244 |
-
"loss": 2.0748,
|
| 4245 |
-
"step": 604
|
| 4246 |
-
},
|
| 4247 |
-
{
|
| 4248 |
-
"epoch": 1.1796246648793565,
|
| 4249 |
-
"grad_norm": 0.4654727280139923,
|
| 4250 |
-
"learning_rate": 3.1905414667580534e-07,
|
| 4251 |
-
"loss": 1.9297,
|
| 4252 |
-
"step": 605
|
| 4253 |
-
},
|
| 4254 |
-
{
|
| 4255 |
-
"epoch": 1.1815744577138678,
|
| 4256 |
-
"grad_norm": 0.45213672518730164,
|
| 4257 |
-
"learning_rate": 3.1871144619602463e-07,
|
| 4258 |
-
"loss": 2.1087,
|
| 4259 |
-
"step": 606
|
| 4260 |
-
},
|
| 4261 |
-
{
|
| 4262 |
-
"epoch": 1.1835242505483792,
|
| 4263 |
-
"grad_norm": 0.45228397846221924,
|
| 4264 |
-
"learning_rate": 3.18368745716244e-07,
|
| 4265 |
-
"loss": 2.0961,
|
| 4266 |
-
"step": 607
|
| 4267 |
-
},
|
| 4268 |
-
{
|
| 4269 |
-
"epoch": 1.1854740433828905,
|
| 4270 |
-
"grad_norm": 0.4470541477203369,
|
| 4271 |
-
"learning_rate": 3.180260452364633e-07,
|
| 4272 |
-
"loss": 2.0073,
|
| 4273 |
-
"step": 608
|
| 4274 |
-
},
|
| 4275 |
-
{
|
| 4276 |
-
"epoch": 1.1874238362174019,
|
| 4277 |
-
"grad_norm": 0.4324132204055786,
|
| 4278 |
-
"learning_rate": 3.176833447566826e-07,
|
| 4279 |
-
"loss": 2.0334,
|
| 4280 |
-
"step": 609
|
| 4281 |
-
},
|
| 4282 |
-
{
|
| 4283 |
-
"epoch": 1.1893736290519132,
|
| 4284 |
-
"grad_norm": 0.47044241428375244,
|
| 4285 |
-
"learning_rate": 3.1734064427690197e-07,
|
| 4286 |
-
"loss": 2.1086,
|
| 4287 |
-
"step": 610
|
| 4288 |
-
},
|
| 4289 |
-
{
|
| 4290 |
-
"epoch": 1.1913234218864246,
|
| 4291 |
-
"grad_norm": 0.43018707633018494,
|
| 4292 |
-
"learning_rate": 3.169979437971213e-07,
|
| 4293 |
-
"loss": 2.0289,
|
| 4294 |
-
"step": 611
|
| 4295 |
-
},
|
| 4296 |
-
{
|
| 4297 |
-
"epoch": 1.193273214720936,
|
| 4298 |
-
"grad_norm": 0.44133853912353516,
|
| 4299 |
-
"learning_rate": 3.166552433173406e-07,
|
| 4300 |
-
"loss": 2.0333,
|
| 4301 |
-
"step": 612
|
| 4302 |
-
},
|
| 4303 |
-
{
|
| 4304 |
-
"epoch": 1.1952230075554473,
|
| 4305 |
-
"grad_norm": 0.45557719469070435,
|
| 4306 |
-
"learning_rate": 3.1631254283755996e-07,
|
| 4307 |
-
"loss": 2.0399,
|
| 4308 |
-
"step": 613
|
| 4309 |
-
},
|
| 4310 |
-
{
|
| 4311 |
-
"epoch": 1.1971728003899587,
|
| 4312 |
-
"grad_norm": 0.4350452125072479,
|
| 4313 |
-
"learning_rate": 3.159698423577793e-07,
|
| 4314 |
-
"loss": 2.0224,
|
| 4315 |
-
"step": 614
|
| 4316 |
-
},
|
| 4317 |
-
{
|
| 4318 |
-
"epoch": 1.19912259322447,
|
| 4319 |
-
"grad_norm": 0.4687999188899994,
|
| 4320 |
-
"learning_rate": 3.156271418779986e-07,
|
| 4321 |
-
"loss": 2.0228,
|
| 4322 |
-
"step": 615
|
| 4323 |
-
},
|
| 4324 |
-
{
|
| 4325 |
-
"epoch": 1.2010723860589811,
|
| 4326 |
-
"grad_norm": 0.43684178590774536,
|
| 4327 |
-
"learning_rate": 3.1528444139821795e-07,
|
| 4328 |
-
"loss": 2.0776,
|
| 4329 |
-
"step": 616
|
| 4330 |
-
},
|
| 4331 |
-
{
|
| 4332 |
-
"epoch": 1.2030221788934925,
|
| 4333 |
-
"grad_norm": 0.45561161637306213,
|
| 4334 |
-
"learning_rate": 3.149417409184373e-07,
|
| 4335 |
-
"loss": 2.0022,
|
| 4336 |
-
"step": 617
|
| 4337 |
-
},
|
| 4338 |
-
{
|
| 4339 |
-
"epoch": 1.2049719717280039,
|
| 4340 |
-
"grad_norm": 0.4689810276031494,
|
| 4341 |
-
"learning_rate": 3.145990404386566e-07,
|
| 4342 |
-
"loss": 2.0173,
|
| 4343 |
-
"step": 618
|
| 4344 |
-
},
|
| 4345 |
-
{
|
| 4346 |
-
"epoch": 1.2069217645625152,
|
| 4347 |
-
"grad_norm": 0.4293496310710907,
|
| 4348 |
-
"learning_rate": 3.142563399588759e-07,
|
| 4349 |
-
"loss": 1.9824,
|
| 4350 |
-
"step": 619
|
| 4351 |
-
},
|
| 4352 |
-
{
|
| 4353 |
-
"epoch": 1.2088715573970266,
|
| 4354 |
-
"grad_norm": 0.4662802219390869,
|
| 4355 |
-
"learning_rate": 3.139136394790953e-07,
|
| 4356 |
-
"loss": 2.0784,
|
| 4357 |
-
"step": 620
|
| 4358 |
-
},
|
| 4359 |
-
{
|
| 4360 |
-
"epoch": 1.210821350231538,
|
| 4361 |
-
"grad_norm": 0.45310187339782715,
|
| 4362 |
-
"learning_rate": 3.135709389993146e-07,
|
| 4363 |
-
"loss": 1.9844,
|
| 4364 |
-
"step": 621
|
| 4365 |
-
},
|
| 4366 |
-
{
|
| 4367 |
-
"epoch": 1.2127711430660493,
|
| 4368 |
-
"grad_norm": 0.4419795870780945,
|
| 4369 |
-
"learning_rate": 3.1322823851953387e-07,
|
| 4370 |
-
"loss": 2.0515,
|
| 4371 |
-
"step": 622
|
| 4372 |
-
},
|
| 4373 |
-
{
|
| 4374 |
-
"epoch": 1.2147209359005606,
|
| 4375 |
-
"grad_norm": 0.4516865611076355,
|
| 4376 |
-
"learning_rate": 3.1288553803975327e-07,
|
| 4377 |
-
"loss": 2.0879,
|
| 4378 |
-
"step": 623
|
| 4379 |
-
},
|
| 4380 |
-
{
|
| 4381 |
-
"epoch": 1.216670728735072,
|
| 4382 |
-
"grad_norm": 0.46178489923477173,
|
| 4383 |
-
"learning_rate": 3.1254283755997257e-07,
|
| 4384 |
-
"loss": 2.0498,
|
| 4385 |
-
"step": 624
|
| 4386 |
-
},
|
| 4387 |
-
{
|
| 4388 |
-
"epoch": 1.2186205215695831,
|
| 4389 |
-
"grad_norm": 0.4678952097892761,
|
| 4390 |
-
"learning_rate": 3.1220013708019186e-07,
|
| 4391 |
-
"loss": 2.0408,
|
| 4392 |
-
"step": 625
|
| 4393 |
-
},
|
| 4394 |
-
{
|
| 4395 |
-
"epoch": 1.2205703144040945,
|
| 4396 |
-
"grad_norm": 0.4456236660480499,
|
| 4397 |
-
"learning_rate": 3.1185743660041126e-07,
|
| 4398 |
-
"loss": 1.9694,
|
| 4399 |
-
"step": 626
|
| 4400 |
-
},
|
| 4401 |
-
{
|
| 4402 |
-
"epoch": 1.2225201072386058,
|
| 4403 |
-
"grad_norm": 0.4397581219673157,
|
| 4404 |
-
"learning_rate": 3.1151473612063056e-07,
|
| 4405 |
-
"loss": 2.0048,
|
| 4406 |
-
"step": 627
|
| 4407 |
-
},
|
| 4408 |
-
{
|
| 4409 |
-
"epoch": 1.2244699000731172,
|
| 4410 |
-
"grad_norm": 0.4338027238845825,
|
| 4411 |
-
"learning_rate": 3.1117203564084985e-07,
|
| 4412 |
-
"loss": 2.0194,
|
| 4413 |
-
"step": 628
|
| 4414 |
-
},
|
| 4415 |
-
{
|
| 4416 |
-
"epoch": 1.2264196929076285,
|
| 4417 |
-
"grad_norm": 0.4413823187351227,
|
| 4418 |
-
"learning_rate": 3.108293351610692e-07,
|
| 4419 |
-
"loss": 2.025,
|
| 4420 |
-
"step": 629
|
| 4421 |
-
},
|
| 4422 |
-
{
|
| 4423 |
-
"epoch": 1.2283694857421399,
|
| 4424 |
-
"grad_norm": 0.43685299158096313,
|
| 4425 |
-
"learning_rate": 3.1048663468128855e-07,
|
| 4426 |
-
"loss": 2.0051,
|
| 4427 |
-
"step": 630
|
| 4428 |
-
},
|
| 4429 |
-
{
|
| 4430 |
-
"epoch": 1.2303192785766512,
|
| 4431 |
-
"grad_norm": 0.4644426107406616,
|
| 4432 |
-
"learning_rate": 3.1014393420150784e-07,
|
| 4433 |
-
"loss": 2.0313,
|
| 4434 |
-
"step": 631
|
| 4435 |
-
},
|
| 4436 |
-
{
|
| 4437 |
-
"epoch": 1.2322690714111626,
|
| 4438 |
-
"grad_norm": 0.4478755593299866,
|
| 4439 |
-
"learning_rate": 3.098012337217272e-07,
|
| 4440 |
-
"loss": 1.9669,
|
| 4441 |
-
"step": 632
|
| 4442 |
-
},
|
| 4443 |
-
{
|
| 4444 |
-
"epoch": 1.234218864245674,
|
| 4445 |
-
"grad_norm": 0.43452218174934387,
|
| 4446 |
-
"learning_rate": 3.0945853324194654e-07,
|
| 4447 |
-
"loss": 1.9927,
|
| 4448 |
-
"step": 633
|
| 4449 |
-
},
|
| 4450 |
-
{
|
| 4451 |
-
"epoch": 1.2361686570801853,
|
| 4452 |
-
"grad_norm": 0.4408141076564789,
|
| 4453 |
-
"learning_rate": 3.0911583276216583e-07,
|
| 4454 |
-
"loss": 2.136,
|
| 4455 |
-
"step": 634
|
| 4456 |
-
},
|
| 4457 |
-
{
|
| 4458 |
-
"epoch": 1.2381184499146967,
|
| 4459 |
-
"grad_norm": 0.42754924297332764,
|
| 4460 |
-
"learning_rate": 3.087731322823852e-07,
|
| 4461 |
-
"loss": 2.0247,
|
| 4462 |
-
"step": 635
|
| 4463 |
-
},
|
| 4464 |
-
{
|
| 4465 |
-
"epoch": 1.2400682427492078,
|
| 4466 |
-
"grad_norm": 0.4387798607349396,
|
| 4467 |
-
"learning_rate": 3.084304318026045e-07,
|
| 4468 |
-
"loss": 1.9643,
|
| 4469 |
-
"step": 636
|
| 4470 |
-
},
|
| 4471 |
-
{
|
| 4472 |
-
"epoch": 1.2420180355837191,
|
| 4473 |
-
"grad_norm": 0.46978920698165894,
|
| 4474 |
-
"learning_rate": 3.080877313228238e-07,
|
| 4475 |
-
"loss": 2.0776,
|
| 4476 |
-
"step": 637
|
| 4477 |
-
},
|
| 4478 |
-
{
|
| 4479 |
-
"epoch": 1.2439678284182305,
|
| 4480 |
-
"grad_norm": 0.41821563243865967,
|
| 4481 |
-
"learning_rate": 3.0774503084304317e-07,
|
| 4482 |
-
"loss": 2.0355,
|
| 4483 |
-
"step": 638
|
| 4484 |
-
},
|
| 4485 |
-
{
|
| 4486 |
-
"epoch": 1.2459176212527419,
|
| 4487 |
-
"grad_norm": 0.4664837419986725,
|
| 4488 |
-
"learning_rate": 3.074023303632625e-07,
|
| 4489 |
-
"loss": 2.0328,
|
| 4490 |
-
"step": 639
|
| 4491 |
-
},
|
| 4492 |
-
{
|
| 4493 |
-
"epoch": 1.2478674140872532,
|
| 4494 |
-
"grad_norm": 0.4467378258705139,
|
| 4495 |
-
"learning_rate": 3.070596298834818e-07,
|
| 4496 |
-
"loss": 2.0058,
|
| 4497 |
-
"step": 640
|
| 4498 |
-
},
|
| 4499 |
-
{
|
| 4500 |
-
"epoch": 1.2498172069217646,
|
| 4501 |
-
"grad_norm": 0.442058265209198,
|
| 4502 |
-
"learning_rate": 3.0671692940370116e-07,
|
| 4503 |
-
"loss": 2.0565,
|
| 4504 |
-
"step": 641
|
| 4505 |
-
},
|
| 4506 |
-
{
|
| 4507 |
-
"epoch": 1.251766999756276,
|
| 4508 |
-
"grad_norm": 0.4655166268348694,
|
| 4509 |
-
"learning_rate": 3.0637422892392045e-07,
|
| 4510 |
-
"loss": 2.0628,
|
| 4511 |
-
"step": 642
|
| 4512 |
-
},
|
| 4513 |
-
{
|
| 4514 |
-
"epoch": 1.2537167925907873,
|
| 4515 |
-
"grad_norm": 0.4388466477394104,
|
| 4516 |
-
"learning_rate": 3.060315284441398e-07,
|
| 4517 |
-
"loss": 2.0716,
|
| 4518 |
-
"step": 643
|
| 4519 |
-
},
|
| 4520 |
-
{
|
| 4521 |
-
"epoch": 1.2556665854252986,
|
| 4522 |
-
"grad_norm": 0.48705416917800903,
|
| 4523 |
-
"learning_rate": 3.0568882796435915e-07,
|
| 4524 |
-
"loss": 1.9872,
|
| 4525 |
-
"step": 644
|
| 4526 |
-
},
|
| 4527 |
-
{
|
| 4528 |
-
"epoch": 1.2576163782598098,
|
| 4529 |
-
"grad_norm": 0.4618842899799347,
|
| 4530 |
-
"learning_rate": 3.0534612748457844e-07,
|
| 4531 |
-
"loss": 2.0306,
|
| 4532 |
-
"step": 645
|
| 4533 |
-
},
|
| 4534 |
-
{
|
| 4535 |
-
"epoch": 1.2595661710943213,
|
| 4536 |
-
"grad_norm": 0.46533843874931335,
|
| 4537 |
-
"learning_rate": 3.050034270047978e-07,
|
| 4538 |
-
"loss": 2.0827,
|
| 4539 |
-
"step": 646
|
| 4540 |
-
},
|
| 4541 |
-
{
|
| 4542 |
-
"epoch": 1.2615159639288325,
|
| 4543 |
-
"grad_norm": 0.4898700714111328,
|
| 4544 |
-
"learning_rate": 3.0466072652501714e-07,
|
| 4545 |
-
"loss": 1.9585,
|
| 4546 |
-
"step": 647
|
| 4547 |
-
},
|
| 4548 |
-
{
|
| 4549 |
-
"epoch": 1.2634657567633438,
|
| 4550 |
-
"grad_norm": 0.4561532735824585,
|
| 4551 |
-
"learning_rate": 3.0431802604523643e-07,
|
| 4552 |
-
"loss": 2.0689,
|
| 4553 |
-
"step": 648
|
| 4554 |
-
},
|
| 4555 |
-
{
|
| 4556 |
-
"epoch": 1.2654155495978552,
|
| 4557 |
-
"grad_norm": 0.4628736078739166,
|
| 4558 |
-
"learning_rate": 3.039753255654558e-07,
|
| 4559 |
-
"loss": 2.0307,
|
| 4560 |
-
"step": 649
|
| 4561 |
-
},
|
| 4562 |
-
{
|
| 4563 |
-
"epoch": 1.2673653424323665,
|
| 4564 |
-
"grad_norm": 0.4475798010826111,
|
| 4565 |
-
"learning_rate": 3.036326250856751e-07,
|
| 4566 |
-
"loss": 2.0372,
|
| 4567 |
-
"step": 650
|
| 4568 |
-
},
|
| 4569 |
-
{
|
| 4570 |
-
"epoch": 1.269315135266878,
|
| 4571 |
-
"grad_norm": 0.44448035955429077,
|
| 4572 |
-
"learning_rate": 3.032899246058944e-07,
|
| 4573 |
-
"loss": 2.0334,
|
| 4574 |
-
"step": 651
|
| 4575 |
-
},
|
| 4576 |
-
{
|
| 4577 |
-
"epoch": 1.2712649281013892,
|
| 4578 |
-
"grad_norm": 0.4554859697818756,
|
| 4579 |
-
"learning_rate": 3.0294722412611377e-07,
|
| 4580 |
-
"loss": 2.0487,
|
| 4581 |
-
"step": 652
|
| 4582 |
-
},
|
| 4583 |
-
{
|
| 4584 |
-
"epoch": 1.2732147209359006,
|
| 4585 |
-
"grad_norm": 0.44150403141975403,
|
| 4586 |
-
"learning_rate": 3.026045236463331e-07,
|
| 4587 |
-
"loss": 2.085,
|
| 4588 |
-
"step": 653
|
| 4589 |
-
},
|
| 4590 |
-
{
|
| 4591 |
-
"epoch": 1.275164513770412,
|
| 4592 |
-
"grad_norm": 0.4476960301399231,
|
| 4593 |
-
"learning_rate": 3.022618231665524e-07,
|
| 4594 |
-
"loss": 1.9762,
|
| 4595 |
-
"step": 654
|
| 4596 |
-
},
|
| 4597 |
-
{
|
| 4598 |
-
"epoch": 1.2771143066049233,
|
| 4599 |
-
"grad_norm": 0.4773290753364563,
|
| 4600 |
-
"learning_rate": 3.019191226867717e-07,
|
| 4601 |
-
"loss": 2.0565,
|
| 4602 |
-
"step": 655
|
| 4603 |
-
},
|
| 4604 |
-
{
|
| 4605 |
-
"epoch": 1.2790640994394344,
|
| 4606 |
-
"grad_norm": 0.43788987398147583,
|
| 4607 |
-
"learning_rate": 3.015764222069911e-07,
|
| 4608 |
-
"loss": 2.0629,
|
| 4609 |
-
"step": 656
|
| 4610 |
-
},
|
| 4611 |
-
{
|
| 4612 |
-
"epoch": 1.281013892273946,
|
| 4613 |
-
"grad_norm": 0.4314157962799072,
|
| 4614 |
-
"learning_rate": 3.012337217272104e-07,
|
| 4615 |
-
"loss": 2.0554,
|
| 4616 |
-
"step": 657
|
| 4617 |
-
},
|
| 4618 |
-
{
|
| 4619 |
-
"epoch": 1.2829636851084572,
|
| 4620 |
-
"grad_norm": 0.45381680130958557,
|
| 4621 |
-
"learning_rate": 3.008910212474297e-07,
|
| 4622 |
-
"loss": 2.0514,
|
| 4623 |
-
"step": 658
|
| 4624 |
-
},
|
| 4625 |
-
{
|
| 4626 |
-
"epoch": 1.2849134779429685,
|
| 4627 |
-
"grad_norm": 0.47213441133499146,
|
| 4628 |
-
"learning_rate": 3.005483207676491e-07,
|
| 4629 |
-
"loss": 2.0267,
|
| 4630 |
-
"step": 659
|
| 4631 |
-
},
|
| 4632 |
-
{
|
| 4633 |
-
"epoch": 1.2868632707774799,
|
| 4634 |
-
"grad_norm": 0.4460486173629761,
|
| 4635 |
-
"learning_rate": 3.002056202878684e-07,
|
| 4636 |
-
"loss": 2.0717,
|
| 4637 |
-
"step": 660
|
| 4638 |
-
},
|
| 4639 |
-
{
|
| 4640 |
-
"epoch": 1.2888130636119912,
|
| 4641 |
-
"grad_norm": 0.452747642993927,
|
| 4642 |
-
"learning_rate": 2.998629198080877e-07,
|
| 4643 |
-
"loss": 2.0634,
|
| 4644 |
-
"step": 661
|
| 4645 |
-
},
|
| 4646 |
-
{
|
| 4647 |
-
"epoch": 1.2907628564465026,
|
| 4648 |
-
"grad_norm": 0.4495120942592621,
|
| 4649 |
-
"learning_rate": 2.995202193283071e-07,
|
| 4650 |
-
"loss": 2.042,
|
| 4651 |
-
"step": 662
|
| 4652 |
-
},
|
| 4653 |
-
{
|
| 4654 |
-
"epoch": 1.292712649281014,
|
| 4655 |
-
"grad_norm": 0.433224081993103,
|
| 4656 |
-
"learning_rate": 2.991775188485264e-07,
|
| 4657 |
-
"loss": 2.0565,
|
| 4658 |
-
"step": 663
|
| 4659 |
-
},
|
| 4660 |
-
{
|
| 4661 |
-
"epoch": 1.2946624421155253,
|
| 4662 |
-
"grad_norm": 0.4596520960330963,
|
| 4663 |
-
"learning_rate": 2.9883481836874567e-07,
|
| 4664 |
-
"loss": 2.0272,
|
| 4665 |
-
"step": 664
|
| 4666 |
-
},
|
| 4667 |
-
{
|
| 4668 |
-
"epoch": 1.2966122349500366,
|
| 4669 |
-
"grad_norm": 0.433887243270874,
|
| 4670 |
-
"learning_rate": 2.9849211788896507e-07,
|
| 4671 |
-
"loss": 1.965,
|
| 4672 |
-
"step": 665
|
| 4673 |
-
},
|
| 4674 |
-
{
|
| 4675 |
-
"epoch": 1.298562027784548,
|
| 4676 |
-
"grad_norm": 0.44755810499191284,
|
| 4677 |
-
"learning_rate": 2.9814941740918437e-07,
|
| 4678 |
-
"loss": 1.9915,
|
| 4679 |
-
"step": 666
|
| 4680 |
-
},
|
| 4681 |
-
{
|
| 4682 |
-
"epoch": 1.3005118206190591,
|
| 4683 |
-
"grad_norm": 0.48203861713409424,
|
| 4684 |
-
"learning_rate": 2.9780671692940366e-07,
|
| 4685 |
-
"loss": 2.0296,
|
| 4686 |
-
"step": 667
|
| 4687 |
-
},
|
| 4688 |
-
{
|
| 4689 |
-
"epoch": 1.3024616134535705,
|
| 4690 |
-
"grad_norm": 0.4314959943294525,
|
| 4691 |
-
"learning_rate": 2.97464016449623e-07,
|
| 4692 |
-
"loss": 2.0282,
|
| 4693 |
-
"step": 668
|
| 4694 |
-
},
|
| 4695 |
-
{
|
| 4696 |
-
"epoch": 1.3044114062880818,
|
| 4697 |
-
"grad_norm": 0.4476211369037628,
|
| 4698 |
-
"learning_rate": 2.9712131596984236e-07,
|
| 4699 |
-
"loss": 2.0348,
|
| 4700 |
-
"step": 669
|
| 4701 |
-
},
|
| 4702 |
-
{
|
| 4703 |
-
"epoch": 1.3063611991225932,
|
| 4704 |
-
"grad_norm": 0.45356854796409607,
|
| 4705 |
-
"learning_rate": 2.9677861549006165e-07,
|
| 4706 |
-
"loss": 2.0369,
|
| 4707 |
-
"step": 670
|
| 4708 |
-
},
|
| 4709 |
-
{
|
| 4710 |
-
"epoch": 1.3083109919571045,
|
| 4711 |
-
"grad_norm": 0.4637032747268677,
|
| 4712 |
-
"learning_rate": 2.96435915010281e-07,
|
| 4713 |
-
"loss": 2.1002,
|
| 4714 |
-
"step": 671
|
| 4715 |
-
},
|
| 4716 |
-
{
|
| 4717 |
-
"epoch": 1.310260784791616,
|
| 4718 |
-
"grad_norm": 0.4258365333080292,
|
| 4719 |
-
"learning_rate": 2.9609321453050035e-07,
|
| 4720 |
-
"loss": 2.0184,
|
| 4721 |
-
"step": 672
|
| 4722 |
-
},
|
| 4723 |
-
{
|
| 4724 |
-
"epoch": 1.3122105776261273,
|
| 4725 |
-
"grad_norm": 0.4571716785430908,
|
| 4726 |
-
"learning_rate": 2.9575051405071964e-07,
|
| 4727 |
-
"loss": 2.0711,
|
| 4728 |
-
"step": 673
|
| 4729 |
-
},
|
| 4730 |
-
{
|
| 4731 |
-
"epoch": 1.3141603704606386,
|
| 4732 |
-
"grad_norm": 0.4479144215583801,
|
| 4733 |
-
"learning_rate": 2.95407813570939e-07,
|
| 4734 |
-
"loss": 2.1037,
|
| 4735 |
-
"step": 674
|
| 4736 |
-
},
|
| 4737 |
-
{
|
| 4738 |
-
"epoch": 1.31611016329515,
|
| 4739 |
-
"grad_norm": 0.463773638010025,
|
| 4740 |
-
"learning_rate": 2.9506511309115834e-07,
|
| 4741 |
-
"loss": 2.087,
|
| 4742 |
-
"step": 675
|
| 4743 |
-
},
|
| 4744 |
-
{
|
| 4745 |
-
"epoch": 1.318059956129661,
|
| 4746 |
-
"grad_norm": 0.4595959782600403,
|
| 4747 |
-
"learning_rate": 2.9472241261137763e-07,
|
| 4748 |
-
"loss": 2.0246,
|
| 4749 |
-
"step": 676
|
| 4750 |
-
},
|
| 4751 |
-
{
|
| 4752 |
-
"epoch": 1.3200097489641727,
|
| 4753 |
-
"grad_norm": 0.41977226734161377,
|
| 4754 |
-
"learning_rate": 2.94379712131597e-07,
|
| 4755 |
-
"loss": 2.0132,
|
| 4756 |
-
"step": 677
|
| 4757 |
-
},
|
| 4758 |
-
{
|
| 4759 |
-
"epoch": 1.3219595417986838,
|
| 4760 |
-
"grad_norm": 0.4429217576980591,
|
| 4761 |
-
"learning_rate": 2.940370116518163e-07,
|
| 4762 |
-
"loss": 2.0414,
|
| 4763 |
-
"step": 678
|
| 4764 |
-
},
|
| 4765 |
-
{
|
| 4766 |
-
"epoch": 1.3239093346331952,
|
| 4767 |
-
"grad_norm": 0.46036285161972046,
|
| 4768 |
-
"learning_rate": 2.936943111720356e-07,
|
| 4769 |
-
"loss": 2.0474,
|
| 4770 |
-
"step": 679
|
| 4771 |
-
},
|
| 4772 |
-
{
|
| 4773 |
-
"epoch": 1.3258591274677065,
|
| 4774 |
-
"grad_norm": 0.4518478512763977,
|
| 4775 |
-
"learning_rate": 2.9335161069225497e-07,
|
| 4776 |
-
"loss": 1.991,
|
| 4777 |
-
"step": 680
|
| 4778 |
-
},
|
| 4779 |
-
{
|
| 4780 |
-
"epoch": 1.3278089203022179,
|
| 4781 |
-
"grad_norm": 0.4507528841495514,
|
| 4782 |
-
"learning_rate": 2.9300891021247426e-07,
|
| 4783 |
-
"loss": 2.0038,
|
| 4784 |
-
"step": 681
|
| 4785 |
-
},
|
| 4786 |
-
{
|
| 4787 |
-
"epoch": 1.3297587131367292,
|
| 4788 |
-
"grad_norm": 0.45446595549583435,
|
| 4789 |
-
"learning_rate": 2.926662097326936e-07,
|
| 4790 |
-
"loss": 1.9257,
|
| 4791 |
-
"step": 682
|
| 4792 |
-
},
|
| 4793 |
-
{
|
| 4794 |
-
"epoch": 1.3317085059712406,
|
| 4795 |
-
"grad_norm": 0.45073091983795166,
|
| 4796 |
-
"learning_rate": 2.9232350925291296e-07,
|
| 4797 |
-
"loss": 2.0667,
|
| 4798 |
-
"step": 683
|
| 4799 |
-
},
|
| 4800 |
-
{
|
| 4801 |
-
"epoch": 1.333658298805752,
|
| 4802 |
-
"grad_norm": 0.43848779797554016,
|
| 4803 |
-
"learning_rate": 2.9198080877313225e-07,
|
| 4804 |
-
"loss": 2.0127,
|
| 4805 |
-
"step": 684
|
| 4806 |
-
},
|
| 4807 |
-
{
|
| 4808 |
-
"epoch": 1.3356080916402633,
|
| 4809 |
-
"grad_norm": 0.44587504863739014,
|
| 4810 |
-
"learning_rate": 2.916381082933516e-07,
|
| 4811 |
-
"loss": 2.0694,
|
| 4812 |
-
"step": 685
|
| 4813 |
-
},
|
| 4814 |
-
{
|
| 4815 |
-
"epoch": 1.3375578844747746,
|
| 4816 |
-
"grad_norm": 0.46157652139663696,
|
| 4817 |
-
"learning_rate": 2.9129540781357095e-07,
|
| 4818 |
-
"loss": 2.112,
|
| 4819 |
-
"step": 686
|
| 4820 |
-
},
|
| 4821 |
-
{
|
| 4822 |
-
"epoch": 1.3395076773092858,
|
| 4823 |
-
"grad_norm": 0.461897075176239,
|
| 4824 |
-
"learning_rate": 2.9095270733379024e-07,
|
| 4825 |
-
"loss": 2.0431,
|
| 4826 |
-
"step": 687
|
| 4827 |
-
},
|
| 4828 |
-
{
|
| 4829 |
-
"epoch": 1.3414574701437973,
|
| 4830 |
-
"grad_norm": 0.42506590485572815,
|
| 4831 |
-
"learning_rate": 2.906100068540096e-07,
|
| 4832 |
-
"loss": 2.0612,
|
| 4833 |
-
"step": 688
|
| 4834 |
-
},
|
| 4835 |
-
{
|
| 4836 |
-
"epoch": 1.3434072629783085,
|
| 4837 |
-
"grad_norm": 0.43368127942085266,
|
| 4838 |
-
"learning_rate": 2.9026730637422894e-07,
|
| 4839 |
-
"loss": 2.0253,
|
| 4840 |
-
"step": 689
|
| 4841 |
-
},
|
| 4842 |
-
{
|
| 4843 |
-
"epoch": 1.3453570558128198,
|
| 4844 |
-
"grad_norm": 0.4484082758426666,
|
| 4845 |
-
"learning_rate": 2.8992460589444823e-07,
|
| 4846 |
-
"loss": 1.9962,
|
| 4847 |
-
"step": 690
|
| 4848 |
-
},
|
| 4849 |
-
{
|
| 4850 |
-
"epoch": 1.3473068486473312,
|
| 4851 |
-
"grad_norm": 0.44570791721343994,
|
| 4852 |
-
"learning_rate": 2.895819054146676e-07,
|
| 4853 |
-
"loss": 2.018,
|
| 4854 |
-
"step": 691
|
| 4855 |
-
},
|
| 4856 |
-
{
|
| 4857 |
-
"epoch": 1.3492566414818425,
|
| 4858 |
-
"grad_norm": 0.4472144842147827,
|
| 4859 |
-
"learning_rate": 2.892392049348869e-07,
|
| 4860 |
-
"loss": 2.0254,
|
| 4861 |
-
"step": 692
|
| 4862 |
-
},
|
| 4863 |
-
{
|
| 4864 |
-
"epoch": 1.351206434316354,
|
| 4865 |
-
"grad_norm": 0.4680030047893524,
|
| 4866 |
-
"learning_rate": 2.888965044551062e-07,
|
| 4867 |
-
"loss": 2.1265,
|
| 4868 |
-
"step": 693
|
| 4869 |
-
},
|
| 4870 |
-
{
|
| 4871 |
-
"epoch": 1.3531562271508653,
|
| 4872 |
-
"grad_norm": 0.44323253631591797,
|
| 4873 |
-
"learning_rate": 2.885538039753255e-07,
|
| 4874 |
-
"loss": 2.0222,
|
| 4875 |
-
"step": 694
|
| 4876 |
-
},
|
| 4877 |
-
{
|
| 4878 |
-
"epoch": 1.3551060199853766,
|
| 4879 |
-
"grad_norm": 0.4732964038848877,
|
| 4880 |
-
"learning_rate": 2.882111034955449e-07,
|
| 4881 |
-
"loss": 2.0219,
|
| 4882 |
-
"step": 695
|
| 4883 |
-
},
|
| 4884 |
-
{
|
| 4885 |
-
"epoch": 1.357055812819888,
|
| 4886 |
-
"grad_norm": 0.4392209053039551,
|
| 4887 |
-
"learning_rate": 2.878684030157642e-07,
|
| 4888 |
-
"loss": 1.9841,
|
| 4889 |
-
"step": 696
|
| 4890 |
-
},
|
| 4891 |
-
{
|
| 4892 |
-
"epoch": 1.3590056056543993,
|
| 4893 |
-
"grad_norm": 0.46177539229393005,
|
| 4894 |
-
"learning_rate": 2.875257025359835e-07,
|
| 4895 |
-
"loss": 2.0461,
|
| 4896 |
-
"step": 697
|
| 4897 |
-
},
|
| 4898 |
-
{
|
| 4899 |
-
"epoch": 1.3609553984889105,
|
| 4900 |
-
"grad_norm": 0.4625999927520752,
|
| 4901 |
-
"learning_rate": 2.871830020562029e-07,
|
| 4902 |
-
"loss": 2.0137,
|
| 4903 |
-
"step": 698
|
| 4904 |
-
},
|
| 4905 |
-
{
|
| 4906 |
-
"epoch": 1.3629051913234218,
|
| 4907 |
-
"grad_norm": 0.43552806973457336,
|
| 4908 |
-
"learning_rate": 2.868403015764222e-07,
|
| 4909 |
-
"loss": 2.0408,
|
| 4910 |
-
"step": 699
|
| 4911 |
-
},
|
| 4912 |
-
{
|
| 4913 |
-
"epoch": 1.3648549841579332,
|
| 4914 |
-
"grad_norm": 0.47674480080604553,
|
| 4915 |
-
"learning_rate": 2.864976010966415e-07,
|
| 4916 |
-
"loss": 2.0021,
|
| 4917 |
-
"step": 700
|
| 4918 |
-
},
|
| 4919 |
-
{
|
| 4920 |
-
"epoch": 1.3668047769924445,
|
| 4921 |
-
"grad_norm": 0.46479421854019165,
|
| 4922 |
-
"learning_rate": 2.861549006168609e-07,
|
| 4923 |
-
"loss": 1.9898,
|
| 4924 |
-
"step": 701
|
| 4925 |
-
},
|
| 4926 |
-
{
|
| 4927 |
-
"epoch": 1.3687545698269559,
|
| 4928 |
-
"grad_norm": 0.4399622976779938,
|
| 4929 |
-
"learning_rate": 2.858122001370802e-07,
|
| 4930 |
-
"loss": 1.9638,
|
| 4931 |
-
"step": 702
|
| 4932 |
-
},
|
| 4933 |
-
{
|
| 4934 |
-
"epoch": 1.3707043626614672,
|
| 4935 |
-
"grad_norm": 0.442557692527771,
|
| 4936 |
-
"learning_rate": 2.854694996572995e-07,
|
| 4937 |
-
"loss": 2.0099,
|
| 4938 |
-
"step": 703
|
| 4939 |
-
},
|
| 4940 |
-
{
|
| 4941 |
-
"epoch": 1.3726541554959786,
|
| 4942 |
-
"grad_norm": 0.4601743817329407,
|
| 4943 |
-
"learning_rate": 2.851267991775189e-07,
|
| 4944 |
-
"loss": 2.057,
|
| 4945 |
-
"step": 704
|
| 4946 |
-
},
|
| 4947 |
-
{
|
| 4948 |
-
"epoch": 1.37460394833049,
|
| 4949 |
-
"grad_norm": 0.4959220290184021,
|
| 4950 |
-
"learning_rate": 2.847840986977382e-07,
|
| 4951 |
-
"loss": 2.092,
|
| 4952 |
-
"step": 705
|
| 4953 |
-
},
|
| 4954 |
-
{
|
| 4955 |
-
"epoch": 1.3765537411650013,
|
| 4956 |
-
"grad_norm": 0.40172404050827026,
|
| 4957 |
-
"learning_rate": 2.8444139821795747e-07,
|
| 4958 |
-
"loss": 2.0074,
|
| 4959 |
-
"step": 706
|
| 4960 |
-
},
|
| 4961 |
-
{
|
| 4962 |
-
"epoch": 1.3785035339995124,
|
| 4963 |
-
"grad_norm": 0.4572814404964447,
|
| 4964 |
-
"learning_rate": 2.840986977381768e-07,
|
| 4965 |
-
"loss": 1.9777,
|
| 4966 |
-
"step": 707
|
| 4967 |
-
},
|
| 4968 |
-
{
|
| 4969 |
-
"epoch": 1.380453326834024,
|
| 4970 |
-
"grad_norm": 0.4464624524116516,
|
| 4971 |
-
"learning_rate": 2.8375599725839617e-07,
|
| 4972 |
-
"loss": 2.0183,
|
| 4973 |
-
"step": 708
|
| 4974 |
-
},
|
| 4975 |
-
{
|
| 4976 |
-
"epoch": 1.3824031196685351,
|
| 4977 |
-
"grad_norm": 0.4498922526836395,
|
| 4978 |
-
"learning_rate": 2.8341329677861546e-07,
|
| 4979 |
-
"loss": 2.0975,
|
| 4980 |
-
"step": 709
|
| 4981 |
-
},
|
| 4982 |
-
{
|
| 4983 |
-
"epoch": 1.3843529125030465,
|
| 4984 |
-
"grad_norm": 0.4430985748767853,
|
| 4985 |
-
"learning_rate": 2.830705962988348e-07,
|
| 4986 |
-
"loss": 2.027,
|
| 4987 |
-
"step": 710
|
| 4988 |
-
},
|
| 4989 |
-
{
|
| 4990 |
-
"epoch": 1.3863027053375578,
|
| 4991 |
-
"grad_norm": 0.4422641694545746,
|
| 4992 |
-
"learning_rate": 2.8272789581905416e-07,
|
| 4993 |
-
"loss": 2.0625,
|
| 4994 |
-
"step": 711
|
| 4995 |
-
},
|
| 4996 |
-
{
|
| 4997 |
-
"epoch": 1.3882524981720692,
|
| 4998 |
-
"grad_norm": 0.46121206879615784,
|
| 4999 |
-
"learning_rate": 2.8238519533927345e-07,
|
| 5000 |
-
"loss": 2.0135,
|
| 5001 |
-
"step": 712
|
| 5002 |
-
},
|
| 5003 |
-
{
|
| 5004 |
-
"epoch": 1.3902022910065805,
|
| 5005 |
-
"grad_norm": 0.4685353934764862,
|
| 5006 |
-
"learning_rate": 2.820424948594928e-07,
|
| 5007 |
-
"loss": 2.071,
|
| 5008 |
-
"step": 713
|
| 5009 |
-
},
|
| 5010 |
-
{
|
| 5011 |
-
"epoch": 1.392152083841092,
|
| 5012 |
-
"grad_norm": 0.43733134865760803,
|
| 5013 |
-
"learning_rate": 2.8169979437971215e-07,
|
| 5014 |
-
"loss": 2.0531,
|
| 5015 |
-
"step": 714
|
| 5016 |
-
},
|
| 5017 |
-
{
|
| 5018 |
-
"epoch": 1.3941018766756033,
|
| 5019 |
-
"grad_norm": 0.4479463994503021,
|
| 5020 |
-
"learning_rate": 2.8135709389993144e-07,
|
| 5021 |
-
"loss": 2.0192,
|
| 5022 |
-
"step": 715
|
| 5023 |
-
},
|
| 5024 |
-
{
|
| 5025 |
-
"epoch": 1.3960516695101146,
|
| 5026 |
-
"grad_norm": 0.4477840065956116,
|
| 5027 |
-
"learning_rate": 2.810143934201508e-07,
|
| 5028 |
-
"loss": 2.0408,
|
| 5029 |
-
"step": 716
|
| 5030 |
-
},
|
| 5031 |
-
{
|
| 5032 |
-
"epoch": 1.398001462344626,
|
| 5033 |
-
"grad_norm": 0.44232964515686035,
|
| 5034 |
-
"learning_rate": 2.8067169294037014e-07,
|
| 5035 |
-
"loss": 2.0992,
|
| 5036 |
-
"step": 717
|
| 5037 |
-
},
|
| 5038 |
-
{
|
| 5039 |
-
"epoch": 1.399951255179137,
|
| 5040 |
-
"grad_norm": 0.4573095142841339,
|
| 5041 |
-
"learning_rate": 2.8032899246058943e-07,
|
| 5042 |
-
"loss": 1.9958,
|
| 5043 |
-
"step": 718
|
| 5044 |
-
},
|
| 5045 |
-
{
|
| 5046 |
-
"epoch": 1.4019010480136487,
|
| 5047 |
-
"grad_norm": 0.4734794497489929,
|
| 5048 |
-
"learning_rate": 2.799862919808088e-07,
|
| 5049 |
-
"loss": 2.0268,
|
| 5050 |
-
"step": 719
|
| 5051 |
-
},
|
| 5052 |
-
{
|
| 5053 |
-
"epoch": 1.4038508408481598,
|
| 5054 |
-
"grad_norm": 0.4753987193107605,
|
| 5055 |
-
"learning_rate": 2.7964359150102807e-07,
|
| 5056 |
-
"loss": 2.0436,
|
| 5057 |
-
"step": 720
|
| 5058 |
-
},
|
| 5059 |
-
{
|
| 5060 |
-
"epoch": 1.4058006336826712,
|
| 5061 |
-
"grad_norm": 0.4515923261642456,
|
| 5062 |
-
"learning_rate": 2.793008910212474e-07,
|
| 5063 |
-
"loss": 2.0018,
|
| 5064 |
-
"step": 721
|
| 5065 |
-
},
|
| 5066 |
-
{
|
| 5067 |
-
"epoch": 1.4077504265171825,
|
| 5068 |
-
"grad_norm": 0.45925289392471313,
|
| 5069 |
-
"learning_rate": 2.7895819054146677e-07,
|
| 5070 |
-
"loss": 2.0454,
|
| 5071 |
-
"step": 722
|
| 5072 |
-
},
|
| 5073 |
-
{
|
| 5074 |
-
"epoch": 1.4097002193516939,
|
| 5075 |
-
"grad_norm": 0.4684261083602905,
|
| 5076 |
-
"learning_rate": 2.7861549006168606e-07,
|
| 5077 |
-
"loss": 2.0355,
|
| 5078 |
-
"step": 723
|
| 5079 |
-
},
|
| 5080 |
-
{
|
| 5081 |
-
"epoch": 1.4116500121862052,
|
| 5082 |
-
"grad_norm": 0.4723130464553833,
|
| 5083 |
-
"learning_rate": 2.782727895819054e-07,
|
| 5084 |
-
"loss": 2.0189,
|
| 5085 |
-
"step": 724
|
| 5086 |
-
},
|
| 5087 |
-
{
|
| 5088 |
-
"epoch": 1.4135998050207166,
|
| 5089 |
-
"grad_norm": 0.43946054577827454,
|
| 5090 |
-
"learning_rate": 2.7793008910212476e-07,
|
| 5091 |
-
"loss": 2.0165,
|
| 5092 |
-
"step": 725
|
| 5093 |
-
},
|
| 5094 |
-
{
|
| 5095 |
-
"epoch": 1.415549597855228,
|
| 5096 |
-
"grad_norm": 0.45172879099845886,
|
| 5097 |
-
"learning_rate": 2.7758738862234405e-07,
|
| 5098 |
-
"loss": 1.9966,
|
| 5099 |
-
"step": 726
|
| 5100 |
-
},
|
| 5101 |
-
{
|
| 5102 |
-
"epoch": 1.4174993906897393,
|
| 5103 |
-
"grad_norm": 0.4361145496368408,
|
| 5104 |
-
"learning_rate": 2.772446881425634e-07,
|
| 5105 |
-
"loss": 1.982,
|
| 5106 |
-
"step": 727
|
| 5107 |
-
},
|
| 5108 |
-
{
|
| 5109 |
-
"epoch": 1.4194491835242506,
|
| 5110 |
-
"grad_norm": 0.4422454237937927,
|
| 5111 |
-
"learning_rate": 2.7690198766278275e-07,
|
| 5112 |
-
"loss": 2.0032,
|
| 5113 |
-
"step": 728
|
| 5114 |
-
},
|
| 5115 |
-
{
|
| 5116 |
-
"epoch": 1.4213989763587618,
|
| 5117 |
-
"grad_norm": 0.4438495934009552,
|
| 5118 |
-
"learning_rate": 2.7655928718300204e-07,
|
| 5119 |
-
"loss": 2.0198,
|
| 5120 |
-
"step": 729
|
| 5121 |
-
},
|
| 5122 |
-
{
|
| 5123 |
-
"epoch": 1.4233487691932731,
|
| 5124 |
-
"grad_norm": 0.4422749876976013,
|
| 5125 |
-
"learning_rate": 2.762165867032214e-07,
|
| 5126 |
-
"loss": 1.992,
|
| 5127 |
-
"step": 730
|
| 5128 |
-
},
|
| 5129 |
-
{
|
| 5130 |
-
"epoch": 1.4252985620277845,
|
| 5131 |
-
"grad_norm": 0.4652174115180969,
|
| 5132 |
-
"learning_rate": 2.7587388622344074e-07,
|
| 5133 |
-
"loss": 2.0345,
|
| 5134 |
-
"step": 731
|
| 5135 |
-
},
|
| 5136 |
-
{
|
| 5137 |
-
"epoch": 1.4272483548622958,
|
| 5138 |
-
"grad_norm": 0.46277597546577454,
|
| 5139 |
-
"learning_rate": 2.7553118574366003e-07,
|
| 5140 |
-
"loss": 2.0406,
|
| 5141 |
-
"step": 732
|
| 5142 |
-
},
|
| 5143 |
-
{
|
| 5144 |
-
"epoch": 1.4291981476968072,
|
| 5145 |
-
"grad_norm": 0.45579442381858826,
|
| 5146 |
-
"learning_rate": 2.751884852638793e-07,
|
| 5147 |
-
"loss": 2.0671,
|
| 5148 |
-
"step": 733
|
| 5149 |
-
},
|
| 5150 |
-
{
|
| 5151 |
-
"epoch": 1.4311479405313186,
|
| 5152 |
-
"grad_norm": 0.43527230620384216,
|
| 5153 |
-
"learning_rate": 2.748457847840987e-07,
|
| 5154 |
-
"loss": 2.0433,
|
| 5155 |
-
"step": 734
|
| 5156 |
-
},
|
| 5157 |
-
{
|
| 5158 |
-
"epoch": 1.43309773336583,
|
| 5159 |
-
"grad_norm": 0.4699551463127136,
|
| 5160 |
-
"learning_rate": 2.74503084304318e-07,
|
| 5161 |
-
"loss": 2.0366,
|
| 5162 |
-
"step": 735
|
| 5163 |
-
},
|
| 5164 |
-
{
|
| 5165 |
-
"epoch": 1.4350475262003413,
|
| 5166 |
-
"grad_norm": 0.4446089565753937,
|
| 5167 |
-
"learning_rate": 2.741603838245373e-07,
|
| 5168 |
-
"loss": 1.9986,
|
| 5169 |
-
"step": 736
|
| 5170 |
-
},
|
| 5171 |
-
{
|
| 5172 |
-
"epoch": 1.4369973190348526,
|
| 5173 |
-
"grad_norm": 0.4645906686782837,
|
| 5174 |
-
"learning_rate": 2.738176833447567e-07,
|
| 5175 |
-
"loss": 2.1331,
|
| 5176 |
-
"step": 737
|
| 5177 |
-
},
|
| 5178 |
-
{
|
| 5179 |
-
"epoch": 1.4389471118693637,
|
| 5180 |
-
"grad_norm": 0.46871501207351685,
|
| 5181 |
-
"learning_rate": 2.73474982864976e-07,
|
| 5182 |
-
"loss": 2.0402,
|
| 5183 |
-
"step": 738
|
| 5184 |
-
},
|
| 5185 |
-
{
|
| 5186 |
-
"epoch": 1.4408969047038753,
|
| 5187 |
-
"grad_norm": 0.4507101774215698,
|
| 5188 |
-
"learning_rate": 2.731322823851953e-07,
|
| 5189 |
-
"loss": 2.0027,
|
| 5190 |
-
"step": 739
|
| 5191 |
-
},
|
| 5192 |
-
{
|
| 5193 |
-
"epoch": 1.4428466975383865,
|
| 5194 |
-
"grad_norm": 0.4642309546470642,
|
| 5195 |
-
"learning_rate": 2.727895819054147e-07,
|
| 5196 |
-
"loss": 2.0613,
|
| 5197 |
-
"step": 740
|
| 5198 |
-
},
|
| 5199 |
-
{
|
| 5200 |
-
"epoch": 1.4447964903728978,
|
| 5201 |
-
"grad_norm": 0.4762292206287384,
|
| 5202 |
-
"learning_rate": 2.72446881425634e-07,
|
| 5203 |
-
"loss": 2.0315,
|
| 5204 |
-
"step": 741
|
| 5205 |
-
},
|
| 5206 |
-
{
|
| 5207 |
-
"epoch": 1.4467462832074092,
|
| 5208 |
-
"grad_norm": 0.4549463391304016,
|
| 5209 |
-
"learning_rate": 2.721041809458533e-07,
|
| 5210 |
-
"loss": 2.0492,
|
| 5211 |
-
"step": 742
|
| 5212 |
-
},
|
| 5213 |
-
{
|
| 5214 |
-
"epoch": 1.4486960760419205,
|
| 5215 |
-
"grad_norm": 0.4566596448421478,
|
| 5216 |
-
"learning_rate": 2.717614804660727e-07,
|
| 5217 |
-
"loss": 1.9571,
|
| 5218 |
-
"step": 743
|
| 5219 |
-
},
|
| 5220 |
-
{
|
| 5221 |
-
"epoch": 1.4506458688764319,
|
| 5222 |
-
"grad_norm": 0.4666212797164917,
|
| 5223 |
-
"learning_rate": 2.71418779986292e-07,
|
| 5224 |
-
"loss": 1.9897,
|
| 5225 |
-
"step": 744
|
| 5226 |
-
},
|
| 5227 |
-
{
|
| 5228 |
-
"epoch": 1.4525956617109432,
|
| 5229 |
-
"grad_norm": 0.45651644468307495,
|
| 5230 |
-
"learning_rate": 2.710760795065113e-07,
|
| 5231 |
-
"loss": 2.0471,
|
| 5232 |
-
"step": 745
|
| 5233 |
-
},
|
| 5234 |
-
{
|
| 5235 |
-
"epoch": 1.4545454545454546,
|
| 5236 |
-
"grad_norm": 0.43935099244117737,
|
| 5237 |
-
"learning_rate": 2.707333790267306e-07,
|
| 5238 |
-
"loss": 1.9525,
|
| 5239 |
-
"step": 746
|
| 5240 |
-
},
|
| 5241 |
-
{
|
| 5242 |
-
"epoch": 1.456495247379966,
|
| 5243 |
-
"grad_norm": 0.4813799560070038,
|
| 5244 |
-
"learning_rate": 2.7039067854695e-07,
|
| 5245 |
-
"loss": 2.0396,
|
| 5246 |
-
"step": 747
|
| 5247 |
-
},
|
| 5248 |
-
{
|
| 5249 |
-
"epoch": 1.4584450402144773,
|
| 5250 |
-
"grad_norm": 0.4743799567222595,
|
| 5251 |
-
"learning_rate": 2.7004797806716927e-07,
|
| 5252 |
-
"loss": 2.0824,
|
| 5253 |
-
"step": 748
|
| 5254 |
-
},
|
| 5255 |
-
{
|
| 5256 |
-
"epoch": 1.4603948330489884,
|
| 5257 |
-
"grad_norm": 0.4927983283996582,
|
| 5258 |
-
"learning_rate": 2.6970527758738857e-07,
|
| 5259 |
-
"loss": 2.0257,
|
| 5260 |
-
"step": 749
|
| 5261 |
-
},
|
| 5262 |
-
{
|
| 5263 |
-
"epoch": 1.4623446258835,
|
| 5264 |
-
"grad_norm": 0.4711035192012787,
|
| 5265 |
-
"learning_rate": 2.6936257710760797e-07,
|
| 5266 |
-
"loss": 2.0487,
|
| 5267 |
-
"step": 750
|
| 5268 |
-
},
|
| 5269 |
-
{
|
| 5270 |
-
"epoch": 1.4642944187180111,
|
| 5271 |
-
"grad_norm": 0.4515864849090576,
|
| 5272 |
-
"learning_rate": 2.6901987662782726e-07,
|
| 5273 |
-
"loss": 2.0244,
|
| 5274 |
-
"step": 751
|
| 5275 |
-
},
|
| 5276 |
-
{
|
| 5277 |
-
"epoch": 1.4662442115525225,
|
| 5278 |
-
"grad_norm": 0.46076542139053345,
|
| 5279 |
-
"learning_rate": 2.6867717614804656e-07,
|
| 5280 |
-
"loss": 2.07,
|
| 5281 |
-
"step": 752
|
| 5282 |
-
},
|
| 5283 |
-
{
|
| 5284 |
-
"epoch": 1.4681940043870338,
|
| 5285 |
-
"grad_norm": 0.44762691855430603,
|
| 5286 |
-
"learning_rate": 2.6833447566826596e-07,
|
| 5287 |
-
"loss": 2.0297,
|
| 5288 |
-
"step": 753
|
| 5289 |
-
},
|
| 5290 |
-
{
|
| 5291 |
-
"epoch": 1.4701437972215452,
|
| 5292 |
-
"grad_norm": 0.4801499843597412,
|
| 5293 |
-
"learning_rate": 2.6799177518848525e-07,
|
| 5294 |
-
"loss": 2.0683,
|
| 5295 |
-
"step": 754
|
| 5296 |
-
},
|
| 5297 |
-
{
|
| 5298 |
-
"epoch": 1.4720935900560566,
|
| 5299 |
-
"grad_norm": 0.45053598284721375,
|
| 5300 |
-
"learning_rate": 2.6764907470870455e-07,
|
| 5301 |
-
"loss": 1.9783,
|
| 5302 |
-
"step": 755
|
| 5303 |
-
},
|
| 5304 |
-
{
|
| 5305 |
-
"epoch": 1.474043382890568,
|
| 5306 |
-
"grad_norm": 0.45730066299438477,
|
| 5307 |
-
"learning_rate": 2.673063742289239e-07,
|
| 5308 |
-
"loss": 2.0548,
|
| 5309 |
-
"step": 756
|
| 5310 |
-
},
|
| 5311 |
-
{
|
| 5312 |
-
"epoch": 1.4759931757250793,
|
| 5313 |
-
"grad_norm": 0.4543995261192322,
|
| 5314 |
-
"learning_rate": 2.6696367374914324e-07,
|
| 5315 |
-
"loss": 2.0306,
|
| 5316 |
-
"step": 757
|
| 5317 |
-
},
|
| 5318 |
-
{
|
| 5319 |
-
"epoch": 1.4779429685595906,
|
| 5320 |
-
"grad_norm": 0.4372531473636627,
|
| 5321 |
-
"learning_rate": 2.6662097326936254e-07,
|
| 5322 |
-
"loss": 2.0164,
|
| 5323 |
-
"step": 758
|
| 5324 |
-
},
|
| 5325 |
-
{
|
| 5326 |
-
"epoch": 1.479892761394102,
|
| 5327 |
-
"grad_norm": 0.44617414474487305,
|
| 5328 |
-
"learning_rate": 2.662782727895819e-07,
|
| 5329 |
-
"loss": 1.9891,
|
| 5330 |
-
"step": 759
|
| 5331 |
-
},
|
| 5332 |
-
{
|
| 5333 |
-
"epoch": 1.481842554228613,
|
| 5334 |
-
"grad_norm": 0.4605617821216583,
|
| 5335 |
-
"learning_rate": 2.6593557230980123e-07,
|
| 5336 |
-
"loss": 2.01,
|
| 5337 |
-
"step": 760
|
| 5338 |
-
},
|
| 5339 |
-
{
|
| 5340 |
-
"epoch": 1.4837923470631245,
|
| 5341 |
-
"grad_norm": 0.4638999402523041,
|
| 5342 |
-
"learning_rate": 2.655928718300205e-07,
|
| 5343 |
-
"loss": 2.0685,
|
| 5344 |
-
"step": 761
|
| 5345 |
-
},
|
| 5346 |
-
{
|
| 5347 |
-
"epoch": 1.4857421398976358,
|
| 5348 |
-
"grad_norm": 0.4548538327217102,
|
| 5349 |
-
"learning_rate": 2.6525017135023987e-07,
|
| 5350 |
-
"loss": 2.0665,
|
| 5351 |
-
"step": 762
|
| 5352 |
-
},
|
| 5353 |
-
{
|
| 5354 |
-
"epoch": 1.4876919327321472,
|
| 5355 |
-
"grad_norm": 0.44948044419288635,
|
| 5356 |
-
"learning_rate": 2.649074708704592e-07,
|
| 5357 |
-
"loss": 1.9921,
|
| 5358 |
-
"step": 763
|
| 5359 |
-
},
|
| 5360 |
-
{
|
| 5361 |
-
"epoch": 1.4896417255666585,
|
| 5362 |
-
"grad_norm": 0.4577581286430359,
|
| 5363 |
-
"learning_rate": 2.645647703906785e-07,
|
| 5364 |
-
"loss": 2.0392,
|
| 5365 |
-
"step": 764
|
| 5366 |
-
},
|
| 5367 |
-
{
|
| 5368 |
-
"epoch": 1.4915915184011699,
|
| 5369 |
-
"grad_norm": 0.4821256101131439,
|
| 5370 |
-
"learning_rate": 2.6422206991089786e-07,
|
| 5371 |
-
"loss": 2.1304,
|
| 5372 |
-
"step": 765
|
| 5373 |
-
},
|
| 5374 |
-
{
|
| 5375 |
-
"epoch": 1.4935413112356812,
|
| 5376 |
-
"grad_norm": 0.48839786648750305,
|
| 5377 |
-
"learning_rate": 2.638793694311172e-07,
|
| 5378 |
-
"loss": 2.0773,
|
| 5379 |
-
"step": 766
|
| 5380 |
-
},
|
| 5381 |
-
{
|
| 5382 |
-
"epoch": 1.4954911040701926,
|
| 5383 |
-
"grad_norm": 0.43702590465545654,
|
| 5384 |
-
"learning_rate": 2.635366689513365e-07,
|
| 5385 |
-
"loss": 2.02,
|
| 5386 |
-
"step": 767
|
| 5387 |
-
},
|
| 5388 |
-
{
|
| 5389 |
-
"epoch": 1.497440896904704,
|
| 5390 |
-
"grad_norm": 0.45477136969566345,
|
| 5391 |
-
"learning_rate": 2.6319396847155585e-07,
|
| 5392 |
-
"loss": 1.9962,
|
| 5393 |
-
"step": 768
|
| 5394 |
-
},
|
| 5395 |
-
{
|
| 5396 |
-
"epoch": 1.499390689739215,
|
| 5397 |
-
"grad_norm": 0.47229456901550293,
|
| 5398 |
-
"learning_rate": 2.6285126799177515e-07,
|
| 5399 |
-
"loss": 2.0281,
|
| 5400 |
-
"step": 769
|
| 5401 |
-
},
|
| 5402 |
-
{
|
| 5403 |
-
"epoch": 1.5013404825737267,
|
| 5404 |
-
"grad_norm": 0.4817400276660919,
|
| 5405 |
-
"learning_rate": 2.625085675119945e-07,
|
| 5406 |
-
"loss": 2.1009,
|
| 5407 |
-
"step": 770
|
| 5408 |
-
},
|
| 5409 |
-
{
|
| 5410 |
-
"epoch": 1.5032902754082378,
|
| 5411 |
-
"grad_norm": 0.4645569324493408,
|
| 5412 |
-
"learning_rate": 2.6216586703221384e-07,
|
| 5413 |
-
"loss": 2.083,
|
| 5414 |
-
"step": 771
|
| 5415 |
-
},
|
| 5416 |
-
{
|
| 5417 |
-
"epoch": 1.5052400682427494,
|
| 5418 |
-
"grad_norm": 0.44810667634010315,
|
| 5419 |
-
"learning_rate": 2.6182316655243314e-07,
|
| 5420 |
-
"loss": 2.09,
|
| 5421 |
-
"step": 772
|
| 5422 |
-
},
|
| 5423 |
-
{
|
| 5424 |
-
"epoch": 1.5071898610772605,
|
| 5425 |
-
"grad_norm": 0.44432902336120605,
|
| 5426 |
-
"learning_rate": 2.614804660726525e-07,
|
| 5427 |
-
"loss": 2.0126,
|
| 5428 |
-
"step": 773
|
| 5429 |
-
},
|
| 5430 |
-
{
|
| 5431 |
-
"epoch": 1.5091396539117719,
|
| 5432 |
-
"grad_norm": 0.4630286991596222,
|
| 5433 |
-
"learning_rate": 2.6113776559287183e-07,
|
| 5434 |
-
"loss": 2.0136,
|
| 5435 |
-
"step": 774
|
| 5436 |
-
},
|
| 5437 |
-
{
|
| 5438 |
-
"epoch": 1.5110894467462832,
|
| 5439 |
-
"grad_norm": 0.44443148374557495,
|
| 5440 |
-
"learning_rate": 2.607950651130911e-07,
|
| 5441 |
-
"loss": 1.9979,
|
| 5442 |
-
"step": 775
|
| 5443 |
-
},
|
| 5444 |
-
{
|
| 5445 |
-
"epoch": 1.5130392395807946,
|
| 5446 |
-
"grad_norm": 0.44903403520584106,
|
| 5447 |
-
"learning_rate": 2.6045236463331047e-07,
|
| 5448 |
-
"loss": 1.9788,
|
| 5449 |
-
"step": 776
|
| 5450 |
-
},
|
| 5451 |
-
{
|
| 5452 |
-
"epoch": 1.514989032415306,
|
| 5453 |
-
"grad_norm": 0.45394134521484375,
|
| 5454 |
-
"learning_rate": 2.601096641535298e-07,
|
| 5455 |
-
"loss": 1.9529,
|
| 5456 |
-
"step": 777
|
| 5457 |
-
},
|
| 5458 |
-
{
|
| 5459 |
-
"epoch": 1.516938825249817,
|
| 5460 |
-
"grad_norm": 0.46713778376579285,
|
| 5461 |
-
"learning_rate": 2.597669636737491e-07,
|
| 5462 |
-
"loss": 2.0212,
|
| 5463 |
-
"step": 778
|
| 5464 |
-
},
|
| 5465 |
-
{
|
| 5466 |
-
"epoch": 1.5188886180843286,
|
| 5467 |
-
"grad_norm": 0.45262840390205383,
|
| 5468 |
-
"learning_rate": 2.5942426319396846e-07,
|
| 5469 |
-
"loss": 2.0723,
|
| 5470 |
-
"step": 779
|
| 5471 |
-
},
|
| 5472 |
-
{
|
| 5473 |
-
"epoch": 1.5208384109188398,
|
| 5474 |
-
"grad_norm": 0.4648626446723938,
|
| 5475 |
-
"learning_rate": 2.590815627141878e-07,
|
| 5476 |
-
"loss": 2.0046,
|
| 5477 |
-
"step": 780
|
| 5478 |
-
},
|
| 5479 |
-
{
|
| 5480 |
-
"epoch": 1.5227882037533513,
|
| 5481 |
-
"grad_norm": 0.4754423201084137,
|
| 5482 |
-
"learning_rate": 2.587388622344071e-07,
|
| 5483 |
-
"loss": 2.0434,
|
| 5484 |
-
"step": 781
|
| 5485 |
-
},
|
| 5486 |
-
{
|
| 5487 |
-
"epoch": 1.5247379965878625,
|
| 5488 |
-
"grad_norm": 0.4271760880947113,
|
| 5489 |
-
"learning_rate": 2.583961617546264e-07,
|
| 5490 |
-
"loss": 2.0843,
|
| 5491 |
-
"step": 782
|
| 5492 |
-
},
|
| 5493 |
-
{
|
| 5494 |
-
"epoch": 1.5266877894223738,
|
| 5495 |
-
"grad_norm": 0.48139727115631104,
|
| 5496 |
-
"learning_rate": 2.580534612748458e-07,
|
| 5497 |
-
"loss": 2.098,
|
| 5498 |
-
"step": 783
|
| 5499 |
-
},
|
| 5500 |
-
{
|
| 5501 |
-
"epoch": 1.5286375822568852,
|
| 5502 |
-
"grad_norm": 0.473366379737854,
|
| 5503 |
-
"learning_rate": 2.577107607950651e-07,
|
| 5504 |
-
"loss": 2.0422,
|
| 5505 |
-
"step": 784
|
| 5506 |
-
},
|
| 5507 |
-
{
|
| 5508 |
-
"epoch": 1.5305873750913965,
|
| 5509 |
-
"grad_norm": 0.4580918848514557,
|
| 5510 |
-
"learning_rate": 2.573680603152844e-07,
|
| 5511 |
-
"loss": 2.006,
|
| 5512 |
-
"step": 785
|
| 5513 |
-
},
|
| 5514 |
-
{
|
| 5515 |
-
"epoch": 1.5325371679259079,
|
| 5516 |
-
"grad_norm": 0.4635441303253174,
|
| 5517 |
-
"learning_rate": 2.570253598355038e-07,
|
| 5518 |
-
"loss": 1.9736,
|
| 5519 |
-
"step": 786
|
| 5520 |
-
},
|
| 5521 |
-
{
|
| 5522 |
-
"epoch": 1.5344869607604192,
|
| 5523 |
-
"grad_norm": 0.4621422290802002,
|
| 5524 |
-
"learning_rate": 2.566826593557231e-07,
|
| 5525 |
-
"loss": 2.1078,
|
| 5526 |
-
"step": 787
|
| 5527 |
-
},
|
| 5528 |
-
{
|
| 5529 |
-
"epoch": 1.5364367535949306,
|
| 5530 |
-
"grad_norm": 0.4151935279369354,
|
| 5531 |
-
"learning_rate": 2.563399588759424e-07,
|
| 5532 |
-
"loss": 2.0092,
|
| 5533 |
-
"step": 788
|
| 5534 |
-
},
|
| 5535 |
-
{
|
| 5536 |
-
"epoch": 1.5383865464294417,
|
| 5537 |
-
"grad_norm": 0.4793336093425751,
|
| 5538 |
-
"learning_rate": 2.559972583961618e-07,
|
| 5539 |
-
"loss": 2.0173,
|
| 5540 |
-
"step": 789
|
| 5541 |
-
},
|
| 5542 |
-
{
|
| 5543 |
-
"epoch": 1.5403363392639533,
|
| 5544 |
-
"grad_norm": 0.4768364429473877,
|
| 5545 |
-
"learning_rate": 2.5565455791638107e-07,
|
| 5546 |
-
"loss": 2.0813,
|
| 5547 |
-
"step": 790
|
| 5548 |
-
},
|
| 5549 |
-
{
|
| 5550 |
-
"epoch": 1.5422861320984644,
|
| 5551 |
-
"grad_norm": 0.452411949634552,
|
| 5552 |
-
"learning_rate": 2.5531185743660037e-07,
|
| 5553 |
-
"loss": 2.0527,
|
| 5554 |
-
"step": 791
|
| 5555 |
-
},
|
| 5556 |
-
{
|
| 5557 |
-
"epoch": 1.544235924932976,
|
| 5558 |
-
"grad_norm": 0.44334676861763,
|
| 5559 |
-
"learning_rate": 2.5496915695681977e-07,
|
| 5560 |
-
"loss": 1.9701,
|
| 5561 |
-
"step": 792
|
| 5562 |
-
},
|
| 5563 |
-
{
|
| 5564 |
-
"epoch": 1.5461857177674871,
|
| 5565 |
-
"grad_norm": 0.4465942978858948,
|
| 5566 |
-
"learning_rate": 2.5462645647703906e-07,
|
| 5567 |
-
"loss": 1.9905,
|
| 5568 |
-
"step": 793
|
| 5569 |
-
},
|
| 5570 |
-
{
|
| 5571 |
-
"epoch": 1.5481355106019985,
|
| 5572 |
-
"grad_norm": 0.4681743085384369,
|
| 5573 |
-
"learning_rate": 2.5428375599725836e-07,
|
| 5574 |
-
"loss": 2.0654,
|
| 5575 |
-
"step": 794
|
| 5576 |
-
},
|
| 5577 |
-
{
|
| 5578 |
-
"epoch": 1.5500853034365099,
|
| 5579 |
-
"grad_norm": 0.46780961751937866,
|
| 5580 |
-
"learning_rate": 2.539410555174777e-07,
|
| 5581 |
-
"loss": 2.0336,
|
| 5582 |
-
"step": 795
|
| 5583 |
-
},
|
| 5584 |
-
{
|
| 5585 |
-
"epoch": 1.5520350962710212,
|
| 5586 |
-
"grad_norm": 0.44133254885673523,
|
| 5587 |
-
"learning_rate": 2.5359835503769705e-07,
|
| 5588 |
-
"loss": 1.9668,
|
| 5589 |
-
"step": 796
|
| 5590 |
-
},
|
| 5591 |
-
{
|
| 5592 |
-
"epoch": 1.5539848891055326,
|
| 5593 |
-
"grad_norm": 0.45011645555496216,
|
| 5594 |
-
"learning_rate": 2.5325565455791635e-07,
|
| 5595 |
-
"loss": 2.0099,
|
| 5596 |
-
"step": 797
|
| 5597 |
-
},
|
| 5598 |
-
{
|
| 5599 |
-
"epoch": 1.555934681940044,
|
| 5600 |
-
"grad_norm": 0.41162246465682983,
|
| 5601 |
-
"learning_rate": 2.529129540781357e-07,
|
| 5602 |
-
"loss": 1.9684,
|
| 5603 |
-
"step": 798
|
| 5604 |
-
},
|
| 5605 |
-
{
|
| 5606 |
-
"epoch": 1.5578844747745553,
|
| 5607 |
-
"grad_norm": 0.438760906457901,
|
| 5608 |
-
"learning_rate": 2.5257025359835504e-07,
|
| 5609 |
-
"loss": 1.9934,
|
| 5610 |
-
"step": 799
|
| 5611 |
-
},
|
| 5612 |
-
{
|
| 5613 |
-
"epoch": 1.5598342676090664,
|
| 5614 |
-
"grad_norm": 0.45921608805656433,
|
| 5615 |
-
"learning_rate": 2.5222755311857434e-07,
|
| 5616 |
-
"loss": 2.0447,
|
| 5617 |
-
"step": 800
|
| 5618 |
-
},
|
| 5619 |
-
{
|
| 5620 |
-
"epoch": 1.561784060443578,
|
| 5621 |
-
"grad_norm": 0.4474433958530426,
|
| 5622 |
-
"learning_rate": 2.518848526387937e-07,
|
| 5623 |
-
"loss": 2.0508,
|
| 5624 |
-
"step": 801
|
| 5625 |
-
},
|
| 5626 |
-
{
|
| 5627 |
-
"epoch": 1.5637338532780891,
|
| 5628 |
-
"grad_norm": 0.42901015281677246,
|
| 5629 |
-
"learning_rate": 2.5154215215901303e-07,
|
| 5630 |
-
"loss": 2.0607,
|
| 5631 |
-
"step": 802
|
| 5632 |
-
},
|
| 5633 |
-
{
|
| 5634 |
-
"epoch": 1.5656836461126007,
|
| 5635 |
-
"grad_norm": 0.4604319632053375,
|
| 5636 |
-
"learning_rate": 2.511994516792323e-07,
|
| 5637 |
-
"loss": 2.0142,
|
| 5638 |
-
"step": 803
|
| 5639 |
-
},
|
| 5640 |
-
{
|
| 5641 |
-
"epoch": 1.5676334389471118,
|
| 5642 |
-
"grad_norm": 0.4305102527141571,
|
| 5643 |
-
"learning_rate": 2.5085675119945167e-07,
|
| 5644 |
-
"loss": 1.9828,
|
| 5645 |
-
"step": 804
|
| 5646 |
-
},
|
| 5647 |
-
{
|
| 5648 |
-
"epoch": 1.5695832317816232,
|
| 5649 |
-
"grad_norm": 0.4656990170478821,
|
| 5650 |
-
"learning_rate": 2.50514050719671e-07,
|
| 5651 |
-
"loss": 2.0302,
|
| 5652 |
-
"step": 805
|
| 5653 |
-
},
|
| 5654 |
-
{
|
| 5655 |
-
"epoch": 1.5715330246161345,
|
| 5656 |
-
"grad_norm": 0.4602496325969696,
|
| 5657 |
-
"learning_rate": 2.501713502398903e-07,
|
| 5658 |
-
"loss": 2.0412,
|
| 5659 |
-
"step": 806
|
| 5660 |
-
},
|
| 5661 |
-
{
|
| 5662 |
-
"epoch": 1.5734828174506459,
|
| 5663 |
-
"grad_norm": 0.4626891314983368,
|
| 5664 |
-
"learning_rate": 2.4982864976010966e-07,
|
| 5665 |
-
"loss": 2.0513,
|
| 5666 |
-
"step": 807
|
| 5667 |
-
},
|
| 5668 |
-
{
|
| 5669 |
-
"epoch": 1.5754326102851572,
|
| 5670 |
-
"grad_norm": 0.4671951234340668,
|
| 5671 |
-
"learning_rate": 2.4948594928032896e-07,
|
| 5672 |
-
"loss": 2.003,
|
| 5673 |
-
"step": 808
|
| 5674 |
-
},
|
| 5675 |
-
{
|
| 5676 |
-
"epoch": 1.5773824031196684,
|
| 5677 |
-
"grad_norm": 0.4399751126766205,
|
| 5678 |
-
"learning_rate": 2.491432488005483e-07,
|
| 5679 |
-
"loss": 2.0532,
|
| 5680 |
-
"step": 809
|
| 5681 |
-
},
|
| 5682 |
-
{
|
| 5683 |
-
"epoch": 1.57933219595418,
|
| 5684 |
-
"grad_norm": 0.4228038191795349,
|
| 5685 |
-
"learning_rate": 2.4880054832076765e-07,
|
| 5686 |
-
"loss": 2.0078,
|
| 5687 |
-
"step": 810
|
| 5688 |
-
},
|
| 5689 |
-
{
|
| 5690 |
-
"epoch": 1.581281988788691,
|
| 5691 |
-
"grad_norm": 0.4445479214191437,
|
| 5692 |
-
"learning_rate": 2.4845784784098695e-07,
|
| 5693 |
-
"loss": 2.0142,
|
| 5694 |
-
"step": 811
|
| 5695 |
-
},
|
| 5696 |
-
{
|
| 5697 |
-
"epoch": 1.5832317816232027,
|
| 5698 |
-
"grad_norm": 0.4397488534450531,
|
| 5699 |
-
"learning_rate": 2.481151473612063e-07,
|
| 5700 |
-
"loss": 2.0468,
|
| 5701 |
-
"step": 812
|
| 5702 |
-
},
|
| 5703 |
-
{
|
| 5704 |
-
"epoch": 1.5851815744577138,
|
| 5705 |
-
"grad_norm": 0.48187440633773804,
|
| 5706 |
-
"learning_rate": 2.4777244688142564e-07,
|
| 5707 |
-
"loss": 2.0444,
|
| 5708 |
-
"step": 813
|
| 5709 |
-
},
|
| 5710 |
-
{
|
| 5711 |
-
"epoch": 1.5871313672922251,
|
| 5712 |
-
"grad_norm": 0.4355807304382324,
|
| 5713 |
-
"learning_rate": 2.4742974640164494e-07,
|
| 5714 |
-
"loss": 1.9955,
|
| 5715 |
-
"step": 814
|
| 5716 |
-
},
|
| 5717 |
-
{
|
| 5718 |
-
"epoch": 1.5890811601267365,
|
| 5719 |
-
"grad_norm": 0.4219972491264343,
|
| 5720 |
-
"learning_rate": 2.470870459218643e-07,
|
| 5721 |
-
"loss": 1.9971,
|
| 5722 |
-
"step": 815
|
| 5723 |
-
},
|
| 5724 |
-
{
|
| 5725 |
-
"epoch": 1.5910309529612479,
|
| 5726 |
-
"grad_norm": 0.44700267910957336,
|
| 5727 |
-
"learning_rate": 2.4674434544208363e-07,
|
| 5728 |
-
"loss": 2.0297,
|
| 5729 |
-
"step": 816
|
| 5730 |
-
},
|
| 5731 |
-
{
|
| 5732 |
-
"epoch": 1.5929807457957592,
|
| 5733 |
-
"grad_norm": 0.45433923602104187,
|
| 5734 |
-
"learning_rate": 2.464016449623029e-07,
|
| 5735 |
-
"loss": 2.0064,
|
| 5736 |
-
"step": 817
|
| 5737 |
-
},
|
| 5738 |
-
{
|
| 5739 |
-
"epoch": 1.5949305386302706,
|
| 5740 |
-
"grad_norm": 0.4188825488090515,
|
| 5741 |
-
"learning_rate": 2.4605894448252227e-07,
|
| 5742 |
-
"loss": 2.0236,
|
| 5743 |
-
"step": 818
|
| 5744 |
-
},
|
| 5745 |
-
{
|
| 5746 |
-
"epoch": 1.596880331464782,
|
| 5747 |
-
"grad_norm": 0.4635048508644104,
|
| 5748 |
-
"learning_rate": 2.457162440027416e-07,
|
| 5749 |
-
"loss": 2.0652,
|
| 5750 |
-
"step": 819
|
| 5751 |
-
},
|
| 5752 |
-
{
|
| 5753 |
-
"epoch": 1.598830124299293,
|
| 5754 |
-
"grad_norm": 0.4555036127567291,
|
| 5755 |
-
"learning_rate": 2.453735435229609e-07,
|
| 5756 |
-
"loss": 2.079,
|
| 5757 |
-
"step": 820
|
| 5758 |
-
},
|
| 5759 |
-
{
|
| 5760 |
-
"epoch": 1.6007799171338046,
|
| 5761 |
-
"grad_norm": 0.45152541995048523,
|
| 5762 |
-
"learning_rate": 2.4503084304318026e-07,
|
| 5763 |
-
"loss": 1.9724,
|
| 5764 |
-
"step": 821
|
| 5765 |
-
},
|
| 5766 |
-
{
|
| 5767 |
-
"epoch": 1.6027297099683158,
|
| 5768 |
-
"grad_norm": 0.4355667233467102,
|
| 5769 |
-
"learning_rate": 2.446881425633996e-07,
|
| 5770 |
-
"loss": 2.0444,
|
| 5771 |
-
"step": 822
|
| 5772 |
-
},
|
| 5773 |
-
{
|
| 5774 |
-
"epoch": 1.6046795028028273,
|
| 5775 |
-
"grad_norm": 0.42853429913520813,
|
| 5776 |
-
"learning_rate": 2.443454420836189e-07,
|
| 5777 |
-
"loss": 1.9451,
|
| 5778 |
-
"step": 823
|
| 5779 |
-
},
|
| 5780 |
-
{
|
| 5781 |
-
"epoch": 1.6066292956373385,
|
| 5782 |
-
"grad_norm": 0.4546351134777069,
|
| 5783 |
-
"learning_rate": 2.4400274160383825e-07,
|
| 5784 |
-
"loss": 2.015,
|
| 5785 |
-
"step": 824
|
| 5786 |
-
},
|
| 5787 |
-
{
|
| 5788 |
-
"epoch": 1.6085790884718498,
|
| 5789 |
-
"grad_norm": 0.45015424489974976,
|
| 5790 |
-
"learning_rate": 2.4366004112405755e-07,
|
| 5791 |
-
"loss": 2.0171,
|
| 5792 |
-
"step": 825
|
| 5793 |
-
},
|
| 5794 |
-
{
|
| 5795 |
-
"epoch": 1.6105288813063612,
|
| 5796 |
-
"grad_norm": 0.446065217256546,
|
| 5797 |
-
"learning_rate": 2.433173406442769e-07,
|
| 5798 |
-
"loss": 2.0085,
|
| 5799 |
-
"step": 826
|
| 5800 |
-
},
|
| 5801 |
-
{
|
| 5802 |
-
"epoch": 1.6124786741408725,
|
| 5803 |
-
"grad_norm": 0.46771183609962463,
|
| 5804 |
-
"learning_rate": 2.4297464016449624e-07,
|
| 5805 |
-
"loss": 1.9844,
|
| 5806 |
-
"step": 827
|
| 5807 |
-
},
|
| 5808 |
-
{
|
| 5809 |
-
"epoch": 1.614428466975384,
|
| 5810 |
-
"grad_norm": 0.4590853452682495,
|
| 5811 |
-
"learning_rate": 2.4263193968471554e-07,
|
| 5812 |
-
"loss": 2.0031,
|
| 5813 |
-
"step": 828
|
| 5814 |
-
},
|
| 5815 |
-
{
|
| 5816 |
-
"epoch": 1.6163782598098952,
|
| 5817 |
-
"grad_norm": 0.4465842545032501,
|
| 5818 |
-
"learning_rate": 2.422892392049349e-07,
|
| 5819 |
-
"loss": 2.0344,
|
| 5820 |
-
"step": 829
|
| 5821 |
-
},
|
| 5822 |
-
{
|
| 5823 |
-
"epoch": 1.6183280526444066,
|
| 5824 |
-
"grad_norm": 0.40251830220222473,
|
| 5825 |
-
"learning_rate": 2.419465387251542e-07,
|
| 5826 |
-
"loss": 2.0129,
|
| 5827 |
-
"step": 830
|
| 5828 |
-
},
|
| 5829 |
-
{
|
| 5830 |
-
"epoch": 1.6202778454789177,
|
| 5831 |
-
"grad_norm": 0.45284631848335266,
|
| 5832 |
-
"learning_rate": 2.416038382453735e-07,
|
| 5833 |
-
"loss": 2.0354,
|
| 5834 |
-
"step": 831
|
| 5835 |
-
},
|
| 5836 |
-
{
|
| 5837 |
-
"epoch": 1.6222276383134293,
|
| 5838 |
-
"grad_norm": 0.4733079969882965,
|
| 5839 |
-
"learning_rate": 2.4126113776559287e-07,
|
| 5840 |
-
"loss": 1.993,
|
| 5841 |
-
"step": 832
|
| 5842 |
-
},
|
| 5843 |
-
{
|
| 5844 |
-
"epoch": 1.6241774311479404,
|
| 5845 |
-
"grad_norm": 0.4264031946659088,
|
| 5846 |
-
"learning_rate": 2.4091843728581217e-07,
|
| 5847 |
-
"loss": 2.007,
|
| 5848 |
-
"step": 833
|
| 5849 |
-
},
|
| 5850 |
-
{
|
| 5851 |
-
"epoch": 1.626127223982452,
|
| 5852 |
-
"grad_norm": 0.46400555968284607,
|
| 5853 |
-
"learning_rate": 2.405757368060315e-07,
|
| 5854 |
-
"loss": 1.9825,
|
| 5855 |
-
"step": 834
|
| 5856 |
-
},
|
| 5857 |
-
{
|
| 5858 |
-
"epoch": 1.6280770168169632,
|
| 5859 |
-
"grad_norm": 0.4408418834209442,
|
| 5860 |
-
"learning_rate": 2.4023303632625086e-07,
|
| 5861 |
-
"loss": 2.0199,
|
| 5862 |
-
"step": 835
|
| 5863 |
-
},
|
| 5864 |
-
{
|
| 5865 |
-
"epoch": 1.6300268096514745,
|
| 5866 |
-
"grad_norm": 0.4353219270706177,
|
| 5867 |
-
"learning_rate": 2.3989033584647016e-07,
|
| 5868 |
-
"loss": 1.9767,
|
| 5869 |
-
"step": 836
|
| 5870 |
-
},
|
| 5871 |
-
{
|
| 5872 |
-
"epoch": 1.6319766024859859,
|
| 5873 |
-
"grad_norm": 0.47256654500961304,
|
| 5874 |
-
"learning_rate": 2.395476353666895e-07,
|
| 5875 |
-
"loss": 2.0708,
|
| 5876 |
-
"step": 837
|
| 5877 |
-
},
|
| 5878 |
-
{
|
| 5879 |
-
"epoch": 1.6339263953204972,
|
| 5880 |
-
"grad_norm": 0.44208547472953796,
|
| 5881 |
-
"learning_rate": 2.392049348869088e-07,
|
| 5882 |
-
"loss": 2.0518,
|
| 5883 |
-
"step": 838
|
| 5884 |
-
},
|
| 5885 |
-
{
|
| 5886 |
-
"epoch": 1.6358761881550086,
|
| 5887 |
-
"grad_norm": 0.4937672019004822,
|
| 5888 |
-
"learning_rate": 2.3886223440712815e-07,
|
| 5889 |
-
"loss": 2.043,
|
| 5890 |
-
"step": 839
|
| 5891 |
-
},
|
| 5892 |
-
{
|
| 5893 |
-
"epoch": 1.6378259809895197,
|
| 5894 |
-
"grad_norm": 0.46095776557922363,
|
| 5895 |
-
"learning_rate": 2.385195339273475e-07,
|
| 5896 |
-
"loss": 2.0421,
|
| 5897 |
-
"step": 840
|
| 5898 |
-
},
|
| 5899 |
-
{
|
| 5900 |
-
"epoch": 1.6397757738240313,
|
| 5901 |
-
"grad_norm": 0.4658643901348114,
|
| 5902 |
-
"learning_rate": 2.3817683344756682e-07,
|
| 5903 |
-
"loss": 2.0225,
|
| 5904 |
-
"step": 841
|
| 5905 |
-
},
|
| 5906 |
-
{
|
| 5907 |
-
"epoch": 1.6417255666585424,
|
| 5908 |
-
"grad_norm": 0.4451207220554352,
|
| 5909 |
-
"learning_rate": 2.3783413296778616e-07,
|
| 5910 |
-
"loss": 2.0244,
|
| 5911 |
-
"step": 842
|
| 5912 |
-
},
|
| 5913 |
-
{
|
| 5914 |
-
"epoch": 1.643675359493054,
|
| 5915 |
-
"grad_norm": 0.43841567635536194,
|
| 5916 |
-
"learning_rate": 2.3749143248800546e-07,
|
| 5917 |
-
"loss": 1.9797,
|
| 5918 |
-
"step": 843
|
| 5919 |
-
},
|
| 5920 |
-
{
|
| 5921 |
-
"epoch": 1.6456251523275651,
|
| 5922 |
-
"grad_norm": 0.45495790243148804,
|
| 5923 |
-
"learning_rate": 2.371487320082248e-07,
|
| 5924 |
-
"loss": 2.039,
|
| 5925 |
-
"step": 844
|
| 5926 |
-
},
|
| 5927 |
-
{
|
| 5928 |
-
"epoch": 1.6475749451620765,
|
| 5929 |
-
"grad_norm": 0.4694961607456207,
|
| 5930 |
-
"learning_rate": 2.3680603152844415e-07,
|
| 5931 |
-
"loss": 2.0232,
|
| 5932 |
-
"step": 845
|
| 5933 |
-
},
|
| 5934 |
-
{
|
| 5935 |
-
"epoch": 1.6495247379965878,
|
| 5936 |
-
"grad_norm": 0.4593546986579895,
|
| 5937 |
-
"learning_rate": 2.3646333104866345e-07,
|
| 5938 |
-
"loss": 2.0495,
|
| 5939 |
-
"step": 846
|
| 5940 |
-
},
|
| 5941 |
-
{
|
| 5942 |
-
"epoch": 1.6514745308310992,
|
| 5943 |
-
"grad_norm": 0.4738862216472626,
|
| 5944 |
-
"learning_rate": 2.361206305688828e-07,
|
| 5945 |
-
"loss": 2.0105,
|
| 5946 |
-
"step": 847
|
| 5947 |
-
},
|
| 5948 |
-
{
|
| 5949 |
-
"epoch": 1.6534243236656105,
|
| 5950 |
-
"grad_norm": 0.45088139176368713,
|
| 5951 |
-
"learning_rate": 2.357779300891021e-07,
|
| 5952 |
-
"loss": 2.0418,
|
| 5953 |
-
"step": 848
|
| 5954 |
-
},
|
| 5955 |
-
{
|
| 5956 |
-
"epoch": 1.655374116500122,
|
| 5957 |
-
"grad_norm": 0.4501790702342987,
|
| 5958 |
-
"learning_rate": 2.3543522960932144e-07,
|
| 5959 |
-
"loss": 2.0531,
|
| 5960 |
-
"step": 849
|
| 5961 |
-
},
|
| 5962 |
-
{
|
| 5963 |
-
"epoch": 1.6573239093346332,
|
| 5964 |
-
"grad_norm": 0.47187909483909607,
|
| 5965 |
-
"learning_rate": 2.3509252912954078e-07,
|
| 5966 |
-
"loss": 1.9907,
|
| 5967 |
-
"step": 850
|
| 5968 |
-
},
|
| 5969 |
-
{
|
| 5970 |
-
"epoch": 1.6592737021691444,
|
| 5971 |
-
"grad_norm": 0.46769675612449646,
|
| 5972 |
-
"learning_rate": 2.3474982864976008e-07,
|
| 5973 |
-
"loss": 2.0145,
|
| 5974 |
-
"step": 851
|
| 5975 |
-
},
|
| 5976 |
-
{
|
| 5977 |
-
"epoch": 1.661223495003656,
|
| 5978 |
-
"grad_norm": 0.44854676723480225,
|
| 5979 |
-
"learning_rate": 2.3440712816997943e-07,
|
| 5980 |
-
"loss": 2.0381,
|
| 5981 |
-
"step": 852
|
| 5982 |
-
},
|
| 5983 |
-
{
|
| 5984 |
-
"epoch": 1.663173287838167,
|
| 5985 |
-
"grad_norm": 0.4576641023159027,
|
| 5986 |
-
"learning_rate": 2.3406442769019877e-07,
|
| 5987 |
-
"loss": 1.9722,
|
| 5988 |
-
"step": 853
|
| 5989 |
-
},
|
| 5990 |
-
{
|
| 5991 |
-
"epoch": 1.6651230806726787,
|
| 5992 |
-
"grad_norm": 0.4568294584751129,
|
| 5993 |
-
"learning_rate": 2.3372172721041807e-07,
|
| 5994 |
-
"loss": 1.9744,
|
| 5995 |
-
"step": 854
|
| 5996 |
-
},
|
| 5997 |
-
{
|
| 5998 |
-
"epoch": 1.6670728735071898,
|
| 5999 |
-
"grad_norm": 0.4591883718967438,
|
| 6000 |
-
"learning_rate": 2.3337902673063742e-07,
|
| 6001 |
-
"loss": 1.9666,
|
| 6002 |
-
"step": 855
|
| 6003 |
-
},
|
| 6004 |
-
{
|
| 6005 |
-
"epoch": 1.6690226663417012,
|
| 6006 |
-
"grad_norm": 0.44672197103500366,
|
| 6007 |
-
"learning_rate": 2.3303632625085674e-07,
|
| 6008 |
-
"loss": 1.9944,
|
| 6009 |
-
"step": 856
|
| 6010 |
-
},
|
| 6011 |
-
{
|
| 6012 |
-
"epoch": 1.6709724591762125,
|
| 6013 |
-
"grad_norm": 0.4896506667137146,
|
| 6014 |
-
"learning_rate": 2.3269362577107606e-07,
|
| 6015 |
-
"loss": 2.0492,
|
| 6016 |
-
"step": 857
|
| 6017 |
-
},
|
| 6018 |
-
{
|
| 6019 |
-
"epoch": 1.6729222520107239,
|
| 6020 |
-
"grad_norm": 0.4453061521053314,
|
| 6021 |
-
"learning_rate": 2.323509252912954e-07,
|
| 6022 |
-
"loss": 1.9757,
|
| 6023 |
-
"step": 858
|
| 6024 |
-
},
|
| 6025 |
-
{
|
| 6026 |
-
"epoch": 1.6748720448452352,
|
| 6027 |
-
"grad_norm": 0.4569021761417389,
|
| 6028 |
-
"learning_rate": 2.3200822481151473e-07,
|
| 6029 |
-
"loss": 2.0523,
|
| 6030 |
-
"step": 859
|
| 6031 |
-
},
|
| 6032 |
-
{
|
| 6033 |
-
"epoch": 1.6768218376797466,
|
| 6034 |
-
"grad_norm": 0.4553905427455902,
|
| 6035 |
-
"learning_rate": 2.3166552433173405e-07,
|
| 6036 |
-
"loss": 2.0189,
|
| 6037 |
-
"step": 860
|
| 6038 |
-
},
|
| 6039 |
-
{
|
| 6040 |
-
"epoch": 1.678771630514258,
|
| 6041 |
-
"grad_norm": 0.4560829699039459,
|
| 6042 |
-
"learning_rate": 2.3132282385195337e-07,
|
| 6043 |
-
"loss": 2.0833,
|
| 6044 |
-
"step": 861
|
| 6045 |
-
},
|
| 6046 |
-
{
|
| 6047 |
-
"epoch": 1.680721423348769,
|
| 6048 |
-
"grad_norm": 0.4487151503562927,
|
| 6049 |
-
"learning_rate": 2.3098012337217272e-07,
|
| 6050 |
-
"loss": 1.9806,
|
| 6051 |
-
"step": 862
|
| 6052 |
-
},
|
| 6053 |
-
{
|
| 6054 |
-
"epoch": 1.6826712161832806,
|
| 6055 |
-
"grad_norm": 0.440891832113266,
|
| 6056 |
-
"learning_rate": 2.3063742289239204e-07,
|
| 6057 |
-
"loss": 1.9989,
|
| 6058 |
-
"step": 863
|
| 6059 |
-
},
|
| 6060 |
-
{
|
| 6061 |
-
"epoch": 1.6846210090177918,
|
| 6062 |
-
"grad_norm": 0.469881534576416,
|
| 6063 |
-
"learning_rate": 2.3029472241261136e-07,
|
| 6064 |
-
"loss": 2.0626,
|
| 6065 |
-
"step": 864
|
| 6066 |
-
},
|
| 6067 |
-
{
|
| 6068 |
-
"epoch": 1.6865708018523033,
|
| 6069 |
-
"grad_norm": 0.43621349334716797,
|
| 6070 |
-
"learning_rate": 2.299520219328307e-07,
|
| 6071 |
-
"loss": 2.063,
|
| 6072 |
-
"step": 865
|
| 6073 |
-
},
|
| 6074 |
-
{
|
| 6075 |
-
"epoch": 1.6885205946868145,
|
| 6076 |
-
"grad_norm": 0.45750436186790466,
|
| 6077 |
-
"learning_rate": 2.2960932145305003e-07,
|
| 6078 |
-
"loss": 2.0164,
|
| 6079 |
-
"step": 866
|
| 6080 |
-
},
|
| 6081 |
-
{
|
| 6082 |
-
"epoch": 1.6904703875213258,
|
| 6083 |
-
"grad_norm": 0.46832090616226196,
|
| 6084 |
-
"learning_rate": 2.2926662097326935e-07,
|
| 6085 |
-
"loss": 2.0459,
|
| 6086 |
-
"step": 867
|
| 6087 |
-
},
|
| 6088 |
-
{
|
| 6089 |
-
"epoch": 1.6924201803558372,
|
| 6090 |
-
"grad_norm": 0.4424852728843689,
|
| 6091 |
-
"learning_rate": 2.289239204934887e-07,
|
| 6092 |
-
"loss": 2.0148,
|
| 6093 |
-
"step": 868
|
| 6094 |
-
},
|
| 6095 |
-
{
|
| 6096 |
-
"epoch": 1.6943699731903485,
|
| 6097 |
-
"grad_norm": 0.4639265239238739,
|
| 6098 |
-
"learning_rate": 2.28581220013708e-07,
|
| 6099 |
-
"loss": 2.0453,
|
| 6100 |
-
"step": 869
|
| 6101 |
-
},
|
| 6102 |
-
{
|
| 6103 |
-
"epoch": 1.69631976602486,
|
| 6104 |
-
"grad_norm": 0.42720574140548706,
|
| 6105 |
-
"learning_rate": 2.2823851953392734e-07,
|
| 6106 |
-
"loss": 2.0164,
|
| 6107 |
-
"step": 870
|
| 6108 |
-
},
|
| 6109 |
-
{
|
| 6110 |
-
"epoch": 1.698269558859371,
|
| 6111 |
-
"grad_norm": 0.46615973114967346,
|
| 6112 |
-
"learning_rate": 2.2789581905414668e-07,
|
| 6113 |
-
"loss": 2.0235,
|
| 6114 |
-
"step": 871
|
| 6115 |
-
},
|
| 6116 |
-
{
|
| 6117 |
-
"epoch": 1.7002193516938826,
|
| 6118 |
-
"grad_norm": 0.46956273913383484,
|
| 6119 |
-
"learning_rate": 2.2755311857436598e-07,
|
| 6120 |
-
"loss": 2.0668,
|
| 6121 |
-
"step": 872
|
| 6122 |
-
},
|
| 6123 |
-
{
|
| 6124 |
-
"epoch": 1.7021691445283937,
|
| 6125 |
-
"grad_norm": 0.45590096712112427,
|
| 6126 |
-
"learning_rate": 2.2721041809458533e-07,
|
| 6127 |
-
"loss": 2.0767,
|
| 6128 |
-
"step": 873
|
| 6129 |
-
},
|
| 6130 |
-
{
|
| 6131 |
-
"epoch": 1.7041189373629053,
|
| 6132 |
-
"grad_norm": 0.4419032037258148,
|
| 6133 |
-
"learning_rate": 2.2686771761480465e-07,
|
| 6134 |
-
"loss": 2.0298,
|
| 6135 |
-
"step": 874
|
| 6136 |
-
},
|
| 6137 |
-
{
|
| 6138 |
-
"epoch": 1.7060687301974164,
|
| 6139 |
-
"grad_norm": 0.48438993096351624,
|
| 6140 |
-
"learning_rate": 2.2652501713502397e-07,
|
| 6141 |
-
"loss": 2.0881,
|
| 6142 |
-
"step": 875
|
| 6143 |
-
},
|
| 6144 |
-
{
|
| 6145 |
-
"epoch": 1.7080185230319278,
|
| 6146 |
-
"grad_norm": 0.4674246609210968,
|
| 6147 |
-
"learning_rate": 2.2618231665524332e-07,
|
| 6148 |
-
"loss": 1.9858,
|
| 6149 |
-
"step": 876
|
| 6150 |
-
},
|
| 6151 |
-
{
|
| 6152 |
-
"epoch": 1.7099683158664392,
|
| 6153 |
-
"grad_norm": 0.4731968641281128,
|
| 6154 |
-
"learning_rate": 2.2583961617546264e-07,
|
| 6155 |
-
"loss": 2.0684,
|
| 6156 |
-
"step": 877
|
| 6157 |
-
},
|
| 6158 |
-
{
|
| 6159 |
-
"epoch": 1.7119181087009505,
|
| 6160 |
-
"grad_norm": 0.44370540976524353,
|
| 6161 |
-
"learning_rate": 2.2549691569568196e-07,
|
| 6162 |
-
"loss": 2.0222,
|
| 6163 |
-
"step": 878
|
| 6164 |
-
},
|
| 6165 |
-
{
|
| 6166 |
-
"epoch": 1.7138679015354619,
|
| 6167 |
-
"grad_norm": 0.43057727813720703,
|
| 6168 |
-
"learning_rate": 2.251542152159013e-07,
|
| 6169 |
-
"loss": 2.0054,
|
| 6170 |
-
"step": 879
|
| 6171 |
-
},
|
| 6172 |
-
{
|
| 6173 |
-
"epoch": 1.7158176943699732,
|
| 6174 |
-
"grad_norm": 0.4575825035572052,
|
| 6175 |
-
"learning_rate": 2.2481151473612063e-07,
|
| 6176 |
-
"loss": 2.0194,
|
| 6177 |
-
"step": 880
|
| 6178 |
-
},
|
| 6179 |
-
{
|
| 6180 |
-
"epoch": 1.7177674872044846,
|
| 6181 |
-
"grad_norm": 0.46100616455078125,
|
| 6182 |
-
"learning_rate": 2.2446881425633995e-07,
|
| 6183 |
-
"loss": 2.0362,
|
| 6184 |
-
"step": 881
|
| 6185 |
-
},
|
| 6186 |
-
{
|
| 6187 |
-
"epoch": 1.7197172800389957,
|
| 6188 |
-
"grad_norm": 0.46780040860176086,
|
| 6189 |
-
"learning_rate": 2.2412611377655927e-07,
|
| 6190 |
-
"loss": 2.0458,
|
| 6191 |
-
"step": 882
|
| 6192 |
-
},
|
| 6193 |
-
{
|
| 6194 |
-
"epoch": 1.7216670728735073,
|
| 6195 |
-
"grad_norm": 0.4316709339618683,
|
| 6196 |
-
"learning_rate": 2.2378341329677862e-07,
|
| 6197 |
-
"loss": 2.0401,
|
| 6198 |
-
"step": 883
|
| 6199 |
-
},
|
| 6200 |
-
{
|
| 6201 |
-
"epoch": 1.7236168657080184,
|
| 6202 |
-
"grad_norm": 0.43883568048477173,
|
| 6203 |
-
"learning_rate": 2.2344071281699794e-07,
|
| 6204 |
-
"loss": 2.0407,
|
| 6205 |
-
"step": 884
|
| 6206 |
-
},
|
| 6207 |
-
{
|
| 6208 |
-
"epoch": 1.72556665854253,
|
| 6209 |
-
"grad_norm": 0.44989317655563354,
|
| 6210 |
-
"learning_rate": 2.2309801233721726e-07,
|
| 6211 |
-
"loss": 2.0253,
|
| 6212 |
-
"step": 885
|
| 6213 |
-
},
|
| 6214 |
-
{
|
| 6215 |
-
"epoch": 1.7275164513770411,
|
| 6216 |
-
"grad_norm": 0.4468737840652466,
|
| 6217 |
-
"learning_rate": 2.227553118574366e-07,
|
| 6218 |
-
"loss": 2.0336,
|
| 6219 |
-
"step": 886
|
| 6220 |
-
},
|
| 6221 |
-
{
|
| 6222 |
-
"epoch": 1.7294662442115525,
|
| 6223 |
-
"grad_norm": 0.45126405358314514,
|
| 6224 |
-
"learning_rate": 2.224126113776559e-07,
|
| 6225 |
-
"loss": 2.0259,
|
| 6226 |
-
"step": 887
|
| 6227 |
-
},
|
| 6228 |
-
{
|
| 6229 |
-
"epoch": 1.7314160370460638,
|
| 6230 |
-
"grad_norm": 0.43270209431648254,
|
| 6231 |
-
"learning_rate": 2.2206991089787525e-07,
|
| 6232 |
-
"loss": 2.0071,
|
| 6233 |
-
"step": 888
|
| 6234 |
-
},
|
| 6235 |
-
{
|
| 6236 |
-
"epoch": 1.7333658298805752,
|
| 6237 |
-
"grad_norm": 0.4503726363182068,
|
| 6238 |
-
"learning_rate": 2.217272104180946e-07,
|
| 6239 |
-
"loss": 2.1025,
|
| 6240 |
-
"step": 889
|
| 6241 |
-
},
|
| 6242 |
-
{
|
| 6243 |
-
"epoch": 1.7353156227150865,
|
| 6244 |
-
"grad_norm": 0.44900792837142944,
|
| 6245 |
-
"learning_rate": 2.213845099383139e-07,
|
| 6246 |
-
"loss": 1.9883,
|
| 6247 |
-
"step": 890
|
| 6248 |
-
},
|
| 6249 |
-
{
|
| 6250 |
-
"epoch": 1.737265415549598,
|
| 6251 |
-
"grad_norm": 0.4531221091747284,
|
| 6252 |
-
"learning_rate": 2.2104180945853324e-07,
|
| 6253 |
-
"loss": 2.0095,
|
| 6254 |
-
"step": 891
|
| 6255 |
-
},
|
| 6256 |
-
{
|
| 6257 |
-
"epoch": 1.7392152083841093,
|
| 6258 |
-
"grad_norm": 0.46359124779701233,
|
| 6259 |
-
"learning_rate": 2.2069910897875258e-07,
|
| 6260 |
-
"loss": 2.003,
|
| 6261 |
-
"step": 892
|
| 6262 |
-
},
|
| 6263 |
-
{
|
| 6264 |
-
"epoch": 1.7411650012186204,
|
| 6265 |
-
"grad_norm": 0.4506163001060486,
|
| 6266 |
-
"learning_rate": 2.2035640849897188e-07,
|
| 6267 |
-
"loss": 1.9438,
|
| 6268 |
-
"step": 893
|
| 6269 |
-
},
|
| 6270 |
-
{
|
| 6271 |
-
"epoch": 1.743114794053132,
|
| 6272 |
-
"grad_norm": 0.4618943929672241,
|
| 6273 |
-
"learning_rate": 2.2001370801919123e-07,
|
| 6274 |
-
"loss": 2.0772,
|
| 6275 |
-
"step": 894
|
| 6276 |
-
},
|
| 6277 |
-
{
|
| 6278 |
-
"epoch": 1.745064586887643,
|
| 6279 |
-
"grad_norm": 0.4341379404067993,
|
| 6280 |
-
"learning_rate": 2.1967100753941055e-07,
|
| 6281 |
-
"loss": 1.9443,
|
| 6282 |
-
"step": 895
|
| 6283 |
-
},
|
| 6284 |
-
{
|
| 6285 |
-
"epoch": 1.7470143797221547,
|
| 6286 |
-
"grad_norm": 0.4800126254558563,
|
| 6287 |
-
"learning_rate": 2.1932830705962987e-07,
|
| 6288 |
-
"loss": 1.9994,
|
| 6289 |
-
"step": 896
|
| 6290 |
-
},
|
| 6291 |
-
{
|
| 6292 |
-
"epoch": 1.7489641725566658,
|
| 6293 |
-
"grad_norm": 0.45474764704704285,
|
| 6294 |
-
"learning_rate": 2.1898560657984922e-07,
|
| 6295 |
-
"loss": 2.0635,
|
| 6296 |
-
"step": 897
|
| 6297 |
-
},
|
| 6298 |
-
{
|
| 6299 |
-
"epoch": 1.7509139653911772,
|
| 6300 |
-
"grad_norm": 0.44301092624664307,
|
| 6301 |
-
"learning_rate": 2.1864290610006854e-07,
|
| 6302 |
-
"loss": 1.9752,
|
| 6303 |
-
"step": 898
|
| 6304 |
-
},
|
| 6305 |
-
{
|
| 6306 |
-
"epoch": 1.7528637582256885,
|
| 6307 |
-
"grad_norm": 0.4428479075431824,
|
| 6308 |
-
"learning_rate": 2.1830020562028786e-07,
|
| 6309 |
-
"loss": 1.9371,
|
| 6310 |
-
"step": 899
|
| 6311 |
-
},
|
| 6312 |
-
{
|
| 6313 |
-
"epoch": 1.7548135510601999,
|
| 6314 |
-
"grad_norm": 0.4576126039028168,
|
| 6315 |
-
"learning_rate": 2.1795750514050718e-07,
|
| 6316 |
-
"loss": 2.063,
|
| 6317 |
-
"step": 900
|
| 6318 |
-
},
|
| 6319 |
-
{
|
| 6320 |
-
"epoch": 1.7567633438947112,
|
| 6321 |
-
"grad_norm": 0.47722387313842773,
|
| 6322 |
-
"learning_rate": 2.1761480466072653e-07,
|
| 6323 |
-
"loss": 2.0743,
|
| 6324 |
-
"step": 901
|
| 6325 |
-
},
|
| 6326 |
-
{
|
| 6327 |
-
"epoch": 1.7587131367292224,
|
| 6328 |
-
"grad_norm": 0.4575481712818146,
|
| 6329 |
-
"learning_rate": 2.1727210418094585e-07,
|
| 6330 |
-
"loss": 1.9873,
|
| 6331 |
-
"step": 902
|
| 6332 |
-
},
|
| 6333 |
-
{
|
| 6334 |
-
"epoch": 1.760662929563734,
|
| 6335 |
-
"grad_norm": 0.4340214729309082,
|
| 6336 |
-
"learning_rate": 2.1692940370116517e-07,
|
| 6337 |
-
"loss": 1.9459,
|
| 6338 |
-
"step": 903
|
| 6339 |
-
},
|
| 6340 |
-
{
|
| 6341 |
-
"epoch": 1.762612722398245,
|
| 6342 |
-
"grad_norm": 0.41616639494895935,
|
| 6343 |
-
"learning_rate": 2.1658670322138452e-07,
|
| 6344 |
-
"loss": 1.9505,
|
| 6345 |
-
"step": 904
|
| 6346 |
-
},
|
| 6347 |
-
{
|
| 6348 |
-
"epoch": 1.7645625152327566,
|
| 6349 |
-
"grad_norm": 0.472650408744812,
|
| 6350 |
-
"learning_rate": 2.162440027416038e-07,
|
| 6351 |
-
"loss": 2.0594,
|
| 6352 |
-
"step": 905
|
| 6353 |
-
},
|
| 6354 |
-
{
|
| 6355 |
-
"epoch": 1.7665123080672678,
|
| 6356 |
-
"grad_norm": 0.4756447374820709,
|
| 6357 |
-
"learning_rate": 2.1590130226182316e-07,
|
| 6358 |
-
"loss": 1.9695,
|
| 6359 |
-
"step": 906
|
| 6360 |
-
},
|
| 6361 |
-
{
|
| 6362 |
-
"epoch": 1.7684621009017791,
|
| 6363 |
-
"grad_norm": 0.44738152623176575,
|
| 6364 |
-
"learning_rate": 2.155586017820425e-07,
|
| 6365 |
-
"loss": 2.0771,
|
| 6366 |
-
"step": 907
|
| 6367 |
-
},
|
| 6368 |
-
{
|
| 6369 |
-
"epoch": 1.7704118937362905,
|
| 6370 |
-
"grad_norm": 0.4602157771587372,
|
| 6371 |
-
"learning_rate": 2.152159013022618e-07,
|
| 6372 |
-
"loss": 2.0813,
|
| 6373 |
-
"step": 908
|
| 6374 |
-
},
|
| 6375 |
-
{
|
| 6376 |
-
"epoch": 1.7723616865708018,
|
| 6377 |
-
"grad_norm": 0.46765050292015076,
|
| 6378 |
-
"learning_rate": 2.1487320082248115e-07,
|
| 6379 |
-
"loss": 2.0801,
|
| 6380 |
-
"step": 909
|
| 6381 |
-
},
|
| 6382 |
-
{
|
| 6383 |
-
"epoch": 1.7743114794053132,
|
| 6384 |
-
"grad_norm": 0.4703747034072876,
|
| 6385 |
-
"learning_rate": 2.145305003427005e-07,
|
| 6386 |
-
"loss": 2.0093,
|
| 6387 |
-
"step": 910
|
| 6388 |
-
},
|
| 6389 |
-
{
|
| 6390 |
-
"epoch": 1.7762612722398246,
|
| 6391 |
-
"grad_norm": 0.48457059264183044,
|
| 6392 |
-
"learning_rate": 2.141877998629198e-07,
|
| 6393 |
-
"loss": 2.0528,
|
| 6394 |
-
"step": 911
|
| 6395 |
-
},
|
| 6396 |
-
{
|
| 6397 |
-
"epoch": 1.778211065074336,
|
| 6398 |
-
"grad_norm": 0.478710412979126,
|
| 6399 |
-
"learning_rate": 2.1384509938313914e-07,
|
| 6400 |
-
"loss": 2.1099,
|
| 6401 |
-
"step": 912
|
| 6402 |
-
},
|
| 6403 |
-
{
|
| 6404 |
-
"epoch": 1.780160857908847,
|
| 6405 |
-
"grad_norm": 0.4458109438419342,
|
| 6406 |
-
"learning_rate": 2.1350239890335843e-07,
|
| 6407 |
-
"loss": 2.0592,
|
| 6408 |
-
"step": 913
|
| 6409 |
-
},
|
| 6410 |
-
{
|
| 6411 |
-
"epoch": 1.7821106507433586,
|
| 6412 |
-
"grad_norm": 0.4474625885486603,
|
| 6413 |
-
"learning_rate": 2.1315969842357778e-07,
|
| 6414 |
-
"loss": 2.0055,
|
| 6415 |
-
"step": 914
|
| 6416 |
-
},
|
| 6417 |
-
{
|
| 6418 |
-
"epoch": 1.7840604435778697,
|
| 6419 |
-
"grad_norm": 0.4586813151836395,
|
| 6420 |
-
"learning_rate": 2.1281699794379713e-07,
|
| 6421 |
-
"loss": 2.0131,
|
| 6422 |
-
"step": 915
|
| 6423 |
-
},
|
| 6424 |
-
{
|
| 6425 |
-
"epoch": 1.7860102364123813,
|
| 6426 |
-
"grad_norm": 0.45083218812942505,
|
| 6427 |
-
"learning_rate": 2.1247429746401642e-07,
|
| 6428 |
-
"loss": 2.0437,
|
| 6429 |
-
"step": 916
|
| 6430 |
-
},
|
| 6431 |
-
{
|
| 6432 |
-
"epoch": 1.7879600292468925,
|
| 6433 |
-
"grad_norm": 0.44078171253204346,
|
| 6434 |
-
"learning_rate": 2.1213159698423577e-07,
|
| 6435 |
-
"loss": 1.9792,
|
| 6436 |
-
"step": 917
|
| 6437 |
-
},
|
| 6438 |
-
{
|
| 6439 |
-
"epoch": 1.7899098220814038,
|
| 6440 |
-
"grad_norm": 0.4346940219402313,
|
| 6441 |
-
"learning_rate": 2.117888965044551e-07,
|
| 6442 |
-
"loss": 1.9933,
|
| 6443 |
-
"step": 918
|
| 6444 |
-
},
|
| 6445 |
-
{
|
| 6446 |
-
"epoch": 1.7918596149159152,
|
| 6447 |
-
"grad_norm": 0.45846906304359436,
|
| 6448 |
-
"learning_rate": 2.114461960246744e-07,
|
| 6449 |
-
"loss": 1.9682,
|
| 6450 |
-
"step": 919
|
| 6451 |
-
},
|
| 6452 |
-
{
|
| 6453 |
-
"epoch": 1.7938094077504265,
|
| 6454 |
-
"grad_norm": 0.4335155785083771,
|
| 6455 |
-
"learning_rate": 2.1110349554489376e-07,
|
| 6456 |
-
"loss": 2.03,
|
| 6457 |
-
"step": 920
|
| 6458 |
-
},
|
| 6459 |
-
{
|
| 6460 |
-
"epoch": 1.7957592005849379,
|
| 6461 |
-
"grad_norm": 0.4618023633956909,
|
| 6462 |
-
"learning_rate": 2.1076079506511308e-07,
|
| 6463 |
-
"loss": 2.0966,
|
| 6464 |
-
"step": 921
|
| 6465 |
-
},
|
| 6466 |
-
{
|
| 6467 |
-
"epoch": 1.7977089934194492,
|
| 6468 |
-
"grad_norm": 0.46044906973838806,
|
| 6469 |
-
"learning_rate": 2.104180945853324e-07,
|
| 6470 |
-
"loss": 2.0873,
|
| 6471 |
-
"step": 922
|
| 6472 |
-
},
|
| 6473 |
-
{
|
| 6474 |
-
"epoch": 1.7996587862539606,
|
| 6475 |
-
"grad_norm": 0.4635170102119446,
|
| 6476 |
-
"learning_rate": 2.1007539410555175e-07,
|
| 6477 |
-
"loss": 1.9897,
|
| 6478 |
-
"step": 923
|
| 6479 |
-
},
|
| 6480 |
-
{
|
| 6481 |
-
"epoch": 1.8016085790884717,
|
| 6482 |
-
"grad_norm": 0.4335494637489319,
|
| 6483 |
-
"learning_rate": 2.0973269362577107e-07,
|
| 6484 |
-
"loss": 2.0228,
|
| 6485 |
-
"step": 924
|
| 6486 |
-
},
|
| 6487 |
-
{
|
| 6488 |
-
"epoch": 1.8035583719229833,
|
| 6489 |
-
"grad_norm": 0.44605642557144165,
|
| 6490 |
-
"learning_rate": 2.093899931459904e-07,
|
| 6491 |
-
"loss": 2.0561,
|
| 6492 |
-
"step": 925
|
| 6493 |
-
},
|
| 6494 |
-
{
|
| 6495 |
-
"epoch": 1.8055081647574944,
|
| 6496 |
-
"grad_norm": 0.4611765146255493,
|
| 6497 |
-
"learning_rate": 2.090472926662097e-07,
|
| 6498 |
-
"loss": 2.0329,
|
| 6499 |
-
"step": 926
|
| 6500 |
-
},
|
| 6501 |
-
{
|
| 6502 |
-
"epoch": 1.807457957592006,
|
| 6503 |
-
"grad_norm": 0.443036288022995,
|
| 6504 |
-
"learning_rate": 2.0870459218642906e-07,
|
| 6505 |
-
"loss": 1.9565,
|
| 6506 |
-
"step": 927
|
| 6507 |
-
},
|
| 6508 |
-
{
|
| 6509 |
-
"epoch": 1.8094077504265171,
|
| 6510 |
-
"grad_norm": 0.4552265405654907,
|
| 6511 |
-
"learning_rate": 2.0836189170664838e-07,
|
| 6512 |
-
"loss": 2.0842,
|
| 6513 |
-
"step": 928
|
| 6514 |
-
},
|
| 6515 |
-
{
|
| 6516 |
-
"epoch": 1.8113575432610285,
|
| 6517 |
-
"grad_norm": 0.41511160135269165,
|
| 6518 |
-
"learning_rate": 2.080191912268677e-07,
|
| 6519 |
-
"loss": 2.0043,
|
| 6520 |
-
"step": 929
|
| 6521 |
-
},
|
| 6522 |
-
{
|
| 6523 |
-
"epoch": 1.8133073360955398,
|
| 6524 |
-
"grad_norm": 0.44421470165252686,
|
| 6525 |
-
"learning_rate": 2.0767649074708705e-07,
|
| 6526 |
-
"loss": 2.0433,
|
| 6527 |
-
"step": 930
|
| 6528 |
-
},
|
| 6529 |
-
{
|
| 6530 |
-
"epoch": 1.8152571289300512,
|
| 6531 |
-
"grad_norm": 0.43709036707878113,
|
| 6532 |
-
"learning_rate": 2.0733379026730634e-07,
|
| 6533 |
-
"loss": 2.0405,
|
| 6534 |
-
"step": 931
|
| 6535 |
-
},
|
| 6536 |
-
{
|
| 6537 |
-
"epoch": 1.8172069217645626,
|
| 6538 |
-
"grad_norm": 0.429074227809906,
|
| 6539 |
-
"learning_rate": 2.069910897875257e-07,
|
| 6540 |
-
"loss": 1.964,
|
| 6541 |
-
"step": 932
|
| 6542 |
-
},
|
| 6543 |
-
{
|
| 6544 |
-
"epoch": 1.8191567145990737,
|
| 6545 |
-
"grad_norm": 0.4392930269241333,
|
| 6546 |
-
"learning_rate": 2.0664838930774504e-07,
|
| 6547 |
-
"loss": 1.9819,
|
| 6548 |
-
"step": 933
|
| 6549 |
-
},
|
| 6550 |
-
{
|
| 6551 |
-
"epoch": 1.8211065074335853,
|
| 6552 |
-
"grad_norm": 0.41590166091918945,
|
| 6553 |
-
"learning_rate": 2.0630568882796433e-07,
|
| 6554 |
-
"loss": 1.9821,
|
| 6555 |
-
"step": 934
|
| 6556 |
-
},
|
| 6557 |
-
{
|
| 6558 |
-
"epoch": 1.8230563002680964,
|
| 6559 |
-
"grad_norm": 0.445362389087677,
|
| 6560 |
-
"learning_rate": 2.0596298834818368e-07,
|
| 6561 |
-
"loss": 2.092,
|
| 6562 |
-
"step": 935
|
| 6563 |
-
},
|
| 6564 |
-
{
|
| 6565 |
-
"epoch": 1.825006093102608,
|
| 6566 |
-
"grad_norm": 0.43674713373184204,
|
| 6567 |
-
"learning_rate": 2.0562028786840303e-07,
|
| 6568 |
-
"loss": 2.0371,
|
| 6569 |
-
"step": 936
|
| 6570 |
-
},
|
| 6571 |
-
{
|
| 6572 |
-
"epoch": 1.826955885937119,
|
| 6573 |
-
"grad_norm": 0.4520663022994995,
|
| 6574 |
-
"learning_rate": 2.0527758738862232e-07,
|
| 6575 |
-
"loss": 2.0329,
|
| 6576 |
-
"step": 937
|
| 6577 |
-
},
|
| 6578 |
-
{
|
| 6579 |
-
"epoch": 1.8289056787716305,
|
| 6580 |
-
"grad_norm": 0.4744395613670349,
|
| 6581 |
-
"learning_rate": 2.0493488690884167e-07,
|
| 6582 |
-
"loss": 2.0828,
|
| 6583 |
-
"step": 938
|
| 6584 |
-
},
|
| 6585 |
-
{
|
| 6586 |
-
"epoch": 1.8308554716061418,
|
| 6587 |
-
"grad_norm": 0.45714208483695984,
|
| 6588 |
-
"learning_rate": 2.04592186429061e-07,
|
| 6589 |
-
"loss": 2.017,
|
| 6590 |
-
"step": 939
|
| 6591 |
-
},
|
| 6592 |
-
{
|
| 6593 |
-
"epoch": 1.8328052644406532,
|
| 6594 |
-
"grad_norm": 0.4604392647743225,
|
| 6595 |
-
"learning_rate": 2.042494859492803e-07,
|
| 6596 |
-
"loss": 1.9813,
|
| 6597 |
-
"step": 940
|
| 6598 |
-
},
|
| 6599 |
-
{
|
| 6600 |
-
"epoch": 1.8347550572751645,
|
| 6601 |
-
"grad_norm": 0.43890222907066345,
|
| 6602 |
-
"learning_rate": 2.0390678546949966e-07,
|
| 6603 |
-
"loss": 1.9902,
|
| 6604 |
-
"step": 941
|
| 6605 |
-
},
|
| 6606 |
-
{
|
| 6607 |
-
"epoch": 1.8367048501096759,
|
| 6608 |
-
"grad_norm": 0.44383513927459717,
|
| 6609 |
-
"learning_rate": 2.0356408498971898e-07,
|
| 6610 |
-
"loss": 2.0434,
|
| 6611 |
-
"step": 942
|
| 6612 |
-
},
|
| 6613 |
-
{
|
| 6614 |
-
"epoch": 1.8386546429441872,
|
| 6615 |
-
"grad_norm": 0.43706512451171875,
|
| 6616 |
-
"learning_rate": 2.032213845099383e-07,
|
| 6617 |
-
"loss": 2.052,
|
| 6618 |
-
"step": 943
|
| 6619 |
-
},
|
| 6620 |
-
{
|
| 6621 |
-
"epoch": 1.8406044357786984,
|
| 6622 |
-
"grad_norm": 0.427843302488327,
|
| 6623 |
-
"learning_rate": 2.0287868403015762e-07,
|
| 6624 |
-
"loss": 1.8841,
|
| 6625 |
-
"step": 944
|
| 6626 |
-
},
|
| 6627 |
-
{
|
| 6628 |
-
"epoch": 1.84255422861321,
|
| 6629 |
-
"grad_norm": 0.4639602601528168,
|
| 6630 |
-
"learning_rate": 2.0253598355037697e-07,
|
| 6631 |
-
"loss": 2.0831,
|
| 6632 |
-
"step": 945
|
| 6633 |
-
},
|
| 6634 |
-
{
|
| 6635 |
-
"epoch": 1.844504021447721,
|
| 6636 |
-
"grad_norm": 0.44139614701271057,
|
| 6637 |
-
"learning_rate": 2.021932830705963e-07,
|
| 6638 |
-
"loss": 1.9867,
|
| 6639 |
-
"step": 946
|
| 6640 |
-
},
|
| 6641 |
-
{
|
| 6642 |
-
"epoch": 1.8464538142822327,
|
| 6643 |
-
"grad_norm": 0.4408351182937622,
|
| 6644 |
-
"learning_rate": 2.018505825908156e-07,
|
| 6645 |
-
"loss": 2.0199,
|
| 6646 |
-
"step": 947
|
| 6647 |
-
},
|
| 6648 |
-
{
|
| 6649 |
-
"epoch": 1.8484036071167438,
|
| 6650 |
-
"grad_norm": 0.49647897481918335,
|
| 6651 |
-
"learning_rate": 2.0150788211103496e-07,
|
| 6652 |
-
"loss": 2.0877,
|
| 6653 |
-
"step": 948
|
| 6654 |
-
},
|
| 6655 |
-
{
|
| 6656 |
-
"epoch": 1.8503533999512551,
|
| 6657 |
-
"grad_norm": 0.46033725142478943,
|
| 6658 |
-
"learning_rate": 2.0116518163125428e-07,
|
| 6659 |
-
"loss": 2.0584,
|
| 6660 |
-
"step": 949
|
| 6661 |
-
},
|
| 6662 |
-
{
|
| 6663 |
-
"epoch": 1.8523031927857665,
|
| 6664 |
-
"grad_norm": 0.4471881687641144,
|
| 6665 |
-
"learning_rate": 2.008224811514736e-07,
|
| 6666 |
-
"loss": 1.9694,
|
| 6667 |
-
"step": 950
|
| 6668 |
-
},
|
| 6669 |
-
{
|
| 6670 |
-
"epoch": 1.8542529856202778,
|
| 6671 |
-
"grad_norm": 0.435660183429718,
|
| 6672 |
-
"learning_rate": 2.0047978067169295e-07,
|
| 6673 |
-
"loss": 2.0025,
|
| 6674 |
-
"step": 951
|
| 6675 |
-
},
|
| 6676 |
-
{
|
| 6677 |
-
"epoch": 1.8562027784547892,
|
| 6678 |
-
"grad_norm": 0.4504587650299072,
|
| 6679 |
-
"learning_rate": 2.0013708019191224e-07,
|
| 6680 |
-
"loss": 2.0403,
|
| 6681 |
-
"step": 952
|
| 6682 |
-
},
|
| 6683 |
-
{
|
| 6684 |
-
"epoch": 1.8581525712893006,
|
| 6685 |
-
"grad_norm": 0.446451336145401,
|
| 6686 |
-
"learning_rate": 1.997943797121316e-07,
|
| 6687 |
-
"loss": 1.9817,
|
| 6688 |
-
"step": 953
|
| 6689 |
-
},
|
| 6690 |
-
{
|
| 6691 |
-
"epoch": 1.860102364123812,
|
| 6692 |
-
"grad_norm": 0.46191105246543884,
|
| 6693 |
-
"learning_rate": 1.9945167923235094e-07,
|
| 6694 |
-
"loss": 2.0329,
|
| 6695 |
-
"step": 954
|
| 6696 |
-
},
|
| 6697 |
-
{
|
| 6698 |
-
"epoch": 1.862052156958323,
|
| 6699 |
-
"grad_norm": 0.4477747976779938,
|
| 6700 |
-
"learning_rate": 1.9910897875257023e-07,
|
| 6701 |
-
"loss": 2.0113,
|
| 6702 |
-
"step": 955
|
| 6703 |
-
},
|
| 6704 |
-
{
|
| 6705 |
-
"epoch": 1.8640019497928346,
|
| 6706 |
-
"grad_norm": 0.46400219202041626,
|
| 6707 |
-
"learning_rate": 1.9876627827278958e-07,
|
| 6708 |
-
"loss": 2.0142,
|
| 6709 |
-
"step": 956
|
| 6710 |
-
},
|
| 6711 |
-
{
|
| 6712 |
-
"epoch": 1.8659517426273458,
|
| 6713 |
-
"grad_norm": 0.45763564109802246,
|
| 6714 |
-
"learning_rate": 1.984235777930089e-07,
|
| 6715 |
-
"loss": 2.0555,
|
| 6716 |
-
"step": 957
|
| 6717 |
-
},
|
| 6718 |
-
{
|
| 6719 |
-
"epoch": 1.8679015354618573,
|
| 6720 |
-
"grad_norm": 0.4603627920150757,
|
| 6721 |
-
"learning_rate": 1.9808087731322822e-07,
|
| 6722 |
-
"loss": 2.0022,
|
| 6723 |
-
"step": 958
|
| 6724 |
-
},
|
| 6725 |
-
{
|
| 6726 |
-
"epoch": 1.8698513282963685,
|
| 6727 |
-
"grad_norm": 0.5134696364402771,
|
| 6728 |
-
"learning_rate": 1.9773817683344757e-07,
|
| 6729 |
-
"loss": 2.0396,
|
| 6730 |
-
"step": 959
|
| 6731 |
-
},
|
| 6732 |
-
{
|
| 6733 |
-
"epoch": 1.8718011211308798,
|
| 6734 |
-
"grad_norm": 0.46097123622894287,
|
| 6735 |
-
"learning_rate": 1.973954763536669e-07,
|
| 6736 |
-
"loss": 2.0887,
|
| 6737 |
-
"step": 960
|
| 6738 |
-
},
|
| 6739 |
-
{
|
| 6740 |
-
"epoch": 1.8737509139653912,
|
| 6741 |
-
"grad_norm": 0.45269545912742615,
|
| 6742 |
-
"learning_rate": 1.970527758738862e-07,
|
| 6743 |
-
"loss": 2.0184,
|
| 6744 |
-
"step": 961
|
| 6745 |
-
},
|
| 6746 |
-
{
|
| 6747 |
-
"epoch": 1.8757007067999025,
|
| 6748 |
-
"grad_norm": 0.463885635137558,
|
| 6749 |
-
"learning_rate": 1.9671007539410553e-07,
|
| 6750 |
-
"loss": 2.0701,
|
| 6751 |
-
"step": 962
|
| 6752 |
-
},
|
| 6753 |
-
{
|
| 6754 |
-
"epoch": 1.8776504996344139,
|
| 6755 |
-
"grad_norm": 0.4765574634075165,
|
| 6756 |
-
"learning_rate": 1.9636737491432488e-07,
|
| 6757 |
-
"loss": 1.9951,
|
| 6758 |
-
"step": 963
|
| 6759 |
-
},
|
| 6760 |
-
{
|
| 6761 |
-
"epoch": 1.879600292468925,
|
| 6762 |
-
"grad_norm": 0.48183631896972656,
|
| 6763 |
-
"learning_rate": 1.960246744345442e-07,
|
| 6764 |
-
"loss": 2.0723,
|
| 6765 |
-
"step": 964
|
| 6766 |
-
},
|
| 6767 |
-
{
|
| 6768 |
-
"epoch": 1.8815500853034366,
|
| 6769 |
-
"grad_norm": 0.44266360998153687,
|
| 6770 |
-
"learning_rate": 1.9568197395476352e-07,
|
| 6771 |
-
"loss": 2.0134,
|
| 6772 |
-
"step": 965
|
| 6773 |
-
},
|
| 6774 |
-
{
|
| 6775 |
-
"epoch": 1.8834998781379477,
|
| 6776 |
-
"grad_norm": 0.4508133828639984,
|
| 6777 |
-
"learning_rate": 1.9533927347498287e-07,
|
| 6778 |
-
"loss": 1.9951,
|
| 6779 |
-
"step": 966
|
| 6780 |
-
},
|
| 6781 |
-
{
|
| 6782 |
-
"epoch": 1.8854496709724593,
|
| 6783 |
-
"grad_norm": 0.4255620539188385,
|
| 6784 |
-
"learning_rate": 1.949965729952022e-07,
|
| 6785 |
-
"loss": 1.9663,
|
| 6786 |
-
"step": 967
|
| 6787 |
-
},
|
| 6788 |
-
{
|
| 6789 |
-
"epoch": 1.8873994638069704,
|
| 6790 |
-
"grad_norm": 0.45423394441604614,
|
| 6791 |
-
"learning_rate": 1.946538725154215e-07,
|
| 6792 |
-
"loss": 2.0072,
|
| 6793 |
-
"step": 968
|
| 6794 |
-
},
|
| 6795 |
-
{
|
| 6796 |
-
"epoch": 1.8893492566414818,
|
| 6797 |
-
"grad_norm": 0.4226663112640381,
|
| 6798 |
-
"learning_rate": 1.9431117203564086e-07,
|
| 6799 |
-
"loss": 1.9598,
|
| 6800 |
-
"step": 969
|
| 6801 |
-
},
|
| 6802 |
-
{
|
| 6803 |
-
"epoch": 1.8912990494759931,
|
| 6804 |
-
"grad_norm": 0.47366762161254883,
|
| 6805 |
-
"learning_rate": 1.9396847155586015e-07,
|
| 6806 |
-
"loss": 1.9927,
|
| 6807 |
-
"step": 970
|
| 6808 |
-
},
|
| 6809 |
-
{
|
| 6810 |
-
"epoch": 1.8932488423105045,
|
| 6811 |
-
"grad_norm": 0.44758790731430054,
|
| 6812 |
-
"learning_rate": 1.936257710760795e-07,
|
| 6813 |
-
"loss": 1.9628,
|
| 6814 |
-
"step": 971
|
| 6815 |
-
},
|
| 6816 |
-
{
|
| 6817 |
-
"epoch": 1.8951986351450159,
|
| 6818 |
-
"grad_norm": 0.48197463154792786,
|
| 6819 |
-
"learning_rate": 1.9328307059629885e-07,
|
| 6820 |
-
"loss": 2.1004,
|
| 6821 |
-
"step": 972
|
| 6822 |
-
},
|
| 6823 |
-
{
|
| 6824 |
-
"epoch": 1.8971484279795272,
|
| 6825 |
-
"grad_norm": 0.4538448750972748,
|
| 6826 |
-
"learning_rate": 1.9294037011651814e-07,
|
| 6827 |
-
"loss": 2.0199,
|
| 6828 |
-
"step": 973
|
| 6829 |
-
},
|
| 6830 |
-
{
|
| 6831 |
-
"epoch": 1.8990982208140386,
|
| 6832 |
-
"grad_norm": 0.47362738847732544,
|
| 6833 |
-
"learning_rate": 1.925976696367375e-07,
|
| 6834 |
-
"loss": 2.0746,
|
| 6835 |
-
"step": 974
|
| 6836 |
-
},
|
| 6837 |
-
{
|
| 6838 |
-
"epoch": 1.9010480136485497,
|
| 6839 |
-
"grad_norm": 0.47095638513565063,
|
| 6840 |
-
"learning_rate": 1.922549691569568e-07,
|
| 6841 |
-
"loss": 1.9897,
|
| 6842 |
-
"step": 975
|
| 6843 |
-
},
|
| 6844 |
-
{
|
| 6845 |
-
"epoch": 1.9029978064830613,
|
| 6846 |
-
"grad_norm": 0.4763641059398651,
|
| 6847 |
-
"learning_rate": 1.9191226867717613e-07,
|
| 6848 |
-
"loss": 2.0156,
|
| 6849 |
-
"step": 976
|
| 6850 |
-
},
|
| 6851 |
-
{
|
| 6852 |
-
"epoch": 1.9049475993175724,
|
| 6853 |
-
"grad_norm": 0.4224942922592163,
|
| 6854 |
-
"learning_rate": 1.9156956819739548e-07,
|
| 6855 |
-
"loss": 2.0114,
|
| 6856 |
-
"step": 977
|
| 6857 |
-
},
|
| 6858 |
-
{
|
| 6859 |
-
"epoch": 1.906897392152084,
|
| 6860 |
-
"grad_norm": 0.44930440187454224,
|
| 6861 |
-
"learning_rate": 1.912268677176148e-07,
|
| 6862 |
-
"loss": 2.0121,
|
| 6863 |
-
"step": 978
|
| 6864 |
-
},
|
| 6865 |
-
{
|
| 6866 |
-
"epoch": 1.9088471849865951,
|
| 6867 |
-
"grad_norm": 0.45916110277175903,
|
| 6868 |
-
"learning_rate": 1.9088416723783412e-07,
|
| 6869 |
-
"loss": 2.0053,
|
| 6870 |
-
"step": 979
|
| 6871 |
-
},
|
| 6872 |
-
{
|
| 6873 |
-
"epoch": 1.9107969778211065,
|
| 6874 |
-
"grad_norm": 0.42759600281715393,
|
| 6875 |
-
"learning_rate": 1.9054146675805347e-07,
|
| 6876 |
-
"loss": 2.0109,
|
| 6877 |
-
"step": 980
|
| 6878 |
-
},
|
| 6879 |
-
{
|
| 6880 |
-
"epoch": 1.9127467706556178,
|
| 6881 |
-
"grad_norm": 0.49347975850105286,
|
| 6882 |
-
"learning_rate": 1.901987662782728e-07,
|
| 6883 |
-
"loss": 2.0657,
|
| 6884 |
-
"step": 981
|
| 6885 |
-
},
|
| 6886 |
-
{
|
| 6887 |
-
"epoch": 1.9146965634901292,
|
| 6888 |
-
"grad_norm": 0.4315294027328491,
|
| 6889 |
-
"learning_rate": 1.898560657984921e-07,
|
| 6890 |
-
"loss": 1.9473,
|
| 6891 |
-
"step": 982
|
| 6892 |
-
},
|
| 6893 |
-
{
|
| 6894 |
-
"epoch": 1.9166463563246405,
|
| 6895 |
-
"grad_norm": 0.42915600538253784,
|
| 6896 |
-
"learning_rate": 1.8951336531871143e-07,
|
| 6897 |
-
"loss": 1.9958,
|
| 6898 |
-
"step": 983
|
| 6899 |
-
},
|
| 6900 |
-
{
|
| 6901 |
-
"epoch": 1.9185961491591519,
|
| 6902 |
-
"grad_norm": 0.48152124881744385,
|
| 6903 |
-
"learning_rate": 1.8917066483893078e-07,
|
| 6904 |
-
"loss": 2.0815,
|
| 6905 |
-
"step": 984
|
| 6906 |
-
},
|
| 6907 |
-
{
|
| 6908 |
-
"epoch": 1.9205459419936632,
|
| 6909 |
-
"grad_norm": 0.44423532485961914,
|
| 6910 |
-
"learning_rate": 1.888279643591501e-07,
|
| 6911 |
-
"loss": 2.0227,
|
| 6912 |
-
"step": 985
|
| 6913 |
-
},
|
| 6914 |
-
{
|
| 6915 |
-
"epoch": 1.9224957348281744,
|
| 6916 |
-
"grad_norm": 0.4499359130859375,
|
| 6917 |
-
"learning_rate": 1.8848526387936942e-07,
|
| 6918 |
-
"loss": 1.961,
|
| 6919 |
-
"step": 986
|
| 6920 |
-
},
|
| 6921 |
-
{
|
| 6922 |
-
"epoch": 1.924445527662686,
|
| 6923 |
-
"grad_norm": 0.4560549855232239,
|
| 6924 |
-
"learning_rate": 1.8814256339958877e-07,
|
| 6925 |
-
"loss": 2.03,
|
| 6926 |
-
"step": 987
|
| 6927 |
-
},
|
| 6928 |
-
{
|
| 6929 |
-
"epoch": 1.926395320497197,
|
| 6930 |
-
"grad_norm": 0.48396381735801697,
|
| 6931 |
-
"learning_rate": 1.8779986291980806e-07,
|
| 6932 |
-
"loss": 1.985,
|
| 6933 |
-
"step": 988
|
| 6934 |
-
},
|
| 6935 |
-
{
|
| 6936 |
-
"epoch": 1.9283451133317087,
|
| 6937 |
-
"grad_norm": 0.456910103559494,
|
| 6938 |
-
"learning_rate": 1.874571624400274e-07,
|
| 6939 |
-
"loss": 1.9802,
|
| 6940 |
-
"step": 989
|
| 6941 |
-
},
|
| 6942 |
-
{
|
| 6943 |
-
"epoch": 1.9302949061662198,
|
| 6944 |
-
"grad_norm": 0.46041303873062134,
|
| 6945 |
-
"learning_rate": 1.8711446196024676e-07,
|
| 6946 |
-
"loss": 1.9507,
|
| 6947 |
-
"step": 990
|
| 6948 |
-
},
|
| 6949 |
-
{
|
| 6950 |
-
"epoch": 1.9322446990007311,
|
| 6951 |
-
"grad_norm": 0.4496663510799408,
|
| 6952 |
-
"learning_rate": 1.8677176148046605e-07,
|
| 6953 |
-
"loss": 2.0329,
|
| 6954 |
-
"step": 991
|
| 6955 |
-
},
|
| 6956 |
-
{
|
| 6957 |
-
"epoch": 1.9341944918352425,
|
| 6958 |
-
"grad_norm": 0.4381345212459564,
|
| 6959 |
-
"learning_rate": 1.864290610006854e-07,
|
| 6960 |
-
"loss": 1.9643,
|
| 6961 |
-
"step": 992
|
| 6962 |
-
},
|
| 6963 |
-
{
|
| 6964 |
-
"epoch": 1.9361442846697539,
|
| 6965 |
-
"grad_norm": 0.43699464201927185,
|
| 6966 |
-
"learning_rate": 1.8608636052090475e-07,
|
| 6967 |
-
"loss": 2.026,
|
| 6968 |
-
"step": 993
|
| 6969 |
-
},
|
| 6970 |
-
{
|
| 6971 |
-
"epoch": 1.9380940775042652,
|
| 6972 |
-
"grad_norm": 0.4496040344238281,
|
| 6973 |
-
"learning_rate": 1.8574366004112404e-07,
|
| 6974 |
-
"loss": 1.9318,
|
| 6975 |
-
"step": 994
|
| 6976 |
-
},
|
| 6977 |
-
{
|
| 6978 |
-
"epoch": 1.9400438703387763,
|
| 6979 |
-
"grad_norm": 0.45028945803642273,
|
| 6980 |
-
"learning_rate": 1.854009595613434e-07,
|
| 6981 |
-
"loss": 2.0254,
|
| 6982 |
-
"step": 995
|
| 6983 |
-
},
|
| 6984 |
-
{
|
| 6985 |
-
"epoch": 1.941993663173288,
|
| 6986 |
-
"grad_norm": 0.46241873502731323,
|
| 6987 |
-
"learning_rate": 1.8505825908156268e-07,
|
| 6988 |
-
"loss": 2.0224,
|
| 6989 |
-
"step": 996
|
| 6990 |
-
},
|
| 6991 |
-
{
|
| 6992 |
-
"epoch": 1.943943456007799,
|
| 6993 |
-
"grad_norm": 0.4494277238845825,
|
| 6994 |
-
"learning_rate": 1.8471555860178203e-07,
|
| 6995 |
-
"loss": 2.0734,
|
| 6996 |
-
"step": 997
|
| 6997 |
-
},
|
| 6998 |
-
{
|
| 6999 |
-
"epoch": 1.9458932488423106,
|
| 7000 |
-
"grad_norm": 0.44225579500198364,
|
| 7001 |
-
"learning_rate": 1.8437285812200138e-07,
|
| 7002 |
-
"loss": 2.0548,
|
| 7003 |
-
"step": 998
|
| 7004 |
-
},
|
| 7005 |
-
{
|
| 7006 |
-
"epoch": 1.9478430416768218,
|
| 7007 |
-
"grad_norm": 0.4850820004940033,
|
| 7008 |
-
"learning_rate": 1.8403015764222067e-07,
|
| 7009 |
-
"loss": 1.9961,
|
| 7010 |
-
"step": 999
|
| 7011 |
-
},
|
| 7012 |
-
{
|
| 7013 |
-
"epoch": 1.9497928345113331,
|
| 7014 |
-
"grad_norm": 0.46442610025405884,
|
| 7015 |
-
"learning_rate": 1.8368745716244002e-07,
|
| 7016 |
-
"loss": 1.9777,
|
| 7017 |
-
"step": 1000
|
| 7018 |
-
},
|
| 7019 |
-
{
|
| 7020 |
-
"epoch": 1.9517426273458445,
|
| 7021 |
-
"grad_norm": 0.457109272480011,
|
| 7022 |
-
"learning_rate": 1.8334475668265934e-07,
|
| 7023 |
-
"loss": 2.0949,
|
| 7024 |
-
"step": 1001
|
| 7025 |
-
},
|
| 7026 |
-
{
|
| 7027 |
-
"epoch": 1.9536924201803558,
|
| 7028 |
-
"grad_norm": 0.4514349699020386,
|
| 7029 |
-
"learning_rate": 1.8300205620287866e-07,
|
| 7030 |
-
"loss": 2.0933,
|
| 7031 |
-
"step": 1002
|
| 7032 |
-
},
|
| 7033 |
-
{
|
| 7034 |
-
"epoch": 1.9556422130148672,
|
| 7035 |
-
"grad_norm": 0.4601777195930481,
|
| 7036 |
-
"learning_rate": 1.82659355723098e-07,
|
| 7037 |
-
"loss": 1.9975,
|
| 7038 |
-
"step": 1003
|
| 7039 |
-
},
|
| 7040 |
-
{
|
| 7041 |
-
"epoch": 1.9575920058493785,
|
| 7042 |
-
"grad_norm": 0.4604569673538208,
|
| 7043 |
-
"learning_rate": 1.8231665524331733e-07,
|
| 7044 |
-
"loss": 2.0364,
|
| 7045 |
-
"step": 1004
|
| 7046 |
-
},
|
| 7047 |
-
{
|
| 7048 |
-
"epoch": 1.95954179868389,
|
| 7049 |
-
"grad_norm": 0.4434170424938202,
|
| 7050 |
-
"learning_rate": 1.8197395476353665e-07,
|
| 7051 |
-
"loss": 1.9835,
|
| 7052 |
-
"step": 1005
|
| 7053 |
-
},
|
| 7054 |
-
{
|
| 7055 |
-
"epoch": 1.961491591518401,
|
| 7056 |
-
"grad_norm": 0.45063334703445435,
|
| 7057 |
-
"learning_rate": 1.81631254283756e-07,
|
| 7058 |
-
"loss": 1.9904,
|
| 7059 |
-
"step": 1006
|
| 7060 |
-
},
|
| 7061 |
-
{
|
| 7062 |
-
"epoch": 1.9634413843529126,
|
| 7063 |
-
"grad_norm": 0.45276153087615967,
|
| 7064 |
-
"learning_rate": 1.8128855380397532e-07,
|
| 7065 |
-
"loss": 2.021,
|
| 7066 |
-
"step": 1007
|
| 7067 |
-
},
|
| 7068 |
-
{
|
| 7069 |
-
"epoch": 1.9653911771874237,
|
| 7070 |
-
"grad_norm": 0.44774502515792847,
|
| 7071 |
-
"learning_rate": 1.8094585332419464e-07,
|
| 7072 |
-
"loss": 2.0024,
|
| 7073 |
-
"step": 1008
|
| 7074 |
-
},
|
| 7075 |
-
{
|
| 7076 |
-
"epoch": 1.9673409700219353,
|
| 7077 |
-
"grad_norm": 0.43734362721443176,
|
| 7078 |
-
"learning_rate": 1.8060315284441396e-07,
|
| 7079 |
-
"loss": 2.0261,
|
| 7080 |
-
"step": 1009
|
| 7081 |
-
},
|
| 7082 |
-
{
|
| 7083 |
-
"epoch": 1.9692907628564464,
|
| 7084 |
-
"grad_norm": 0.45293501019477844,
|
| 7085 |
-
"learning_rate": 1.802604523646333e-07,
|
| 7086 |
-
"loss": 2.0781,
|
| 7087 |
-
"step": 1010
|
| 7088 |
-
},
|
| 7089 |
-
{
|
| 7090 |
-
"epoch": 1.9712405556909578,
|
| 7091 |
-
"grad_norm": 0.4538004994392395,
|
| 7092 |
-
"learning_rate": 1.7991775188485263e-07,
|
| 7093 |
-
"loss": 2.0081,
|
| 7094 |
-
"step": 1011
|
| 7095 |
-
},
|
| 7096 |
-
{
|
| 7097 |
-
"epoch": 1.9731903485254692,
|
| 7098 |
-
"grad_norm": 0.45042964816093445,
|
| 7099 |
-
"learning_rate": 1.7957505140507195e-07,
|
| 7100 |
-
"loss": 2.0121,
|
| 7101 |
-
"step": 1012
|
| 7102 |
-
},
|
| 7103 |
-
{
|
| 7104 |
-
"epoch": 1.9751401413599805,
|
| 7105 |
-
"grad_norm": 0.4721399247646332,
|
| 7106 |
-
"learning_rate": 1.792323509252913e-07,
|
| 7107 |
-
"loss": 2.0071,
|
| 7108 |
-
"step": 1013
|
| 7109 |
-
},
|
| 7110 |
-
{
|
| 7111 |
-
"epoch": 1.9770899341944919,
|
| 7112 |
-
"grad_norm": 0.4297287166118622,
|
| 7113 |
-
"learning_rate": 1.788896504455106e-07,
|
| 7114 |
-
"loss": 2.0213,
|
| 7115 |
-
"step": 1014
|
| 7116 |
-
},
|
| 7117 |
-
{
|
| 7118 |
-
"epoch": 1.9790397270290032,
|
| 7119 |
-
"grad_norm": 0.4454828202724457,
|
| 7120 |
-
"learning_rate": 1.7854694996572994e-07,
|
| 7121 |
-
"loss": 2.0093,
|
| 7122 |
-
"step": 1015
|
| 7123 |
-
},
|
| 7124 |
-
{
|
| 7125 |
-
"epoch": 1.9809895198635146,
|
| 7126 |
-
"grad_norm": 0.4550788700580597,
|
| 7127 |
-
"learning_rate": 1.782042494859493e-07,
|
| 7128 |
-
"loss": 2.0599,
|
| 7129 |
-
"step": 1016
|
| 7130 |
-
},
|
| 7131 |
-
{
|
| 7132 |
-
"epoch": 1.9829393126980257,
|
| 7133 |
-
"grad_norm": 0.44854849576950073,
|
| 7134 |
-
"learning_rate": 1.7786154900616858e-07,
|
| 7135 |
-
"loss": 2.0262,
|
| 7136 |
-
"step": 1017
|
| 7137 |
-
},
|
| 7138 |
-
{
|
| 7139 |
-
"epoch": 1.9848891055325373,
|
| 7140 |
-
"grad_norm": 0.4477459192276001,
|
| 7141 |
-
"learning_rate": 1.7751884852638793e-07,
|
| 7142 |
-
"loss": 1.9533,
|
| 7143 |
-
"step": 1018
|
| 7144 |
-
},
|
| 7145 |
-
{
|
| 7146 |
-
"epoch": 1.9868388983670484,
|
| 7147 |
-
"grad_norm": 0.43663471937179565,
|
| 7148 |
-
"learning_rate": 1.7717614804660728e-07,
|
| 7149 |
-
"loss": 2.0122,
|
| 7150 |
-
"step": 1019
|
| 7151 |
-
},
|
| 7152 |
-
{
|
| 7153 |
-
"epoch": 1.98878869120156,
|
| 7154 |
-
"grad_norm": 0.45281800627708435,
|
| 7155 |
-
"learning_rate": 1.7683344756682657e-07,
|
| 7156 |
-
"loss": 2.0711,
|
| 7157 |
-
"step": 1020
|
| 7158 |
-
},
|
| 7159 |
-
{
|
| 7160 |
-
"epoch": 1.9907384840360711,
|
| 7161 |
-
"grad_norm": 0.44143861532211304,
|
| 7162 |
-
"learning_rate": 1.7649074708704592e-07,
|
| 7163 |
-
"loss": 2.0198,
|
| 7164 |
-
"step": 1021
|
| 7165 |
-
},
|
| 7166 |
-
{
|
| 7167 |
-
"epoch": 1.9926882768705825,
|
| 7168 |
-
"grad_norm": 0.4464763402938843,
|
| 7169 |
-
"learning_rate": 1.7614804660726524e-07,
|
| 7170 |
-
"loss": 2.0117,
|
| 7171 |
-
"step": 1022
|
| 7172 |
-
},
|
| 7173 |
-
{
|
| 7174 |
-
"epoch": 1.9946380697050938,
|
| 7175 |
-
"grad_norm": 0.42707762122154236,
|
| 7176 |
-
"learning_rate": 1.7580534612748456e-07,
|
| 7177 |
-
"loss": 1.9629,
|
| 7178 |
-
"step": 1023
|
| 7179 |
-
},
|
| 7180 |
-
{
|
| 7181 |
-
"epoch": 1.9965878625396052,
|
| 7182 |
-
"grad_norm": 0.4683617949485779,
|
| 7183 |
-
"learning_rate": 1.754626456477039e-07,
|
| 7184 |
-
"loss": 2.0467,
|
| 7185 |
-
"step": 1024
|
| 7186 |
-
},
|
| 7187 |
-
{
|
| 7188 |
-
"epoch": 1.9985376553741165,
|
| 7189 |
-
"grad_norm": 0.4215565025806427,
|
| 7190 |
-
"learning_rate": 1.7511994516792323e-07,
|
| 7191 |
-
"loss": 1.9545,
|
| 7192 |
-
"step": 1025
|
| 7193 |
-
},
|
| 7194 |
-
{
|
| 7195 |
-
"epoch": 1.9985376553741165,
|
| 7196 |
-
"eval_loss": 2.0196783542633057,
|
| 7197 |
-
"eval_runtime": 480.5583,
|
| 7198 |
-
"eval_samples_per_second": 1.294,
|
| 7199 |
-
"eval_steps_per_second": 0.325,
|
| 7200 |
-
"step": 1025
|
| 7201 |
}
|
| 7202 |
],
|
| 7203 |
"logging_steps": 1,
|
|
@@ -7217,7 +3618,7 @@
|
|
| 7217 |
"attributes": {}
|
| 7218 |
}
|
| 7219 |
},
|
| 7220 |
-
"total_flos":
|
| 7221 |
"train_batch_size": 2,
|
| 7222 |
"trial_name": null,
|
| 7223 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 2.046032667160034,
|
| 3 |
+
"best_model_checkpoint": "/home/sunggeunan/data/ICL/outputs/lora/SKIML-ICL_mrqa_nq_v3/Meta-Llama-3-8B-Instruct-unanswerable-5Q-0U-0C-qa_first/checkpoint-512",
|
| 4 |
+
"epoch": 0.9982939312698026,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 512,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 3599 |
"eval_samples_per_second": 1.293,
|
| 3600 |
"eval_steps_per_second": 0.324,
|
| 3601 |
"step": 512
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3602 |
}
|
| 3603 |
],
|
| 3604 |
"logging_steps": 1,
|
|
|
|
| 3618 |
"attributes": {}
|
| 3619 |
}
|
| 3620 |
},
|
| 3621 |
+
"total_flos": 1.5142103496146289e+18,
|
| 3622 |
"train_batch_size": 2,
|
| 3623 |
"trial_name": null,
|
| 3624 |
"trial_params": null
|