Training in progress, step 30000, checkpoint
Browse files- last-checkpoint/adapter_model.safetensors +1 -1
- last-checkpoint/global_step30000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step30000/mp_rank_00_model_states.pt +3 -0
- last-checkpoint/latest +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/trainer_state.json +1403 -3
last-checkpoint/adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 42002584
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:68c58d4316cbc14ab7cac78de3b40dc31048d94feea3c3258056c9246529c02f
|
3 |
size 42002584
|
last-checkpoint/global_step30000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8fb0e8c2240d1e6a36c6f2b0bc4c7c8a30636f02e0ae999e4ffa546318ee6777
|
3 |
+
size 251710672
|
last-checkpoint/global_step30000/mp_rank_00_model_states.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d28de1c4a1926e05a13af93c413d15c685991639ed23a59950e0f523327947af
|
3 |
+
size 153747385
|
last-checkpoint/latest
CHANGED
@@ -1 +1 @@
|
|
1 |
-
|
|
|
1 |
+
global_step30000
|
last-checkpoint/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 sha256:81410e8c68ff09be250a30bb210a84a3afd289539e330cebd1667ccee91efaa6
|
3 |
size 14244
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
5 |
"eval_steps": 1000,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -2814,6 +2814,1406 @@
|
|
2814 |
"learning_rate": 0.00019428739592797673,
|
2815 |
"loss": 1.3316,
|
2816 |
"step": 20000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2817 |
}
|
2818 |
],
|
2819 |
"logging_steps": 50,
|
@@ -2833,7 +4233,7 @@
|
|
2833 |
"attributes": {}
|
2834 |
}
|
2835 |
},
|
2836 |
-
"total_flos":
|
2837 |
"train_batch_size": 2,
|
2838 |
"trial_name": null,
|
2839 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.8957095512495148,
|
5 |
"eval_steps": 1000,
|
6 |
+
"global_step": 30000,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
2814 |
"learning_rate": 0.00019428739592797673,
|
2815 |
"loss": 1.3316,
|
2816 |
"step": 20000
|
2817 |
+
},
|
2818 |
+
{
|
2819 |
+
"epoch": 0.5986325500850924,
|
2820 |
+
"grad_norm": 4.055521488189697,
|
2821 |
+
"learning_rate": 0.00019427311013205792,
|
2822 |
+
"loss": 1.3784,
|
2823 |
+
"step": 20050
|
2824 |
+
},
|
2825 |
+
{
|
2826 |
+
"epoch": 0.6001253993371749,
|
2827 |
+
"grad_norm": 4.933177471160889,
|
2828 |
+
"learning_rate": 0.00019425882433613906,
|
2829 |
+
"loss": 1.3352,
|
2830 |
+
"step": 20100
|
2831 |
+
},
|
2832 |
+
{
|
2833 |
+
"epoch": 0.6016182485892575,
|
2834 |
+
"grad_norm": 3.8867061138153076,
|
2835 |
+
"learning_rate": 0.00019424453854022025,
|
2836 |
+
"loss": 1.4169,
|
2837 |
+
"step": 20150
|
2838 |
+
},
|
2839 |
+
{
|
2840 |
+
"epoch": 0.60311109784134,
|
2841 |
+
"grad_norm": 3.364475727081299,
|
2842 |
+
"learning_rate": 0.0001942302527443014,
|
2843 |
+
"loss": 1.3767,
|
2844 |
+
"step": 20200
|
2845 |
+
},
|
2846 |
+
{
|
2847 |
+
"epoch": 0.6046039470934225,
|
2848 |
+
"grad_norm": 3.48152232170105,
|
2849 |
+
"learning_rate": 0.00019421596694838258,
|
2850 |
+
"loss": 1.3659,
|
2851 |
+
"step": 20250
|
2852 |
+
},
|
2853 |
+
{
|
2854 |
+
"epoch": 0.6060967963455051,
|
2855 |
+
"grad_norm": 3.3658649921417236,
|
2856 |
+
"learning_rate": 0.00019420168115246374,
|
2857 |
+
"loss": 1.3745,
|
2858 |
+
"step": 20300
|
2859 |
+
},
|
2860 |
+
{
|
2861 |
+
"epoch": 0.6075896455975875,
|
2862 |
+
"grad_norm": 4.441917896270752,
|
2863 |
+
"learning_rate": 0.0001941873953565449,
|
2864 |
+
"loss": 1.3827,
|
2865 |
+
"step": 20350
|
2866 |
+
},
|
2867 |
+
{
|
2868 |
+
"epoch": 0.60908249484967,
|
2869 |
+
"grad_norm": 5.014800548553467,
|
2870 |
+
"learning_rate": 0.00019417310956062607,
|
2871 |
+
"loss": 1.3792,
|
2872 |
+
"step": 20400
|
2873 |
+
},
|
2874 |
+
{
|
2875 |
+
"epoch": 0.6105753441017526,
|
2876 |
+
"grad_norm": 5.472316265106201,
|
2877 |
+
"learning_rate": 0.00019415882376470724,
|
2878 |
+
"loss": 1.394,
|
2879 |
+
"step": 20450
|
2880 |
+
},
|
2881 |
+
{
|
2882 |
+
"epoch": 0.6120681933538351,
|
2883 |
+
"grad_norm": 5.35073184967041,
|
2884 |
+
"learning_rate": 0.0001941445379687884,
|
2885 |
+
"loss": 1.401,
|
2886 |
+
"step": 20500
|
2887 |
+
},
|
2888 |
+
{
|
2889 |
+
"epoch": 0.6135610426059177,
|
2890 |
+
"grad_norm": 4.284445762634277,
|
2891 |
+
"learning_rate": 0.00019413025217286957,
|
2892 |
+
"loss": 1.4354,
|
2893 |
+
"step": 20550
|
2894 |
+
},
|
2895 |
+
{
|
2896 |
+
"epoch": 0.6150538918580002,
|
2897 |
+
"grad_norm": 3.561774492263794,
|
2898 |
+
"learning_rate": 0.00019411596637695073,
|
2899 |
+
"loss": 1.3295,
|
2900 |
+
"step": 20600
|
2901 |
+
},
|
2902 |
+
{
|
2903 |
+
"epoch": 0.6165467411100827,
|
2904 |
+
"grad_norm": 4.479186534881592,
|
2905 |
+
"learning_rate": 0.0001941016805810319,
|
2906 |
+
"loss": 1.3629,
|
2907 |
+
"step": 20650
|
2908 |
+
},
|
2909 |
+
{
|
2910 |
+
"epoch": 0.6180395903621653,
|
2911 |
+
"grad_norm": 4.186618804931641,
|
2912 |
+
"learning_rate": 0.00019408739478511306,
|
2913 |
+
"loss": 1.3399,
|
2914 |
+
"step": 20700
|
2915 |
+
},
|
2916 |
+
{
|
2917 |
+
"epoch": 0.6195324396142478,
|
2918 |
+
"grad_norm": 3.589655637741089,
|
2919 |
+
"learning_rate": 0.00019407310898919425,
|
2920 |
+
"loss": 1.3516,
|
2921 |
+
"step": 20750
|
2922 |
+
},
|
2923 |
+
{
|
2924 |
+
"epoch": 0.6210252888663302,
|
2925 |
+
"grad_norm": 4.330646514892578,
|
2926 |
+
"learning_rate": 0.0001940588231932754,
|
2927 |
+
"loss": 1.3148,
|
2928 |
+
"step": 20800
|
2929 |
+
},
|
2930 |
+
{
|
2931 |
+
"epoch": 0.6225181381184128,
|
2932 |
+
"grad_norm": 6.324933052062988,
|
2933 |
+
"learning_rate": 0.00019404453739735658,
|
2934 |
+
"loss": 1.331,
|
2935 |
+
"step": 20850
|
2936 |
+
},
|
2937 |
+
{
|
2938 |
+
"epoch": 0.6240109873704953,
|
2939 |
+
"grad_norm": 4.652800559997559,
|
2940 |
+
"learning_rate": 0.00019403025160143772,
|
2941 |
+
"loss": 1.3604,
|
2942 |
+
"step": 20900
|
2943 |
+
},
|
2944 |
+
{
|
2945 |
+
"epoch": 0.6255038366225778,
|
2946 |
+
"grad_norm": 4.831106662750244,
|
2947 |
+
"learning_rate": 0.0001940159658055189,
|
2948 |
+
"loss": 1.3322,
|
2949 |
+
"step": 20950
|
2950 |
+
},
|
2951 |
+
{
|
2952 |
+
"epoch": 0.6269966858746604,
|
2953 |
+
"grad_norm": 6.0938920974731445,
|
2954 |
+
"learning_rate": 0.00019400168000960005,
|
2955 |
+
"loss": 1.4106,
|
2956 |
+
"step": 21000
|
2957 |
+
},
|
2958 |
+
{
|
2959 |
+
"epoch": 0.6284895351267429,
|
2960 |
+
"grad_norm": 4.424108028411865,
|
2961 |
+
"learning_rate": 0.00019398739421368124,
|
2962 |
+
"loss": 1.4001,
|
2963 |
+
"step": 21050
|
2964 |
+
},
|
2965 |
+
{
|
2966 |
+
"epoch": 0.6299823843788255,
|
2967 |
+
"grad_norm": 4.329803466796875,
|
2968 |
+
"learning_rate": 0.0001939731084177624,
|
2969 |
+
"loss": 1.4048,
|
2970 |
+
"step": 21100
|
2971 |
+
},
|
2972 |
+
{
|
2973 |
+
"epoch": 0.631475233630908,
|
2974 |
+
"grad_norm": 3.915818929672241,
|
2975 |
+
"learning_rate": 0.00019395882262184357,
|
2976 |
+
"loss": 1.3619,
|
2977 |
+
"step": 21150
|
2978 |
+
},
|
2979 |
+
{
|
2980 |
+
"epoch": 0.6329680828829904,
|
2981 |
+
"grad_norm": 3.9562571048736572,
|
2982 |
+
"learning_rate": 0.00019394453682592474,
|
2983 |
+
"loss": 1.3636,
|
2984 |
+
"step": 21200
|
2985 |
+
},
|
2986 |
+
{
|
2987 |
+
"epoch": 0.634460932135073,
|
2988 |
+
"grad_norm": 5.229249954223633,
|
2989 |
+
"learning_rate": 0.0001939302510300059,
|
2990 |
+
"loss": 1.3502,
|
2991 |
+
"step": 21250
|
2992 |
+
},
|
2993 |
+
{
|
2994 |
+
"epoch": 0.6359537813871555,
|
2995 |
+
"grad_norm": 4.149145126342773,
|
2996 |
+
"learning_rate": 0.00019391596523408706,
|
2997 |
+
"loss": 1.3603,
|
2998 |
+
"step": 21300
|
2999 |
+
},
|
3000 |
+
{
|
3001 |
+
"epoch": 0.637446630639238,
|
3002 |
+
"grad_norm": 4.068868637084961,
|
3003 |
+
"learning_rate": 0.00019390167943816823,
|
3004 |
+
"loss": 1.3945,
|
3005 |
+
"step": 21350
|
3006 |
+
},
|
3007 |
+
{
|
3008 |
+
"epoch": 0.6389394798913206,
|
3009 |
+
"grad_norm": 3.6808931827545166,
|
3010 |
+
"learning_rate": 0.0001938873936422494,
|
3011 |
+
"loss": 1.4015,
|
3012 |
+
"step": 21400
|
3013 |
+
},
|
3014 |
+
{
|
3015 |
+
"epoch": 0.6404323291434031,
|
3016 |
+
"grad_norm": 4.391795635223389,
|
3017 |
+
"learning_rate": 0.00019387310784633056,
|
3018 |
+
"loss": 1.3428,
|
3019 |
+
"step": 21450
|
3020 |
+
},
|
3021 |
+
{
|
3022 |
+
"epoch": 0.6419251783954856,
|
3023 |
+
"grad_norm": 5.109655380249023,
|
3024 |
+
"learning_rate": 0.00019385882205041172,
|
3025 |
+
"loss": 1.3874,
|
3026 |
+
"step": 21500
|
3027 |
+
},
|
3028 |
+
{
|
3029 |
+
"epoch": 0.6434180276475682,
|
3030 |
+
"grad_norm": 4.560513496398926,
|
3031 |
+
"learning_rate": 0.0001938445362544929,
|
3032 |
+
"loss": 1.3833,
|
3033 |
+
"step": 21550
|
3034 |
+
},
|
3035 |
+
{
|
3036 |
+
"epoch": 0.6449108768996507,
|
3037 |
+
"grad_norm": 3.6108620166778564,
|
3038 |
+
"learning_rate": 0.00019383025045857405,
|
3039 |
+
"loss": 1.3981,
|
3040 |
+
"step": 21600
|
3041 |
+
},
|
3042 |
+
{
|
3043 |
+
"epoch": 0.6464037261517332,
|
3044 |
+
"grad_norm": 4.100367546081543,
|
3045 |
+
"learning_rate": 0.00019381596466265522,
|
3046 |
+
"loss": 1.3139,
|
3047 |
+
"step": 21650
|
3048 |
+
},
|
3049 |
+
{
|
3050 |
+
"epoch": 0.6478965754038157,
|
3051 |
+
"grad_norm": 3.1513540744781494,
|
3052 |
+
"learning_rate": 0.00019380167886673638,
|
3053 |
+
"loss": 1.3359,
|
3054 |
+
"step": 21700
|
3055 |
+
},
|
3056 |
+
{
|
3057 |
+
"epoch": 0.6493894246558982,
|
3058 |
+
"grad_norm": 4.793807506561279,
|
3059 |
+
"learning_rate": 0.00019378739307081755,
|
3060 |
+
"loss": 1.4025,
|
3061 |
+
"step": 21750
|
3062 |
+
},
|
3063 |
+
{
|
3064 |
+
"epoch": 0.6508822739079808,
|
3065 |
+
"grad_norm": 4.403114318847656,
|
3066 |
+
"learning_rate": 0.0001937731072748987,
|
3067 |
+
"loss": 1.349,
|
3068 |
+
"step": 21800
|
3069 |
+
},
|
3070 |
+
{
|
3071 |
+
"epoch": 0.6523751231600633,
|
3072 |
+
"grad_norm": 4.445423603057861,
|
3073 |
+
"learning_rate": 0.00019375882147897988,
|
3074 |
+
"loss": 1.4641,
|
3075 |
+
"step": 21850
|
3076 |
+
},
|
3077 |
+
{
|
3078 |
+
"epoch": 0.6538679724121458,
|
3079 |
+
"grad_norm": 5.721547603607178,
|
3080 |
+
"learning_rate": 0.00019374453568306107,
|
3081 |
+
"loss": 1.3451,
|
3082 |
+
"step": 21900
|
3083 |
+
},
|
3084 |
+
{
|
3085 |
+
"epoch": 0.6553608216642284,
|
3086 |
+
"grad_norm": 4.411465644836426,
|
3087 |
+
"learning_rate": 0.0001937302498871422,
|
3088 |
+
"loss": 1.3893,
|
3089 |
+
"step": 21950
|
3090 |
+
},
|
3091 |
+
{
|
3092 |
+
"epoch": 0.6568536709163109,
|
3093 |
+
"grad_norm": 3.181412696838379,
|
3094 |
+
"learning_rate": 0.0001937159640912234,
|
3095 |
+
"loss": 1.385,
|
3096 |
+
"step": 22000
|
3097 |
+
},
|
3098 |
+
{
|
3099 |
+
"epoch": 0.6583465201683933,
|
3100 |
+
"grad_norm": 7.374556541442871,
|
3101 |
+
"learning_rate": 0.00019370167829530453,
|
3102 |
+
"loss": 1.4051,
|
3103 |
+
"step": 22050
|
3104 |
+
},
|
3105 |
+
{
|
3106 |
+
"epoch": 0.6598393694204759,
|
3107 |
+
"grad_norm": 4.452084064483643,
|
3108 |
+
"learning_rate": 0.00019368739249938573,
|
3109 |
+
"loss": 1.3389,
|
3110 |
+
"step": 22100
|
3111 |
+
},
|
3112 |
+
{
|
3113 |
+
"epoch": 0.6613322186725584,
|
3114 |
+
"grad_norm": 5.60634708404541,
|
3115 |
+
"learning_rate": 0.0001936731067034669,
|
3116 |
+
"loss": 1.3924,
|
3117 |
+
"step": 22150
|
3118 |
+
},
|
3119 |
+
{
|
3120 |
+
"epoch": 0.662825067924641,
|
3121 |
+
"grad_norm": 4.35741662979126,
|
3122 |
+
"learning_rate": 0.00019365882090754806,
|
3123 |
+
"loss": 1.3957,
|
3124 |
+
"step": 22200
|
3125 |
+
},
|
3126 |
+
{
|
3127 |
+
"epoch": 0.6643179171767235,
|
3128 |
+
"grad_norm": 4.173916816711426,
|
3129 |
+
"learning_rate": 0.00019364453511162922,
|
3130 |
+
"loss": 1.3669,
|
3131 |
+
"step": 22250
|
3132 |
+
},
|
3133 |
+
{
|
3134 |
+
"epoch": 0.665810766428806,
|
3135 |
+
"grad_norm": 3.8707377910614014,
|
3136 |
+
"learning_rate": 0.00019363024931571038,
|
3137 |
+
"loss": 1.3478,
|
3138 |
+
"step": 22300
|
3139 |
+
},
|
3140 |
+
{
|
3141 |
+
"epoch": 0.6673036156808886,
|
3142 |
+
"grad_norm": 4.765937328338623,
|
3143 |
+
"learning_rate": 0.00019361596351979155,
|
3144 |
+
"loss": 1.4007,
|
3145 |
+
"step": 22350
|
3146 |
+
},
|
3147 |
+
{
|
3148 |
+
"epoch": 0.6687964649329711,
|
3149 |
+
"grad_norm": 4.315809726715088,
|
3150 |
+
"learning_rate": 0.00019360167772387271,
|
3151 |
+
"loss": 1.3622,
|
3152 |
+
"step": 22400
|
3153 |
+
},
|
3154 |
+
{
|
3155 |
+
"epoch": 0.6702893141850536,
|
3156 |
+
"grad_norm": 5.977219104766846,
|
3157 |
+
"learning_rate": 0.00019358739192795388,
|
3158 |
+
"loss": 1.3836,
|
3159 |
+
"step": 22450
|
3160 |
+
},
|
3161 |
+
{
|
3162 |
+
"epoch": 0.6717821634371361,
|
3163 |
+
"grad_norm": 4.420149803161621,
|
3164 |
+
"learning_rate": 0.00019357310613203504,
|
3165 |
+
"loss": 1.3488,
|
3166 |
+
"step": 22500
|
3167 |
+
},
|
3168 |
+
{
|
3169 |
+
"epoch": 0.6732750126892186,
|
3170 |
+
"grad_norm": 3.9537293910980225,
|
3171 |
+
"learning_rate": 0.0001935588203361162,
|
3172 |
+
"loss": 1.3892,
|
3173 |
+
"step": 22550
|
3174 |
+
},
|
3175 |
+
{
|
3176 |
+
"epoch": 0.6747678619413011,
|
3177 |
+
"grad_norm": 5.468355655670166,
|
3178 |
+
"learning_rate": 0.0001935445345401974,
|
3179 |
+
"loss": 1.4475,
|
3180 |
+
"step": 22600
|
3181 |
+
},
|
3182 |
+
{
|
3183 |
+
"epoch": 0.6762607111933837,
|
3184 |
+
"grad_norm": 4.3148674964904785,
|
3185 |
+
"learning_rate": 0.00019353024874427854,
|
3186 |
+
"loss": 1.3866,
|
3187 |
+
"step": 22650
|
3188 |
+
},
|
3189 |
+
{
|
3190 |
+
"epoch": 0.6777535604454662,
|
3191 |
+
"grad_norm": 4.004809379577637,
|
3192 |
+
"learning_rate": 0.00019351596294835973,
|
3193 |
+
"loss": 1.3811,
|
3194 |
+
"step": 22700
|
3195 |
+
},
|
3196 |
+
{
|
3197 |
+
"epoch": 0.6792464096975488,
|
3198 |
+
"grad_norm": 4.404988765716553,
|
3199 |
+
"learning_rate": 0.00019350167715244087,
|
3200 |
+
"loss": 1.3648,
|
3201 |
+
"step": 22750
|
3202 |
+
},
|
3203 |
+
{
|
3204 |
+
"epoch": 0.6807392589496313,
|
3205 |
+
"grad_norm": 5.115052223205566,
|
3206 |
+
"learning_rate": 0.00019348739135652206,
|
3207 |
+
"loss": 1.3523,
|
3208 |
+
"step": 22800
|
3209 |
+
},
|
3210 |
+
{
|
3211 |
+
"epoch": 0.6822321082017138,
|
3212 |
+
"grad_norm": 5.083719730377197,
|
3213 |
+
"learning_rate": 0.0001934731055606032,
|
3214 |
+
"loss": 1.3978,
|
3215 |
+
"step": 22850
|
3216 |
+
},
|
3217 |
+
{
|
3218 |
+
"epoch": 0.6837249574537964,
|
3219 |
+
"grad_norm": 4.038282871246338,
|
3220 |
+
"learning_rate": 0.0001934588197646844,
|
3221 |
+
"loss": 1.2788,
|
3222 |
+
"step": 22900
|
3223 |
+
},
|
3224 |
+
{
|
3225 |
+
"epoch": 0.6852178067058788,
|
3226 |
+
"grad_norm": 5.438407897949219,
|
3227 |
+
"learning_rate": 0.00019344453396876555,
|
3228 |
+
"loss": 1.4181,
|
3229 |
+
"step": 22950
|
3230 |
+
},
|
3231 |
+
{
|
3232 |
+
"epoch": 0.6867106559579613,
|
3233 |
+
"grad_norm": 5.381191730499268,
|
3234 |
+
"learning_rate": 0.00019343024817284672,
|
3235 |
+
"loss": 1.4145,
|
3236 |
+
"step": 23000
|
3237 |
+
},
|
3238 |
+
{
|
3239 |
+
"epoch": 0.6882035052100439,
|
3240 |
+
"grad_norm": 3.183706283569336,
|
3241 |
+
"learning_rate": 0.00019341596237692788,
|
3242 |
+
"loss": 1.3271,
|
3243 |
+
"step": 23050
|
3244 |
+
},
|
3245 |
+
{
|
3246 |
+
"epoch": 0.6896963544621264,
|
3247 |
+
"grad_norm": 4.063404083251953,
|
3248 |
+
"learning_rate": 0.00019340167658100905,
|
3249 |
+
"loss": 1.3717,
|
3250 |
+
"step": 23100
|
3251 |
+
},
|
3252 |
+
{
|
3253 |
+
"epoch": 0.6911892037142089,
|
3254 |
+
"grad_norm": 5.677481651306152,
|
3255 |
+
"learning_rate": 0.0001933873907850902,
|
3256 |
+
"loss": 1.344,
|
3257 |
+
"step": 23150
|
3258 |
+
},
|
3259 |
+
{
|
3260 |
+
"epoch": 0.6926820529662915,
|
3261 |
+
"grad_norm": 5.376470565795898,
|
3262 |
+
"learning_rate": 0.00019337310498917138,
|
3263 |
+
"loss": 1.295,
|
3264 |
+
"step": 23200
|
3265 |
+
},
|
3266 |
+
{
|
3267 |
+
"epoch": 0.694174902218374,
|
3268 |
+
"grad_norm": 3.8844990730285645,
|
3269 |
+
"learning_rate": 0.00019335881919325254,
|
3270 |
+
"loss": 1.381,
|
3271 |
+
"step": 23250
|
3272 |
+
},
|
3273 |
+
{
|
3274 |
+
"epoch": 0.6956677514704566,
|
3275 |
+
"grad_norm": 5.285017013549805,
|
3276 |
+
"learning_rate": 0.0001933445333973337,
|
3277 |
+
"loss": 1.3505,
|
3278 |
+
"step": 23300
|
3279 |
+
},
|
3280 |
+
{
|
3281 |
+
"epoch": 0.697160600722539,
|
3282 |
+
"grad_norm": 4.133642196655273,
|
3283 |
+
"learning_rate": 0.00019333024760141487,
|
3284 |
+
"loss": 1.3604,
|
3285 |
+
"step": 23350
|
3286 |
+
},
|
3287 |
+
{
|
3288 |
+
"epoch": 0.6986534499746215,
|
3289 |
+
"grad_norm": 5.037447929382324,
|
3290 |
+
"learning_rate": 0.00019331596180549606,
|
3291 |
+
"loss": 1.3209,
|
3292 |
+
"step": 23400
|
3293 |
+
},
|
3294 |
+
{
|
3295 |
+
"epoch": 0.7001462992267041,
|
3296 |
+
"grad_norm": 4.916257858276367,
|
3297 |
+
"learning_rate": 0.0001933016760095772,
|
3298 |
+
"loss": 1.4084,
|
3299 |
+
"step": 23450
|
3300 |
+
},
|
3301 |
+
{
|
3302 |
+
"epoch": 0.7016391484787866,
|
3303 |
+
"grad_norm": 3.469505786895752,
|
3304 |
+
"learning_rate": 0.0001932873902136584,
|
3305 |
+
"loss": 1.3705,
|
3306 |
+
"step": 23500
|
3307 |
+
},
|
3308 |
+
{
|
3309 |
+
"epoch": 0.7031319977308691,
|
3310 |
+
"grad_norm": 3.624896287918091,
|
3311 |
+
"learning_rate": 0.00019327310441773953,
|
3312 |
+
"loss": 1.3507,
|
3313 |
+
"step": 23550
|
3314 |
+
},
|
3315 |
+
{
|
3316 |
+
"epoch": 0.7046248469829517,
|
3317 |
+
"grad_norm": 4.352174758911133,
|
3318 |
+
"learning_rate": 0.00019325881862182072,
|
3319 |
+
"loss": 1.3184,
|
3320 |
+
"step": 23600
|
3321 |
+
},
|
3322 |
+
{
|
3323 |
+
"epoch": 0.7061176962350342,
|
3324 |
+
"grad_norm": 6.4549150466918945,
|
3325 |
+
"learning_rate": 0.00019324453282590186,
|
3326 |
+
"loss": 1.4161,
|
3327 |
+
"step": 23650
|
3328 |
+
},
|
3329 |
+
{
|
3330 |
+
"epoch": 0.7076105454871167,
|
3331 |
+
"grad_norm": 6.221735000610352,
|
3332 |
+
"learning_rate": 0.00019323024702998305,
|
3333 |
+
"loss": 1.3425,
|
3334 |
+
"step": 23700
|
3335 |
+
},
|
3336 |
+
{
|
3337 |
+
"epoch": 0.7091033947391993,
|
3338 |
+
"grad_norm": 4.285811901092529,
|
3339 |
+
"learning_rate": 0.00019321596123406421,
|
3340 |
+
"loss": 1.3418,
|
3341 |
+
"step": 23750
|
3342 |
+
},
|
3343 |
+
{
|
3344 |
+
"epoch": 0.7105962439912817,
|
3345 |
+
"grad_norm": 4.533527374267578,
|
3346 |
+
"learning_rate": 0.00019320167543814538,
|
3347 |
+
"loss": 1.3723,
|
3348 |
+
"step": 23800
|
3349 |
+
},
|
3350 |
+
{
|
3351 |
+
"epoch": 0.7120890932433643,
|
3352 |
+
"grad_norm": 3.838109016418457,
|
3353 |
+
"learning_rate": 0.00019318738964222654,
|
3354 |
+
"loss": 1.3624,
|
3355 |
+
"step": 23850
|
3356 |
+
},
|
3357 |
+
{
|
3358 |
+
"epoch": 0.7135819424954468,
|
3359 |
+
"grad_norm": 3.8401126861572266,
|
3360 |
+
"learning_rate": 0.0001931731038463077,
|
3361 |
+
"loss": 1.4239,
|
3362 |
+
"step": 23900
|
3363 |
+
},
|
3364 |
+
{
|
3365 |
+
"epoch": 0.7150747917475293,
|
3366 |
+
"grad_norm": 5.9049787521362305,
|
3367 |
+
"learning_rate": 0.00019315881805038887,
|
3368 |
+
"loss": 1.3642,
|
3369 |
+
"step": 23950
|
3370 |
+
},
|
3371 |
+
{
|
3372 |
+
"epoch": 0.7165676409996119,
|
3373 |
+
"grad_norm": 5.397033214569092,
|
3374 |
+
"learning_rate": 0.00019314453225447004,
|
3375 |
+
"loss": 1.3298,
|
3376 |
+
"step": 24000
|
3377 |
+
},
|
3378 |
+
{
|
3379 |
+
"epoch": 0.7180604902516944,
|
3380 |
+
"grad_norm": 5.649387359619141,
|
3381 |
+
"learning_rate": 0.0001931302464585512,
|
3382 |
+
"loss": 1.4102,
|
3383 |
+
"step": 24050
|
3384 |
+
},
|
3385 |
+
{
|
3386 |
+
"epoch": 0.7195533395037769,
|
3387 |
+
"grad_norm": 5.697938919067383,
|
3388 |
+
"learning_rate": 0.00019311596066263237,
|
3389 |
+
"loss": 1.3765,
|
3390 |
+
"step": 24100
|
3391 |
+
},
|
3392 |
+
{
|
3393 |
+
"epoch": 0.7210461887558595,
|
3394 |
+
"grad_norm": 4.244998455047607,
|
3395 |
+
"learning_rate": 0.00019310167486671353,
|
3396 |
+
"loss": 1.3471,
|
3397 |
+
"step": 24150
|
3398 |
+
},
|
3399 |
+
{
|
3400 |
+
"epoch": 0.722539038007942,
|
3401 |
+
"grad_norm": 4.579226493835449,
|
3402 |
+
"learning_rate": 0.0001930873890707947,
|
3403 |
+
"loss": 1.3897,
|
3404 |
+
"step": 24200
|
3405 |
+
},
|
3406 |
+
{
|
3407 |
+
"epoch": 0.7240318872600244,
|
3408 |
+
"grad_norm": 8.80657958984375,
|
3409 |
+
"learning_rate": 0.00019307310327487586,
|
3410 |
+
"loss": 1.3628,
|
3411 |
+
"step": 24250
|
3412 |
+
},
|
3413 |
+
{
|
3414 |
+
"epoch": 0.725524736512107,
|
3415 |
+
"grad_norm": 5.42709493637085,
|
3416 |
+
"learning_rate": 0.00019305881747895703,
|
3417 |
+
"loss": 1.3587,
|
3418 |
+
"step": 24300
|
3419 |
+
},
|
3420 |
+
{
|
3421 |
+
"epoch": 0.7270175857641895,
|
3422 |
+
"grad_norm": 4.2680840492248535,
|
3423 |
+
"learning_rate": 0.0001930445316830382,
|
3424 |
+
"loss": 1.3414,
|
3425 |
+
"step": 24350
|
3426 |
+
},
|
3427 |
+
{
|
3428 |
+
"epoch": 0.7285104350162721,
|
3429 |
+
"grad_norm": 5.641107559204102,
|
3430 |
+
"learning_rate": 0.00019303024588711935,
|
3431 |
+
"loss": 1.4095,
|
3432 |
+
"step": 24400
|
3433 |
+
},
|
3434 |
+
{
|
3435 |
+
"epoch": 0.7300032842683546,
|
3436 |
+
"grad_norm": 5.29530143737793,
|
3437 |
+
"learning_rate": 0.00019301596009120052,
|
3438 |
+
"loss": 1.389,
|
3439 |
+
"step": 24450
|
3440 |
+
},
|
3441 |
+
{
|
3442 |
+
"epoch": 0.7314961335204371,
|
3443 |
+
"grad_norm": 4.110136032104492,
|
3444 |
+
"learning_rate": 0.00019300167429528168,
|
3445 |
+
"loss": 1.4641,
|
3446 |
+
"step": 24500
|
3447 |
+
},
|
3448 |
+
{
|
3449 |
+
"epoch": 0.7329889827725197,
|
3450 |
+
"grad_norm": 4.390302658081055,
|
3451 |
+
"learning_rate": 0.00019298738849936288,
|
3452 |
+
"loss": 1.4028,
|
3453 |
+
"step": 24550
|
3454 |
+
},
|
3455 |
+
{
|
3456 |
+
"epoch": 0.7344818320246022,
|
3457 |
+
"grad_norm": 3.717409372329712,
|
3458 |
+
"learning_rate": 0.000192973102703444,
|
3459 |
+
"loss": 1.436,
|
3460 |
+
"step": 24600
|
3461 |
+
},
|
3462 |
+
{
|
3463 |
+
"epoch": 0.7359746812766846,
|
3464 |
+
"grad_norm": 5.399808406829834,
|
3465 |
+
"learning_rate": 0.0001929588169075252,
|
3466 |
+
"loss": 1.2887,
|
3467 |
+
"step": 24650
|
3468 |
+
},
|
3469 |
+
{
|
3470 |
+
"epoch": 0.7374675305287672,
|
3471 |
+
"grad_norm": 5.027936935424805,
|
3472 |
+
"learning_rate": 0.00019294453111160634,
|
3473 |
+
"loss": 1.3591,
|
3474 |
+
"step": 24700
|
3475 |
+
},
|
3476 |
+
{
|
3477 |
+
"epoch": 0.7389603797808497,
|
3478 |
+
"grad_norm": 5.12235689163208,
|
3479 |
+
"learning_rate": 0.00019293024531568753,
|
3480 |
+
"loss": 1.4305,
|
3481 |
+
"step": 24750
|
3482 |
+
},
|
3483 |
+
{
|
3484 |
+
"epoch": 0.7404532290329322,
|
3485 |
+
"grad_norm": 4.445746898651123,
|
3486 |
+
"learning_rate": 0.0001929159595197687,
|
3487 |
+
"loss": 1.3609,
|
3488 |
+
"step": 24800
|
3489 |
+
},
|
3490 |
+
{
|
3491 |
+
"epoch": 0.7419460782850148,
|
3492 |
+
"grad_norm": 4.590288162231445,
|
3493 |
+
"learning_rate": 0.00019290167372384986,
|
3494 |
+
"loss": 1.3811,
|
3495 |
+
"step": 24850
|
3496 |
+
},
|
3497 |
+
{
|
3498 |
+
"epoch": 0.7434389275370973,
|
3499 |
+
"grad_norm": 3.841573476791382,
|
3500 |
+
"learning_rate": 0.00019288738792793103,
|
3501 |
+
"loss": 1.3632,
|
3502 |
+
"step": 24900
|
3503 |
+
},
|
3504 |
+
{
|
3505 |
+
"epoch": 0.7449317767891799,
|
3506 |
+
"grad_norm": 5.2155327796936035,
|
3507 |
+
"learning_rate": 0.0001928731021320122,
|
3508 |
+
"loss": 1.4422,
|
3509 |
+
"step": 24950
|
3510 |
+
},
|
3511 |
+
{
|
3512 |
+
"epoch": 0.7464246260412624,
|
3513 |
+
"grad_norm": 4.942720413208008,
|
3514 |
+
"learning_rate": 0.00019285881633609336,
|
3515 |
+
"loss": 1.3916,
|
3516 |
+
"step": 25000
|
3517 |
+
},
|
3518 |
+
{
|
3519 |
+
"epoch": 0.7479174752933448,
|
3520 |
+
"grad_norm": 5.292357921600342,
|
3521 |
+
"learning_rate": 0.00019284453054017452,
|
3522 |
+
"loss": 1.3048,
|
3523 |
+
"step": 25050
|
3524 |
+
},
|
3525 |
+
{
|
3526 |
+
"epoch": 0.7494103245454274,
|
3527 |
+
"grad_norm": 6.690375804901123,
|
3528 |
+
"learning_rate": 0.0001928302447442557,
|
3529 |
+
"loss": 1.3967,
|
3530 |
+
"step": 25100
|
3531 |
+
},
|
3532 |
+
{
|
3533 |
+
"epoch": 0.7509031737975099,
|
3534 |
+
"grad_norm": 4.590394496917725,
|
3535 |
+
"learning_rate": 0.00019281595894833685,
|
3536 |
+
"loss": 1.3925,
|
3537 |
+
"step": 25150
|
3538 |
+
},
|
3539 |
+
{
|
3540 |
+
"epoch": 0.7523960230495924,
|
3541 |
+
"grad_norm": 4.220889091491699,
|
3542 |
+
"learning_rate": 0.00019280167315241802,
|
3543 |
+
"loss": 1.3263,
|
3544 |
+
"step": 25200
|
3545 |
+
},
|
3546 |
+
{
|
3547 |
+
"epoch": 0.753888872301675,
|
3548 |
+
"grad_norm": 4.847371578216553,
|
3549 |
+
"learning_rate": 0.00019278738735649918,
|
3550 |
+
"loss": 1.4141,
|
3551 |
+
"step": 25250
|
3552 |
+
},
|
3553 |
+
{
|
3554 |
+
"epoch": 0.7553817215537575,
|
3555 |
+
"grad_norm": 5.392793655395508,
|
3556 |
+
"learning_rate": 0.00019277310156058035,
|
3557 |
+
"loss": 1.3147,
|
3558 |
+
"step": 25300
|
3559 |
+
},
|
3560 |
+
{
|
3561 |
+
"epoch": 0.75687457080584,
|
3562 |
+
"grad_norm": 4.261468887329102,
|
3563 |
+
"learning_rate": 0.00019275881576466154,
|
3564 |
+
"loss": 1.3102,
|
3565 |
+
"step": 25350
|
3566 |
+
},
|
3567 |
+
{
|
3568 |
+
"epoch": 0.7583674200579226,
|
3569 |
+
"grad_norm": 4.498802661895752,
|
3570 |
+
"learning_rate": 0.00019274452996874267,
|
3571 |
+
"loss": 1.3912,
|
3572 |
+
"step": 25400
|
3573 |
+
},
|
3574 |
+
{
|
3575 |
+
"epoch": 0.759860269310005,
|
3576 |
+
"grad_norm": 5.2235283851623535,
|
3577 |
+
"learning_rate": 0.00019273024417282387,
|
3578 |
+
"loss": 1.3375,
|
3579 |
+
"step": 25450
|
3580 |
+
},
|
3581 |
+
{
|
3582 |
+
"epoch": 0.7613531185620876,
|
3583 |
+
"grad_norm": 6.409016132354736,
|
3584 |
+
"learning_rate": 0.000192715958376905,
|
3585 |
+
"loss": 1.3339,
|
3586 |
+
"step": 25500
|
3587 |
+
},
|
3588 |
+
{
|
3589 |
+
"epoch": 0.7628459678141701,
|
3590 |
+
"grad_norm": 4.4392805099487305,
|
3591 |
+
"learning_rate": 0.0001927016725809862,
|
3592 |
+
"loss": 1.3599,
|
3593 |
+
"step": 25550
|
3594 |
+
},
|
3595 |
+
{
|
3596 |
+
"epoch": 0.7643388170662526,
|
3597 |
+
"grad_norm": 5.580776691436768,
|
3598 |
+
"learning_rate": 0.00019268738678506736,
|
3599 |
+
"loss": 1.3812,
|
3600 |
+
"step": 25600
|
3601 |
+
},
|
3602 |
+
{
|
3603 |
+
"epoch": 0.7658316663183352,
|
3604 |
+
"grad_norm": 5.473046779632568,
|
3605 |
+
"learning_rate": 0.00019267310098914852,
|
3606 |
+
"loss": 1.3582,
|
3607 |
+
"step": 25650
|
3608 |
+
},
|
3609 |
+
{
|
3610 |
+
"epoch": 0.7673245155704177,
|
3611 |
+
"grad_norm": 6.353076934814453,
|
3612 |
+
"learning_rate": 0.0001926588151932297,
|
3613 |
+
"loss": 1.3388,
|
3614 |
+
"step": 25700
|
3615 |
+
},
|
3616 |
+
{
|
3617 |
+
"epoch": 0.7688173648225002,
|
3618 |
+
"grad_norm": 4.247453689575195,
|
3619 |
+
"learning_rate": 0.00019264452939731085,
|
3620 |
+
"loss": 1.3745,
|
3621 |
+
"step": 25750
|
3622 |
+
},
|
3623 |
+
{
|
3624 |
+
"epoch": 0.7703102140745828,
|
3625 |
+
"grad_norm": 5.048892498016357,
|
3626 |
+
"learning_rate": 0.00019263024360139202,
|
3627 |
+
"loss": 1.405,
|
3628 |
+
"step": 25800
|
3629 |
+
},
|
3630 |
+
{
|
3631 |
+
"epoch": 0.7718030633266653,
|
3632 |
+
"grad_norm": 4.883440017700195,
|
3633 |
+
"learning_rate": 0.00019261595780547318,
|
3634 |
+
"loss": 1.4082,
|
3635 |
+
"step": 25850
|
3636 |
+
},
|
3637 |
+
{
|
3638 |
+
"epoch": 0.7732959125787477,
|
3639 |
+
"grad_norm": 4.221151828765869,
|
3640 |
+
"learning_rate": 0.00019260167200955435,
|
3641 |
+
"loss": 1.3766,
|
3642 |
+
"step": 25900
|
3643 |
+
},
|
3644 |
+
{
|
3645 |
+
"epoch": 0.7747887618308303,
|
3646 |
+
"grad_norm": 8.68738079071045,
|
3647 |
+
"learning_rate": 0.0001925873862136355,
|
3648 |
+
"loss": 1.4046,
|
3649 |
+
"step": 25950
|
3650 |
+
},
|
3651 |
+
{
|
3652 |
+
"epoch": 0.7762816110829128,
|
3653 |
+
"grad_norm": 4.479017734527588,
|
3654 |
+
"learning_rate": 0.00019257310041771668,
|
3655 |
+
"loss": 1.3452,
|
3656 |
+
"step": 26000
|
3657 |
+
},
|
3658 |
+
{
|
3659 |
+
"epoch": 0.7777744603349954,
|
3660 |
+
"grad_norm": 4.299473285675049,
|
3661 |
+
"learning_rate": 0.00019255881462179787,
|
3662 |
+
"loss": 1.3702,
|
3663 |
+
"step": 26050
|
3664 |
+
},
|
3665 |
+
{
|
3666 |
+
"epoch": 0.7792673095870779,
|
3667 |
+
"grad_norm": 5.272589683532715,
|
3668 |
+
"learning_rate": 0.000192544528825879,
|
3669 |
+
"loss": 1.3725,
|
3670 |
+
"step": 26100
|
3671 |
+
},
|
3672 |
+
{
|
3673 |
+
"epoch": 0.7807601588391604,
|
3674 |
+
"grad_norm": 4.644061088562012,
|
3675 |
+
"learning_rate": 0.0001925302430299602,
|
3676 |
+
"loss": 1.4143,
|
3677 |
+
"step": 26150
|
3678 |
+
},
|
3679 |
+
{
|
3680 |
+
"epoch": 0.782253008091243,
|
3681 |
+
"grad_norm": 4.7033185958862305,
|
3682 |
+
"learning_rate": 0.00019251595723404134,
|
3683 |
+
"loss": 1.3666,
|
3684 |
+
"step": 26200
|
3685 |
+
},
|
3686 |
+
{
|
3687 |
+
"epoch": 0.7837458573433255,
|
3688 |
+
"grad_norm": 3.5262560844421387,
|
3689 |
+
"learning_rate": 0.00019250167143812253,
|
3690 |
+
"loss": 1.4135,
|
3691 |
+
"step": 26250
|
3692 |
+
},
|
3693 |
+
{
|
3694 |
+
"epoch": 0.785238706595408,
|
3695 |
+
"grad_norm": 3.8599159717559814,
|
3696 |
+
"learning_rate": 0.00019248738564220367,
|
3697 |
+
"loss": 1.3258,
|
3698 |
+
"step": 26300
|
3699 |
+
},
|
3700 |
+
{
|
3701 |
+
"epoch": 0.7867315558474905,
|
3702 |
+
"grad_norm": 5.743364334106445,
|
3703 |
+
"learning_rate": 0.00019247309984628486,
|
3704 |
+
"loss": 1.3946,
|
3705 |
+
"step": 26350
|
3706 |
+
},
|
3707 |
+
{
|
3708 |
+
"epoch": 0.788224405099573,
|
3709 |
+
"grad_norm": 5.478078365325928,
|
3710 |
+
"learning_rate": 0.00019245881405036602,
|
3711 |
+
"loss": 1.3881,
|
3712 |
+
"step": 26400
|
3713 |
+
},
|
3714 |
+
{
|
3715 |
+
"epoch": 0.7897172543516555,
|
3716 |
+
"grad_norm": 5.912649631500244,
|
3717 |
+
"learning_rate": 0.0001924445282544472,
|
3718 |
+
"loss": 1.4139,
|
3719 |
+
"step": 26450
|
3720 |
+
},
|
3721 |
+
{
|
3722 |
+
"epoch": 0.7912101036037381,
|
3723 |
+
"grad_norm": 3.753570079803467,
|
3724 |
+
"learning_rate": 0.00019243024245852835,
|
3725 |
+
"loss": 1.3267,
|
3726 |
+
"step": 26500
|
3727 |
+
},
|
3728 |
+
{
|
3729 |
+
"epoch": 0.7927029528558206,
|
3730 |
+
"grad_norm": 4.155190944671631,
|
3731 |
+
"learning_rate": 0.00019241595666260952,
|
3732 |
+
"loss": 1.3539,
|
3733 |
+
"step": 26550
|
3734 |
+
},
|
3735 |
+
{
|
3736 |
+
"epoch": 0.7941958021079032,
|
3737 |
+
"grad_norm": 4.314638614654541,
|
3738 |
+
"learning_rate": 0.00019240167086669068,
|
3739 |
+
"loss": 1.3635,
|
3740 |
+
"step": 26600
|
3741 |
+
},
|
3742 |
+
{
|
3743 |
+
"epoch": 0.7956886513599857,
|
3744 |
+
"grad_norm": 4.177329063415527,
|
3745 |
+
"learning_rate": 0.00019238738507077185,
|
3746 |
+
"loss": 1.3642,
|
3747 |
+
"step": 26650
|
3748 |
+
},
|
3749 |
+
{
|
3750 |
+
"epoch": 0.7971815006120682,
|
3751 |
+
"grad_norm": 4.3052144050598145,
|
3752 |
+
"learning_rate": 0.000192373099274853,
|
3753 |
+
"loss": 1.3011,
|
3754 |
+
"step": 26700
|
3755 |
+
},
|
3756 |
+
{
|
3757 |
+
"epoch": 0.7986743498641508,
|
3758 |
+
"grad_norm": 5.800657272338867,
|
3759 |
+
"learning_rate": 0.00019235881347893417,
|
3760 |
+
"loss": 1.3247,
|
3761 |
+
"step": 26750
|
3762 |
+
},
|
3763 |
+
{
|
3764 |
+
"epoch": 0.8001671991162332,
|
3765 |
+
"grad_norm": 4.211179733276367,
|
3766 |
+
"learning_rate": 0.00019234452768301534,
|
3767 |
+
"loss": 1.3399,
|
3768 |
+
"step": 26800
|
3769 |
+
},
|
3770 |
+
{
|
3771 |
+
"epoch": 0.8016600483683157,
|
3772 |
+
"grad_norm": 4.8080315589904785,
|
3773 |
+
"learning_rate": 0.0001923302418870965,
|
3774 |
+
"loss": 1.3439,
|
3775 |
+
"step": 26850
|
3776 |
+
},
|
3777 |
+
{
|
3778 |
+
"epoch": 0.8031528976203983,
|
3779 |
+
"grad_norm": 5.03045129776001,
|
3780 |
+
"learning_rate": 0.00019231595609117767,
|
3781 |
+
"loss": 1.4009,
|
3782 |
+
"step": 26900
|
3783 |
+
},
|
3784 |
+
{
|
3785 |
+
"epoch": 0.8046457468724808,
|
3786 |
+
"grad_norm": 8.932162284851074,
|
3787 |
+
"learning_rate": 0.00019230167029525883,
|
3788 |
+
"loss": 1.3298,
|
3789 |
+
"step": 26950
|
3790 |
+
},
|
3791 |
+
{
|
3792 |
+
"epoch": 0.8061385961245633,
|
3793 |
+
"grad_norm": 5.651429176330566,
|
3794 |
+
"learning_rate": 0.00019228738449934,
|
3795 |
+
"loss": 1.4044,
|
3796 |
+
"step": 27000
|
3797 |
+
},
|
3798 |
+
{
|
3799 |
+
"epoch": 0.8076314453766459,
|
3800 |
+
"grad_norm": 4.428074836730957,
|
3801 |
+
"learning_rate": 0.00019227309870342116,
|
3802 |
+
"loss": 1.3988,
|
3803 |
+
"step": 27050
|
3804 |
+
},
|
3805 |
+
{
|
3806 |
+
"epoch": 0.8091242946287284,
|
3807 |
+
"grad_norm": 6.512781620025635,
|
3808 |
+
"learning_rate": 0.00019225881290750233,
|
3809 |
+
"loss": 1.3529,
|
3810 |
+
"step": 27100
|
3811 |
+
},
|
3812 |
+
{
|
3813 |
+
"epoch": 0.810617143880811,
|
3814 |
+
"grad_norm": 5.393406867980957,
|
3815 |
+
"learning_rate": 0.0001922445271115835,
|
3816 |
+
"loss": 1.3266,
|
3817 |
+
"step": 27150
|
3818 |
+
},
|
3819 |
+
{
|
3820 |
+
"epoch": 0.8121099931328934,
|
3821 |
+
"grad_norm": 3.74702525138855,
|
3822 |
+
"learning_rate": 0.00019223024131566468,
|
3823 |
+
"loss": 1.416,
|
3824 |
+
"step": 27200
|
3825 |
+
},
|
3826 |
+
{
|
3827 |
+
"epoch": 0.8136028423849759,
|
3828 |
+
"grad_norm": 3.856290817260742,
|
3829 |
+
"learning_rate": 0.00019221595551974582,
|
3830 |
+
"loss": 1.4082,
|
3831 |
+
"step": 27250
|
3832 |
+
},
|
3833 |
+
{
|
3834 |
+
"epoch": 0.8150956916370585,
|
3835 |
+
"grad_norm": 4.597848415374756,
|
3836 |
+
"learning_rate": 0.000192201669723827,
|
3837 |
+
"loss": 1.3751,
|
3838 |
+
"step": 27300
|
3839 |
+
},
|
3840 |
+
{
|
3841 |
+
"epoch": 0.816588540889141,
|
3842 |
+
"grad_norm": 3.875593662261963,
|
3843 |
+
"learning_rate": 0.00019218738392790815,
|
3844 |
+
"loss": 1.4024,
|
3845 |
+
"step": 27350
|
3846 |
+
},
|
3847 |
+
{
|
3848 |
+
"epoch": 0.8180813901412235,
|
3849 |
+
"grad_norm": 5.470495700836182,
|
3850 |
+
"learning_rate": 0.00019217309813198934,
|
3851 |
+
"loss": 1.394,
|
3852 |
+
"step": 27400
|
3853 |
+
},
|
3854 |
+
{
|
3855 |
+
"epoch": 0.8195742393933061,
|
3856 |
+
"grad_norm": 4.562834739685059,
|
3857 |
+
"learning_rate": 0.00019215881233607048,
|
3858 |
+
"loss": 1.3161,
|
3859 |
+
"step": 27450
|
3860 |
+
},
|
3861 |
+
{
|
3862 |
+
"epoch": 0.8210670886453886,
|
3863 |
+
"grad_norm": 4.183352470397949,
|
3864 |
+
"learning_rate": 0.00019214452654015167,
|
3865 |
+
"loss": 1.3427,
|
3866 |
+
"step": 27500
|
3867 |
+
},
|
3868 |
+
{
|
3869 |
+
"epoch": 0.8225599378974711,
|
3870 |
+
"grad_norm": 5.345236301422119,
|
3871 |
+
"learning_rate": 0.00019213024074423284,
|
3872 |
+
"loss": 1.3551,
|
3873 |
+
"step": 27550
|
3874 |
+
},
|
3875 |
+
{
|
3876 |
+
"epoch": 0.8240527871495537,
|
3877 |
+
"grad_norm": 5.217405319213867,
|
3878 |
+
"learning_rate": 0.000192115954948314,
|
3879 |
+
"loss": 1.3174,
|
3880 |
+
"step": 27600
|
3881 |
+
},
|
3882 |
+
{
|
3883 |
+
"epoch": 0.8255456364016361,
|
3884 |
+
"grad_norm": 4.486965179443359,
|
3885 |
+
"learning_rate": 0.00019210166915239517,
|
3886 |
+
"loss": 1.3931,
|
3887 |
+
"step": 27650
|
3888 |
+
},
|
3889 |
+
{
|
3890 |
+
"epoch": 0.8270384856537187,
|
3891 |
+
"grad_norm": 4.183075904846191,
|
3892 |
+
"learning_rate": 0.00019208738335647633,
|
3893 |
+
"loss": 1.3672,
|
3894 |
+
"step": 27700
|
3895 |
+
},
|
3896 |
+
{
|
3897 |
+
"epoch": 0.8285313349058012,
|
3898 |
+
"grad_norm": 4.916398525238037,
|
3899 |
+
"learning_rate": 0.0001920730975605575,
|
3900 |
+
"loss": 1.3993,
|
3901 |
+
"step": 27750
|
3902 |
+
},
|
3903 |
+
{
|
3904 |
+
"epoch": 0.8300241841578837,
|
3905 |
+
"grad_norm": 3.99765682220459,
|
3906 |
+
"learning_rate": 0.00019205881176463866,
|
3907 |
+
"loss": 1.362,
|
3908 |
+
"step": 27800
|
3909 |
+
},
|
3910 |
+
{
|
3911 |
+
"epoch": 0.8315170334099663,
|
3912 |
+
"grad_norm": 3.948981523513794,
|
3913 |
+
"learning_rate": 0.00019204452596871982,
|
3914 |
+
"loss": 1.3912,
|
3915 |
+
"step": 27850
|
3916 |
+
},
|
3917 |
+
{
|
3918 |
+
"epoch": 0.8330098826620488,
|
3919 |
+
"grad_norm": 6.524683475494385,
|
3920 |
+
"learning_rate": 0.000192030240172801,
|
3921 |
+
"loss": 1.3845,
|
3922 |
+
"step": 27900
|
3923 |
+
},
|
3924 |
+
{
|
3925 |
+
"epoch": 0.8345027319141313,
|
3926 |
+
"grad_norm": 4.598319053649902,
|
3927 |
+
"learning_rate": 0.00019201595437688215,
|
3928 |
+
"loss": 1.38,
|
3929 |
+
"step": 27950
|
3930 |
+
},
|
3931 |
+
{
|
3932 |
+
"epoch": 0.8359955811662139,
|
3933 |
+
"grad_norm": 4.287365436553955,
|
3934 |
+
"learning_rate": 0.00019200166858096335,
|
3935 |
+
"loss": 1.3687,
|
3936 |
+
"step": 28000
|
3937 |
+
},
|
3938 |
+
{
|
3939 |
+
"epoch": 0.8374884304182963,
|
3940 |
+
"grad_norm": 5.497170448303223,
|
3941 |
+
"learning_rate": 0.00019198738278504448,
|
3942 |
+
"loss": 1.4324,
|
3943 |
+
"step": 28050
|
3944 |
+
},
|
3945 |
+
{
|
3946 |
+
"epoch": 0.8389812796703788,
|
3947 |
+
"grad_norm": 3.9373042583465576,
|
3948 |
+
"learning_rate": 0.00019197309698912567,
|
3949 |
+
"loss": 1.3892,
|
3950 |
+
"step": 28100
|
3951 |
+
},
|
3952 |
+
{
|
3953 |
+
"epoch": 0.8404741289224614,
|
3954 |
+
"grad_norm": 5.157922267913818,
|
3955 |
+
"learning_rate": 0.0001919588111932068,
|
3956 |
+
"loss": 1.3764,
|
3957 |
+
"step": 28150
|
3958 |
+
},
|
3959 |
+
{
|
3960 |
+
"epoch": 0.8419669781745439,
|
3961 |
+
"grad_norm": 4.0024895668029785,
|
3962 |
+
"learning_rate": 0.000191944525397288,
|
3963 |
+
"loss": 1.3571,
|
3964 |
+
"step": 28200
|
3965 |
+
},
|
3966 |
+
{
|
3967 |
+
"epoch": 0.8434598274266265,
|
3968 |
+
"grad_norm": 4.894957542419434,
|
3969 |
+
"learning_rate": 0.00019193023960136917,
|
3970 |
+
"loss": 1.3997,
|
3971 |
+
"step": 28250
|
3972 |
+
},
|
3973 |
+
{
|
3974 |
+
"epoch": 0.844952676678709,
|
3975 |
+
"grad_norm": 4.829247951507568,
|
3976 |
+
"learning_rate": 0.00019191595380545033,
|
3977 |
+
"loss": 1.3751,
|
3978 |
+
"step": 28300
|
3979 |
+
},
|
3980 |
+
{
|
3981 |
+
"epoch": 0.8464455259307915,
|
3982 |
+
"grad_norm": 4.123593330383301,
|
3983 |
+
"learning_rate": 0.0001919016680095315,
|
3984 |
+
"loss": 1.4183,
|
3985 |
+
"step": 28350
|
3986 |
+
},
|
3987 |
+
{
|
3988 |
+
"epoch": 0.8479383751828741,
|
3989 |
+
"grad_norm": 4.012912273406982,
|
3990 |
+
"learning_rate": 0.00019188738221361266,
|
3991 |
+
"loss": 1.3757,
|
3992 |
+
"step": 28400
|
3993 |
+
},
|
3994 |
+
{
|
3995 |
+
"epoch": 0.8494312244349566,
|
3996 |
+
"grad_norm": 5.17979097366333,
|
3997 |
+
"learning_rate": 0.00019187309641769383,
|
3998 |
+
"loss": 1.3975,
|
3999 |
+
"step": 28450
|
4000 |
+
},
|
4001 |
+
{
|
4002 |
+
"epoch": 0.850924073687039,
|
4003 |
+
"grad_norm": 4.665642261505127,
|
4004 |
+
"learning_rate": 0.000191858810621775,
|
4005 |
+
"loss": 1.4539,
|
4006 |
+
"step": 28500
|
4007 |
+
},
|
4008 |
+
{
|
4009 |
+
"epoch": 0.8524169229391216,
|
4010 |
+
"grad_norm": 4.333798408508301,
|
4011 |
+
"learning_rate": 0.00019184452482585616,
|
4012 |
+
"loss": 1.3711,
|
4013 |
+
"step": 28550
|
4014 |
+
},
|
4015 |
+
{
|
4016 |
+
"epoch": 0.8539097721912041,
|
4017 |
+
"grad_norm": 4.485865116119385,
|
4018 |
+
"learning_rate": 0.00019183023902993732,
|
4019 |
+
"loss": 1.4299,
|
4020 |
+
"step": 28600
|
4021 |
+
},
|
4022 |
+
{
|
4023 |
+
"epoch": 0.8554026214432866,
|
4024 |
+
"grad_norm": 6.3318328857421875,
|
4025 |
+
"learning_rate": 0.00019181595323401849,
|
4026 |
+
"loss": 1.3621,
|
4027 |
+
"step": 28650
|
4028 |
+
},
|
4029 |
+
{
|
4030 |
+
"epoch": 0.8568954706953692,
|
4031 |
+
"grad_norm": 3.915356159210205,
|
4032 |
+
"learning_rate": 0.00019180166743809965,
|
4033 |
+
"loss": 1.4561,
|
4034 |
+
"step": 28700
|
4035 |
+
},
|
4036 |
+
{
|
4037 |
+
"epoch": 0.8583883199474517,
|
4038 |
+
"grad_norm": 4.613978862762451,
|
4039 |
+
"learning_rate": 0.00019178738164218081,
|
4040 |
+
"loss": 1.3267,
|
4041 |
+
"step": 28750
|
4042 |
+
},
|
4043 |
+
{
|
4044 |
+
"epoch": 0.8598811691995343,
|
4045 |
+
"grad_norm": 4.935427188873291,
|
4046 |
+
"learning_rate": 0.000191773095846262,
|
4047 |
+
"loss": 1.3962,
|
4048 |
+
"step": 28800
|
4049 |
+
},
|
4050 |
+
{
|
4051 |
+
"epoch": 0.8613740184516168,
|
4052 |
+
"grad_norm": 4.204285144805908,
|
4053 |
+
"learning_rate": 0.00019175881005034314,
|
4054 |
+
"loss": 1.3644,
|
4055 |
+
"step": 28850
|
4056 |
+
},
|
4057 |
+
{
|
4058 |
+
"epoch": 0.8628668677036992,
|
4059 |
+
"grad_norm": 3.901266098022461,
|
4060 |
+
"learning_rate": 0.00019174452425442434,
|
4061 |
+
"loss": 1.344,
|
4062 |
+
"step": 28900
|
4063 |
+
},
|
4064 |
+
{
|
4065 |
+
"epoch": 0.8643597169557818,
|
4066 |
+
"grad_norm": 5.1740922927856445,
|
4067 |
+
"learning_rate": 0.00019173023845850547,
|
4068 |
+
"loss": 1.3852,
|
4069 |
+
"step": 28950
|
4070 |
+
},
|
4071 |
+
{
|
4072 |
+
"epoch": 0.8658525662078643,
|
4073 |
+
"grad_norm": 4.984529972076416,
|
4074 |
+
"learning_rate": 0.00019171595266258667,
|
4075 |
+
"loss": 1.3853,
|
4076 |
+
"step": 29000
|
4077 |
+
},
|
4078 |
+
{
|
4079 |
+
"epoch": 0.8673454154599468,
|
4080 |
+
"grad_norm": 4.719232082366943,
|
4081 |
+
"learning_rate": 0.00019170166686666783,
|
4082 |
+
"loss": 1.3962,
|
4083 |
+
"step": 29050
|
4084 |
+
},
|
4085 |
+
{
|
4086 |
+
"epoch": 0.8688382647120294,
|
4087 |
+
"grad_norm": 6.084499359130859,
|
4088 |
+
"learning_rate": 0.000191687381070749,
|
4089 |
+
"loss": 1.3377,
|
4090 |
+
"step": 29100
|
4091 |
+
},
|
4092 |
+
{
|
4093 |
+
"epoch": 0.8703311139641119,
|
4094 |
+
"grad_norm": 5.261173248291016,
|
4095 |
+
"learning_rate": 0.00019167309527483016,
|
4096 |
+
"loss": 1.3673,
|
4097 |
+
"step": 29150
|
4098 |
+
},
|
4099 |
+
{
|
4100 |
+
"epoch": 0.8718239632161944,
|
4101 |
+
"grad_norm": 5.18539571762085,
|
4102 |
+
"learning_rate": 0.00019165880947891132,
|
4103 |
+
"loss": 1.358,
|
4104 |
+
"step": 29200
|
4105 |
+
},
|
4106 |
+
{
|
4107 |
+
"epoch": 0.873316812468277,
|
4108 |
+
"grad_norm": 4.789880275726318,
|
4109 |
+
"learning_rate": 0.0001916445236829925,
|
4110 |
+
"loss": 1.3426,
|
4111 |
+
"step": 29250
|
4112 |
+
},
|
4113 |
+
{
|
4114 |
+
"epoch": 0.8748096617203595,
|
4115 |
+
"grad_norm": 3.2945287227630615,
|
4116 |
+
"learning_rate": 0.00019163023788707365,
|
4117 |
+
"loss": 1.3308,
|
4118 |
+
"step": 29300
|
4119 |
+
},
|
4120 |
+
{
|
4121 |
+
"epoch": 0.876302510972442,
|
4122 |
+
"grad_norm": 5.060371398925781,
|
4123 |
+
"learning_rate": 0.00019161595209115482,
|
4124 |
+
"loss": 1.3497,
|
4125 |
+
"step": 29350
|
4126 |
+
},
|
4127 |
+
{
|
4128 |
+
"epoch": 0.8777953602245245,
|
4129 |
+
"grad_norm": 4.771635055541992,
|
4130 |
+
"learning_rate": 0.00019160166629523598,
|
4131 |
+
"loss": 1.3517,
|
4132 |
+
"step": 29400
|
4133 |
+
},
|
4134 |
+
{
|
4135 |
+
"epoch": 0.879288209476607,
|
4136 |
+
"grad_norm": 4.408042907714844,
|
4137 |
+
"learning_rate": 0.00019158738049931715,
|
4138 |
+
"loss": 1.3636,
|
4139 |
+
"step": 29450
|
4140 |
+
},
|
4141 |
+
{
|
4142 |
+
"epoch": 0.8807810587286896,
|
4143 |
+
"grad_norm": 5.183899879455566,
|
4144 |
+
"learning_rate": 0.0001915730947033983,
|
4145 |
+
"loss": 1.3569,
|
4146 |
+
"step": 29500
|
4147 |
+
},
|
4148 |
+
{
|
4149 |
+
"epoch": 0.8822739079807721,
|
4150 |
+
"grad_norm": 4.176016807556152,
|
4151 |
+
"learning_rate": 0.00019155880890747948,
|
4152 |
+
"loss": 1.3659,
|
4153 |
+
"step": 29550
|
4154 |
+
},
|
4155 |
+
{
|
4156 |
+
"epoch": 0.8837667572328546,
|
4157 |
+
"grad_norm": 4.783830165863037,
|
4158 |
+
"learning_rate": 0.00019154452311156064,
|
4159 |
+
"loss": 1.368,
|
4160 |
+
"step": 29600
|
4161 |
+
},
|
4162 |
+
{
|
4163 |
+
"epoch": 0.8852596064849372,
|
4164 |
+
"grad_norm": 3.7985739707946777,
|
4165 |
+
"learning_rate": 0.0001915302373156418,
|
4166 |
+
"loss": 1.4091,
|
4167 |
+
"step": 29650
|
4168 |
+
},
|
4169 |
+
{
|
4170 |
+
"epoch": 0.8867524557370197,
|
4171 |
+
"grad_norm": 4.401581287384033,
|
4172 |
+
"learning_rate": 0.00019151595151972297,
|
4173 |
+
"loss": 1.3613,
|
4174 |
+
"step": 29700
|
4175 |
+
},
|
4176 |
+
{
|
4177 |
+
"epoch": 0.8882453049891021,
|
4178 |
+
"grad_norm": 6.214754581451416,
|
4179 |
+
"learning_rate": 0.00019150166572380414,
|
4180 |
+
"loss": 1.4077,
|
4181 |
+
"step": 29750
|
4182 |
+
},
|
4183 |
+
{
|
4184 |
+
"epoch": 0.8897381542411847,
|
4185 |
+
"grad_norm": 4.197826862335205,
|
4186 |
+
"learning_rate": 0.0001914873799278853,
|
4187 |
+
"loss": 1.347,
|
4188 |
+
"step": 29800
|
4189 |
+
},
|
4190 |
+
{
|
4191 |
+
"epoch": 0.8912310034932672,
|
4192 |
+
"grad_norm": 3.1535027027130127,
|
4193 |
+
"learning_rate": 0.0001914730941319665,
|
4194 |
+
"loss": 1.3452,
|
4195 |
+
"step": 29850
|
4196 |
+
},
|
4197 |
+
{
|
4198 |
+
"epoch": 0.8927238527453498,
|
4199 |
+
"grad_norm": 4.618538856506348,
|
4200 |
+
"learning_rate": 0.00019145880833604763,
|
4201 |
+
"loss": 1.3417,
|
4202 |
+
"step": 29900
|
4203 |
+
},
|
4204 |
+
{
|
4205 |
+
"epoch": 0.8942167019974323,
|
4206 |
+
"grad_norm": 3.913506031036377,
|
4207 |
+
"learning_rate": 0.00019144452254012882,
|
4208 |
+
"loss": 1.3503,
|
4209 |
+
"step": 29950
|
4210 |
+
},
|
4211 |
+
{
|
4212 |
+
"epoch": 0.8957095512495148,
|
4213 |
+
"grad_norm": 4.013761043548584,
|
4214 |
+
"learning_rate": 0.00019143023674420996,
|
4215 |
+
"loss": 1.4109,
|
4216 |
+
"step": 30000
|
4217 |
}
|
4218 |
],
|
4219 |
"logging_steps": 50,
|
|
|
4233 |
"attributes": {}
|
4234 |
}
|
4235 |
},
|
4236 |
+
"total_flos": 7.580165416503214e+17,
|
4237 |
"train_batch_size": 2,
|
4238 |
"trial_name": null,
|
4239 |
"trial_params": null
|