Add files using upload-large-folder tool
Browse files- model.safetensors +1 -1
- trainer_state.json +1419 -3
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2740113872
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c613193c24d4be76cc18609b14643006fa41b211a3afb34ad44440d8b9f5f954
|
3 |
size 2740113872
|
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,
|
@@ -4956,6 +4956,1422 @@
|
|
4956 |
"learning_rate": 2.5718163788181375e-05,
|
4957 |
"loss": 6.4442,
|
4958 |
"step": 6990
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4959 |
}
|
4960 |
],
|
4961 |
"logging_steps": 10,
|
@@ -4975,7 +6391,7 @@
|
|
4975 |
"attributes": {}
|
4976 |
}
|
4977 |
},
|
4978 |
-
"total_flos": 1.
|
4979 |
"train_batch_size": 30,
|
4980 |
"trial_name": null,
|
4981 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.6510342497163959,
|
5 |
"eval_steps": 1000,
|
6 |
+
"global_step": 8991,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
4956 |
"learning_rate": 2.5718163788181375e-05,
|
4957 |
"loss": 6.4442,
|
4958 |
"step": 6990
|
4959 |
+
},
|
4960 |
+
{
|
4961 |
+
"epoch": 0.5068668388404818,
|
4962 |
+
"grad_norm": 1.2421875,
|
4963 |
+
"learning_rate": 2.5659552382933732e-05,
|
4964 |
+
"loss": 6.4818,
|
4965 |
+
"step": 7000
|
4966 |
+
},
|
4967 |
+
{
|
4968 |
+
"epoch": 0.5068668388404818,
|
4969 |
+
"eval_loss": 6.613395690917969,
|
4970 |
+
"eval_runtime": 1.5015,
|
4971 |
+
"eval_samples_per_second": 7.992,
|
4972 |
+
"eval_steps_per_second": 1.998,
|
4973 |
+
"step": 7000
|
4974 |
+
},
|
4975 |
+
{
|
4976 |
+
"epoch": 0.5075909343245396,
|
4977 |
+
"grad_norm": 1.21875,
|
4978 |
+
"learning_rate": 2.5600937349718462e-05,
|
4979 |
+
"loss": 6.4476,
|
4980 |
+
"step": 7010
|
4981 |
+
},
|
4982 |
+
{
|
4983 |
+
"epoch": 0.5083150298085974,
|
4984 |
+
"grad_norm": 1.1953125,
|
4985 |
+
"learning_rate": 2.554231901095641e-05,
|
4986 |
+
"loss": 6.459,
|
4987 |
+
"step": 7020
|
4988 |
+
},
|
4989 |
+
{
|
4990 |
+
"epoch": 0.5090391252926553,
|
4991 |
+
"grad_norm": 1.25,
|
4992 |
+
"learning_rate": 2.5483697689086615e-05,
|
4993 |
+
"loss": 6.4414,
|
4994 |
+
"step": 7030
|
4995 |
+
},
|
4996 |
+
{
|
4997 |
+
"epoch": 0.5097632207767131,
|
4998 |
+
"grad_norm": 1.1796875,
|
4999 |
+
"learning_rate": 2.5425073706564528e-05,
|
5000 |
+
"loss": 6.4691,
|
5001 |
+
"step": 7040
|
5002 |
+
},
|
5003 |
+
{
|
5004 |
+
"epoch": 0.5104873162607709,
|
5005 |
+
"grad_norm": 1.2421875,
|
5006 |
+
"learning_rate": 2.5366447385860225e-05,
|
5007 |
+
"loss": 6.4414,
|
5008 |
+
"step": 7050
|
5009 |
+
},
|
5010 |
+
{
|
5011 |
+
"epoch": 0.5112114117448288,
|
5012 |
+
"grad_norm": 1.21875,
|
5013 |
+
"learning_rate": 2.5307819049456655e-05,
|
5014 |
+
"loss": 6.478,
|
5015 |
+
"step": 7060
|
5016 |
+
},
|
5017 |
+
{
|
5018 |
+
"epoch": 0.5119355072288866,
|
5019 |
+
"grad_norm": 1.2109375,
|
5020 |
+
"learning_rate": 2.524918901984784e-05,
|
5021 |
+
"loss": 6.4605,
|
5022 |
+
"step": 7070
|
5023 |
+
},
|
5024 |
+
{
|
5025 |
+
"epoch": 0.5126596027129444,
|
5026 |
+
"grad_norm": 1.2421875,
|
5027 |
+
"learning_rate": 2.519055761953714e-05,
|
5028 |
+
"loss": 6.4395,
|
5029 |
+
"step": 7080
|
5030 |
+
},
|
5031 |
+
{
|
5032 |
+
"epoch": 0.5133836981970022,
|
5033 |
+
"grad_norm": 1.2109375,
|
5034 |
+
"learning_rate": 2.5131925171035438e-05,
|
5035 |
+
"loss": 6.4684,
|
5036 |
+
"step": 7090
|
5037 |
+
},
|
5038 |
+
{
|
5039 |
+
"epoch": 0.5141077936810601,
|
5040 |
+
"grad_norm": 1.2265625,
|
5041 |
+
"learning_rate": 2.5073291996859372e-05,
|
5042 |
+
"loss": 6.4657,
|
5043 |
+
"step": 7100
|
5044 |
+
},
|
5045 |
+
{
|
5046 |
+
"epoch": 0.5148318891651179,
|
5047 |
+
"grad_norm": 1.2109375,
|
5048 |
+
"learning_rate": 2.501465841952959e-05,
|
5049 |
+
"loss": 6.4963,
|
5050 |
+
"step": 7110
|
5051 |
+
},
|
5052 |
+
{
|
5053 |
+
"epoch": 0.5155559846491757,
|
5054 |
+
"grad_norm": 1.1875,
|
5055 |
+
"learning_rate": 2.4956024761568965e-05,
|
5056 |
+
"loss": 6.4635,
|
5057 |
+
"step": 7120
|
5058 |
+
},
|
5059 |
+
{
|
5060 |
+
"epoch": 0.5162800801332336,
|
5061 |
+
"grad_norm": 1.2890625,
|
5062 |
+
"learning_rate": 2.4897391345500777e-05,
|
5063 |
+
"loss": 6.4686,
|
5064 |
+
"step": 7130
|
5065 |
+
},
|
5066 |
+
{
|
5067 |
+
"epoch": 0.5170041756172914,
|
5068 |
+
"grad_norm": 1.1875,
|
5069 |
+
"learning_rate": 2.483875849384702e-05,
|
5070 |
+
"loss": 6.4515,
|
5071 |
+
"step": 7140
|
5072 |
+
},
|
5073 |
+
{
|
5074 |
+
"epoch": 0.5177282711013492,
|
5075 |
+
"grad_norm": 1.1953125,
|
5076 |
+
"learning_rate": 2.4780126529126555e-05,
|
5077 |
+
"loss": 6.4655,
|
5078 |
+
"step": 7150
|
5079 |
+
},
|
5080 |
+
{
|
5081 |
+
"epoch": 0.518452366585407,
|
5082 |
+
"grad_norm": 1.2109375,
|
5083 |
+
"learning_rate": 2.4721495773853366e-05,
|
5084 |
+
"loss": 6.457,
|
5085 |
+
"step": 7160
|
5086 |
+
},
|
5087 |
+
{
|
5088 |
+
"epoch": 0.5191764620694649,
|
5089 |
+
"grad_norm": 1.1796875,
|
5090 |
+
"learning_rate": 2.4662866550534798e-05,
|
5091 |
+
"loss": 6.4679,
|
5092 |
+
"step": 7170
|
5093 |
+
},
|
5094 |
+
{
|
5095 |
+
"epoch": 0.5199005575535227,
|
5096 |
+
"grad_norm": 1.25,
|
5097 |
+
"learning_rate": 2.460423918166975e-05,
|
5098 |
+
"loss": 6.4634,
|
5099 |
+
"step": 7180
|
5100 |
+
},
|
5101 |
+
{
|
5102 |
+
"epoch": 0.5206246530375805,
|
5103 |
+
"grad_norm": 1.15625,
|
5104 |
+
"learning_rate": 2.4545613989746953e-05,
|
5105 |
+
"loss": 6.4335,
|
5106 |
+
"step": 7190
|
5107 |
+
},
|
5108 |
+
{
|
5109 |
+
"epoch": 0.5213487485216384,
|
5110 |
+
"grad_norm": 1.2421875,
|
5111 |
+
"learning_rate": 2.4486991297243123e-05,
|
5112 |
+
"loss": 6.4535,
|
5113 |
+
"step": 7200
|
5114 |
+
},
|
5115 |
+
{
|
5116 |
+
"epoch": 0.5220728440056962,
|
5117 |
+
"grad_norm": 1.171875,
|
5118 |
+
"learning_rate": 2.4428371426621248e-05,
|
5119 |
+
"loss": 6.4829,
|
5120 |
+
"step": 7210
|
5121 |
+
},
|
5122 |
+
{
|
5123 |
+
"epoch": 0.522796939489754,
|
5124 |
+
"grad_norm": 1.2265625,
|
5125 |
+
"learning_rate": 2.4369754700328805e-05,
|
5126 |
+
"loss": 6.4657,
|
5127 |
+
"step": 7220
|
5128 |
+
},
|
5129 |
+
{
|
5130 |
+
"epoch": 0.5235210349738119,
|
5131 |
+
"grad_norm": 1.1875,
|
5132 |
+
"learning_rate": 2.4311141440795953e-05,
|
5133 |
+
"loss": 6.4412,
|
5134 |
+
"step": 7230
|
5135 |
+
},
|
5136 |
+
{
|
5137 |
+
"epoch": 0.5242451304578697,
|
5138 |
+
"grad_norm": 1.1796875,
|
5139 |
+
"learning_rate": 2.4252531970433786e-05,
|
5140 |
+
"loss": 6.4693,
|
5141 |
+
"step": 7240
|
5142 |
+
},
|
5143 |
+
{
|
5144 |
+
"epoch": 0.5249692259419275,
|
5145 |
+
"grad_norm": 1.1640625,
|
5146 |
+
"learning_rate": 2.419392661163257e-05,
|
5147 |
+
"loss": 6.46,
|
5148 |
+
"step": 7250
|
5149 |
+
},
|
5150 |
+
{
|
5151 |
+
"epoch": 0.5256933214259853,
|
5152 |
+
"grad_norm": 1.171875,
|
5153 |
+
"learning_rate": 2.4135325686759942e-05,
|
5154 |
+
"loss": 6.4613,
|
5155 |
+
"step": 7260
|
5156 |
+
},
|
5157 |
+
{
|
5158 |
+
"epoch": 0.5264174169100432,
|
5159 |
+
"grad_norm": 1.1796875,
|
5160 |
+
"learning_rate": 2.4076729518159146e-05,
|
5161 |
+
"loss": 6.463,
|
5162 |
+
"step": 7270
|
5163 |
+
},
|
5164 |
+
{
|
5165 |
+
"epoch": 0.527141512394101,
|
5166 |
+
"grad_norm": 1.265625,
|
5167 |
+
"learning_rate": 2.401813842814728e-05,
|
5168 |
+
"loss": 6.4323,
|
5169 |
+
"step": 7280
|
5170 |
+
},
|
5171 |
+
{
|
5172 |
+
"epoch": 0.5278656078781588,
|
5173 |
+
"grad_norm": 1.203125,
|
5174 |
+
"learning_rate": 2.395955273901349e-05,
|
5175 |
+
"loss": 6.4312,
|
5176 |
+
"step": 7290
|
5177 |
+
},
|
5178 |
+
{
|
5179 |
+
"epoch": 0.5285897033622167,
|
5180 |
+
"grad_norm": 1.2109375,
|
5181 |
+
"learning_rate": 2.390097277301722e-05,
|
5182 |
+
"loss": 6.4483,
|
5183 |
+
"step": 7300
|
5184 |
+
},
|
5185 |
+
{
|
5186 |
+
"epoch": 0.5293137988462745,
|
5187 |
+
"grad_norm": 1.1796875,
|
5188 |
+
"learning_rate": 2.3842398852386434e-05,
|
5189 |
+
"loss": 6.468,
|
5190 |
+
"step": 7310
|
5191 |
+
},
|
5192 |
+
{
|
5193 |
+
"epoch": 0.5300378943303323,
|
5194 |
+
"grad_norm": 1.234375,
|
5195 |
+
"learning_rate": 2.3783831299315845e-05,
|
5196 |
+
"loss": 6.4574,
|
5197 |
+
"step": 7320
|
5198 |
+
},
|
5199 |
+
{
|
5200 |
+
"epoch": 0.5307619898143902,
|
5201 |
+
"grad_norm": 1.21875,
|
5202 |
+
"learning_rate": 2.3725270435965123e-05,
|
5203 |
+
"loss": 6.4518,
|
5204 |
+
"step": 7330
|
5205 |
+
},
|
5206 |
+
{
|
5207 |
+
"epoch": 0.531486085298448,
|
5208 |
+
"grad_norm": 1.2265625,
|
5209 |
+
"learning_rate": 2.3666716584457174e-05,
|
5210 |
+
"loss": 6.4598,
|
5211 |
+
"step": 7340
|
5212 |
+
},
|
5213 |
+
{
|
5214 |
+
"epoch": 0.5322101807825058,
|
5215 |
+
"grad_norm": 1.21875,
|
5216 |
+
"learning_rate": 2.36081700668763e-05,
|
5217 |
+
"loss": 6.4511,
|
5218 |
+
"step": 7350
|
5219 |
+
},
|
5220 |
+
{
|
5221 |
+
"epoch": 0.5329342762665636,
|
5222 |
+
"grad_norm": 1.328125,
|
5223 |
+
"learning_rate": 2.3549631205266482e-05,
|
5224 |
+
"loss": 6.477,
|
5225 |
+
"step": 7360
|
5226 |
+
},
|
5227 |
+
{
|
5228 |
+
"epoch": 0.5336583717506215,
|
5229 |
+
"grad_norm": 1.1484375,
|
5230 |
+
"learning_rate": 2.3491100321629573e-05,
|
5231 |
+
"loss": 6.4593,
|
5232 |
+
"step": 7370
|
5233 |
+
},
|
5234 |
+
{
|
5235 |
+
"epoch": 0.5343824672346793,
|
5236 |
+
"grad_norm": 1.203125,
|
5237 |
+
"learning_rate": 2.3432577737923564e-05,
|
5238 |
+
"loss": 6.4507,
|
5239 |
+
"step": 7380
|
5240 |
+
},
|
5241 |
+
{
|
5242 |
+
"epoch": 0.5351065627187371,
|
5243 |
+
"grad_norm": 1.1640625,
|
5244 |
+
"learning_rate": 2.337406377606077e-05,
|
5245 |
+
"loss": 6.4771,
|
5246 |
+
"step": 7390
|
5247 |
+
},
|
5248 |
+
{
|
5249 |
+
"epoch": 0.535830658202795,
|
5250 |
+
"grad_norm": 1.2265625,
|
5251 |
+
"learning_rate": 2.3315558757906087e-05,
|
5252 |
+
"loss": 6.4647,
|
5253 |
+
"step": 7400
|
5254 |
+
},
|
5255 |
+
{
|
5256 |
+
"epoch": 0.5365547536868528,
|
5257 |
+
"grad_norm": 1.1640625,
|
5258 |
+
"learning_rate": 2.325706300527521e-05,
|
5259 |
+
"loss": 6.4545,
|
5260 |
+
"step": 7410
|
5261 |
+
},
|
5262 |
+
{
|
5263 |
+
"epoch": 0.5372788491709106,
|
5264 |
+
"grad_norm": 1.21875,
|
5265 |
+
"learning_rate": 2.3198576839932893e-05,
|
5266 |
+
"loss": 6.4633,
|
5267 |
+
"step": 7420
|
5268 |
+
},
|
5269 |
+
{
|
5270 |
+
"epoch": 0.5380029446549685,
|
5271 |
+
"grad_norm": 1.203125,
|
5272 |
+
"learning_rate": 2.3140100583591124e-05,
|
5273 |
+
"loss": 6.4553,
|
5274 |
+
"step": 7430
|
5275 |
+
},
|
5276 |
+
{
|
5277 |
+
"epoch": 0.5387270401390263,
|
5278 |
+
"grad_norm": 1.171875,
|
5279 |
+
"learning_rate": 2.3081634557907388e-05,
|
5280 |
+
"loss": 6.4596,
|
5281 |
+
"step": 7440
|
5282 |
+
},
|
5283 |
+
{
|
5284 |
+
"epoch": 0.5394511356230841,
|
5285 |
+
"grad_norm": 1.2109375,
|
5286 |
+
"learning_rate": 2.3023179084482916e-05,
|
5287 |
+
"loss": 6.4385,
|
5288 |
+
"step": 7450
|
5289 |
+
},
|
5290 |
+
{
|
5291 |
+
"epoch": 0.5401752311071419,
|
5292 |
+
"grad_norm": 1.21875,
|
5293 |
+
"learning_rate": 2.296473448486087e-05,
|
5294 |
+
"loss": 6.4597,
|
5295 |
+
"step": 7460
|
5296 |
+
},
|
5297 |
+
{
|
5298 |
+
"epoch": 0.5408993265911999,
|
5299 |
+
"grad_norm": 1.1328125,
|
5300 |
+
"learning_rate": 2.2906301080524614e-05,
|
5301 |
+
"loss": 6.4686,
|
5302 |
+
"step": 7470
|
5303 |
+
},
|
5304 |
+
{
|
5305 |
+
"epoch": 0.5416234220752577,
|
5306 |
+
"grad_norm": 1.1484375,
|
5307 |
+
"learning_rate": 2.2847879192895922e-05,
|
5308 |
+
"loss": 6.4299,
|
5309 |
+
"step": 7480
|
5310 |
+
},
|
5311 |
+
{
|
5312 |
+
"epoch": 0.5423475175593155,
|
5313 |
+
"grad_norm": 1.15625,
|
5314 |
+
"learning_rate": 2.278946914333323e-05,
|
5315 |
+
"loss": 6.4546,
|
5316 |
+
"step": 7490
|
5317 |
+
},
|
5318 |
+
{
|
5319 |
+
"epoch": 0.5430716130433734,
|
5320 |
+
"grad_norm": 1.1484375,
|
5321 |
+
"learning_rate": 2.2731071253129844e-05,
|
5322 |
+
"loss": 6.4454,
|
5323 |
+
"step": 7500
|
5324 |
+
},
|
5325 |
+
{
|
5326 |
+
"epoch": 0.5437957085274312,
|
5327 |
+
"grad_norm": 1.234375,
|
5328 |
+
"learning_rate": 2.2672685843512187e-05,
|
5329 |
+
"loss": 6.471,
|
5330 |
+
"step": 7510
|
5331 |
+
},
|
5332 |
+
{
|
5333 |
+
"epoch": 0.544519804011489,
|
5334 |
+
"grad_norm": 1.234375,
|
5335 |
+
"learning_rate": 2.2614313235638047e-05,
|
5336 |
+
"loss": 6.4412,
|
5337 |
+
"step": 7520
|
5338 |
+
},
|
5339 |
+
{
|
5340 |
+
"epoch": 0.5452438994955469,
|
5341 |
+
"grad_norm": 1.2109375,
|
5342 |
+
"learning_rate": 2.2555953750594777e-05,
|
5343 |
+
"loss": 6.4591,
|
5344 |
+
"step": 7530
|
5345 |
+
},
|
5346 |
+
{
|
5347 |
+
"epoch": 0.5459679949796047,
|
5348 |
+
"grad_norm": 1.1484375,
|
5349 |
+
"learning_rate": 2.2497607709397543e-05,
|
5350 |
+
"loss": 6.4524,
|
5351 |
+
"step": 7540
|
5352 |
+
},
|
5353 |
+
{
|
5354 |
+
"epoch": 0.5466920904636625,
|
5355 |
+
"grad_norm": 1.2265625,
|
5356 |
+
"learning_rate": 2.2439275432987585e-05,
|
5357 |
+
"loss": 6.4796,
|
5358 |
+
"step": 7550
|
5359 |
+
},
|
5360 |
+
{
|
5361 |
+
"epoch": 0.5474161859477203,
|
5362 |
+
"grad_norm": 1.2734375,
|
5363 |
+
"learning_rate": 2.23809572422304e-05,
|
5364 |
+
"loss": 6.4641,
|
5365 |
+
"step": 7560
|
5366 |
+
},
|
5367 |
+
{
|
5368 |
+
"epoch": 0.5481402814317782,
|
5369 |
+
"grad_norm": 1.125,
|
5370 |
+
"learning_rate": 2.2322653457914014e-05,
|
5371 |
+
"loss": 6.4564,
|
5372 |
+
"step": 7570
|
5373 |
+
},
|
5374 |
+
{
|
5375 |
+
"epoch": 0.548864376915836,
|
5376 |
+
"grad_norm": 1.1796875,
|
5377 |
+
"learning_rate": 2.2264364400747226e-05,
|
5378 |
+
"loss": 6.4588,
|
5379 |
+
"step": 7580
|
5380 |
+
},
|
5381 |
+
{
|
5382 |
+
"epoch": 0.5495884723998938,
|
5383 |
+
"grad_norm": 1.296875,
|
5384 |
+
"learning_rate": 2.2206090391357797e-05,
|
5385 |
+
"loss": 6.4472,
|
5386 |
+
"step": 7590
|
5387 |
+
},
|
5388 |
+
{
|
5389 |
+
"epoch": 0.5503125678839517,
|
5390 |
+
"grad_norm": 1.21875,
|
5391 |
+
"learning_rate": 2.2147831750290732e-05,
|
5392 |
+
"loss": 6.4609,
|
5393 |
+
"step": 7600
|
5394 |
+
},
|
5395 |
+
{
|
5396 |
+
"epoch": 0.5510366633680095,
|
5397 |
+
"grad_norm": 1.171875,
|
5398 |
+
"learning_rate": 2.20895887980065e-05,
|
5399 |
+
"loss": 6.4575,
|
5400 |
+
"step": 7610
|
5401 |
+
},
|
5402 |
+
{
|
5403 |
+
"epoch": 0.5517607588520673,
|
5404 |
+
"grad_norm": 1.140625,
|
5405 |
+
"learning_rate": 2.203136185487927e-05,
|
5406 |
+
"loss": 6.4575,
|
5407 |
+
"step": 7620
|
5408 |
+
},
|
5409 |
+
{
|
5410 |
+
"epoch": 0.5524848543361252,
|
5411 |
+
"grad_norm": 1.1875,
|
5412 |
+
"learning_rate": 2.1973151241195143e-05,
|
5413 |
+
"loss": 6.4774,
|
5414 |
+
"step": 7630
|
5415 |
+
},
|
5416 |
+
{
|
5417 |
+
"epoch": 0.553208949820183,
|
5418 |
+
"grad_norm": 1.2421875,
|
5419 |
+
"learning_rate": 2.191495727715041e-05,
|
5420 |
+
"loss": 6.4593,
|
5421 |
+
"step": 7640
|
5422 |
+
},
|
5423 |
+
{
|
5424 |
+
"epoch": 0.5539330453042408,
|
5425 |
+
"grad_norm": 1.1875,
|
5426 |
+
"learning_rate": 2.1856780282849765e-05,
|
5427 |
+
"loss": 6.4552,
|
5428 |
+
"step": 7650
|
5429 |
+
},
|
5430 |
+
{
|
5431 |
+
"epoch": 0.5546571407882986,
|
5432 |
+
"grad_norm": 1.1640625,
|
5433 |
+
"learning_rate": 2.1798620578304575e-05,
|
5434 |
+
"loss": 6.4388,
|
5435 |
+
"step": 7660
|
5436 |
+
},
|
5437 |
+
{
|
5438 |
+
"epoch": 0.5553812362723565,
|
5439 |
+
"grad_norm": 1.171875,
|
5440 |
+
"learning_rate": 2.1740478483431088e-05,
|
5441 |
+
"loss": 6.4495,
|
5442 |
+
"step": 7670
|
5443 |
+
},
|
5444 |
+
{
|
5445 |
+
"epoch": 0.5561053317564143,
|
5446 |
+
"grad_norm": 1.1640625,
|
5447 |
+
"learning_rate": 2.1682354318048676e-05,
|
5448 |
+
"loss": 6.4629,
|
5449 |
+
"step": 7680
|
5450 |
+
},
|
5451 |
+
{
|
5452 |
+
"epoch": 0.5568294272404721,
|
5453 |
+
"grad_norm": 1.2265625,
|
5454 |
+
"learning_rate": 2.1624248401878123e-05,
|
5455 |
+
"loss": 6.4558,
|
5456 |
+
"step": 7690
|
5457 |
+
},
|
5458 |
+
{
|
5459 |
+
"epoch": 0.55755352272453,
|
5460 |
+
"grad_norm": 1.1328125,
|
5461 |
+
"learning_rate": 2.1566161054539798e-05,
|
5462 |
+
"loss": 6.4757,
|
5463 |
+
"step": 7700
|
5464 |
+
},
|
5465 |
+
{
|
5466 |
+
"epoch": 0.5582776182085878,
|
5467 |
+
"grad_norm": 1.2578125,
|
5468 |
+
"learning_rate": 2.1508092595551938e-05,
|
5469 |
+
"loss": 6.4555,
|
5470 |
+
"step": 7710
|
5471 |
+
},
|
5472 |
+
{
|
5473 |
+
"epoch": 0.5590017136926456,
|
5474 |
+
"grad_norm": 1.1953125,
|
5475 |
+
"learning_rate": 2.1450043344328892e-05,
|
5476 |
+
"loss": 6.4687,
|
5477 |
+
"step": 7720
|
5478 |
+
},
|
5479 |
+
{
|
5480 |
+
"epoch": 0.5597258091767034,
|
5481 |
+
"grad_norm": 1.171875,
|
5482 |
+
"learning_rate": 2.1392013620179337e-05,
|
5483 |
+
"loss": 6.4549,
|
5484 |
+
"step": 7730
|
5485 |
+
},
|
5486 |
+
{
|
5487 |
+
"epoch": 0.5604499046607613,
|
5488 |
+
"grad_norm": 1.1875,
|
5489 |
+
"learning_rate": 2.133400374230455e-05,
|
5490 |
+
"loss": 6.4584,
|
5491 |
+
"step": 7740
|
5492 |
+
},
|
5493 |
+
{
|
5494 |
+
"epoch": 0.5611740001448191,
|
5495 |
+
"grad_norm": 1.15625,
|
5496 |
+
"learning_rate": 2.127601402979665e-05,
|
5497 |
+
"loss": 6.4676,
|
5498 |
+
"step": 7750
|
5499 |
+
},
|
5500 |
+
{
|
5501 |
+
"epoch": 0.5618980956288769,
|
5502 |
+
"grad_norm": 1.21875,
|
5503 |
+
"learning_rate": 2.1218044801636805e-05,
|
5504 |
+
"loss": 6.4536,
|
5505 |
+
"step": 7760
|
5506 |
+
},
|
5507 |
+
{
|
5508 |
+
"epoch": 0.5626221911129348,
|
5509 |
+
"grad_norm": 1.1953125,
|
5510 |
+
"learning_rate": 2.116009637669353e-05,
|
5511 |
+
"loss": 6.4334,
|
5512 |
+
"step": 7770
|
5513 |
+
},
|
5514 |
+
{
|
5515 |
+
"epoch": 0.5633462865969926,
|
5516 |
+
"grad_norm": 1.1953125,
|
5517 |
+
"learning_rate": 2.1102169073720894e-05,
|
5518 |
+
"loss": 6.4507,
|
5519 |
+
"step": 7780
|
5520 |
+
},
|
5521 |
+
{
|
5522 |
+
"epoch": 0.5640703820810504,
|
5523 |
+
"grad_norm": 1.203125,
|
5524 |
+
"learning_rate": 2.1044263211356797e-05,
|
5525 |
+
"loss": 6.436,
|
5526 |
+
"step": 7790
|
5527 |
+
},
|
5528 |
+
{
|
5529 |
+
"epoch": 0.5647944775651083,
|
5530 |
+
"grad_norm": 1.1328125,
|
5531 |
+
"learning_rate": 2.098637910812119e-05,
|
5532 |
+
"loss": 6.4561,
|
5533 |
+
"step": 7800
|
5534 |
+
},
|
5535 |
+
{
|
5536 |
+
"epoch": 0.5655185730491661,
|
5537 |
+
"grad_norm": 1.1953125,
|
5538 |
+
"learning_rate": 2.0928517082414335e-05,
|
5539 |
+
"loss": 6.454,
|
5540 |
+
"step": 7810
|
5541 |
+
},
|
5542 |
+
{
|
5543 |
+
"epoch": 0.5662426685332239,
|
5544 |
+
"grad_norm": 1.1796875,
|
5545 |
+
"learning_rate": 2.0870677452515057e-05,
|
5546 |
+
"loss": 6.4499,
|
5547 |
+
"step": 7820
|
5548 |
+
},
|
5549 |
+
{
|
5550 |
+
"epoch": 0.5669667640172817,
|
5551 |
+
"grad_norm": 1.2265625,
|
5552 |
+
"learning_rate": 2.081286053657899e-05,
|
5553 |
+
"loss": 6.4552,
|
5554 |
+
"step": 7830
|
5555 |
+
},
|
5556 |
+
{
|
5557 |
+
"epoch": 0.5676908595013396,
|
5558 |
+
"grad_norm": 1.203125,
|
5559 |
+
"learning_rate": 2.0755066652636817e-05,
|
5560 |
+
"loss": 6.4605,
|
5561 |
+
"step": 7840
|
5562 |
+
},
|
5563 |
+
{
|
5564 |
+
"epoch": 0.5684149549853974,
|
5565 |
+
"grad_norm": 1.203125,
|
5566 |
+
"learning_rate": 2.0697296118592553e-05,
|
5567 |
+
"loss": 6.4492,
|
5568 |
+
"step": 7850
|
5569 |
+
},
|
5570 |
+
{
|
5571 |
+
"epoch": 0.5691390504694552,
|
5572 |
+
"grad_norm": 1.2734375,
|
5573 |
+
"learning_rate": 2.0639549252221745e-05,
|
5574 |
+
"loss": 6.4476,
|
5575 |
+
"step": 7860
|
5576 |
+
},
|
5577 |
+
{
|
5578 |
+
"epoch": 0.5698631459535131,
|
5579 |
+
"grad_norm": 1.1953125,
|
5580 |
+
"learning_rate": 2.058182637116976e-05,
|
5581 |
+
"loss": 6.4568,
|
5582 |
+
"step": 7870
|
5583 |
+
},
|
5584 |
+
{
|
5585 |
+
"epoch": 0.5705872414375709,
|
5586 |
+
"grad_norm": 1.2734375,
|
5587 |
+
"learning_rate": 2.0524127792950056e-05,
|
5588 |
+
"loss": 6.4512,
|
5589 |
+
"step": 7880
|
5590 |
+
},
|
5591 |
+
{
|
5592 |
+
"epoch": 0.5713113369216287,
|
5593 |
+
"grad_norm": 1.1875,
|
5594 |
+
"learning_rate": 2.046645383494237e-05,
|
5595 |
+
"loss": 6.4383,
|
5596 |
+
"step": 7890
|
5597 |
+
},
|
5598 |
+
{
|
5599 |
+
"epoch": 0.5720354324056866,
|
5600 |
+
"grad_norm": 1.2265625,
|
5601 |
+
"learning_rate": 2.0408804814391043e-05,
|
5602 |
+
"loss": 6.4694,
|
5603 |
+
"step": 7900
|
5604 |
+
},
|
5605 |
+
{
|
5606 |
+
"epoch": 0.5727595278897444,
|
5607 |
+
"grad_norm": 1.1640625,
|
5608 |
+
"learning_rate": 2.0351181048403227e-05,
|
5609 |
+
"loss": 6.4535,
|
5610 |
+
"step": 7910
|
5611 |
+
},
|
5612 |
+
{
|
5613 |
+
"epoch": 0.5734836233738022,
|
5614 |
+
"grad_norm": 1.2109375,
|
5615 |
+
"learning_rate": 2.0293582853947164e-05,
|
5616 |
+
"loss": 6.4736,
|
5617 |
+
"step": 7920
|
5618 |
+
},
|
5619 |
+
{
|
5620 |
+
"epoch": 0.57420771885786,
|
5621 |
+
"grad_norm": 1.203125,
|
5622 |
+
"learning_rate": 2.0236010547850432e-05,
|
5623 |
+
"loss": 6.4463,
|
5624 |
+
"step": 7930
|
5625 |
+
},
|
5626 |
+
{
|
5627 |
+
"epoch": 0.5749318143419179,
|
5628 |
+
"grad_norm": 1.2421875,
|
5629 |
+
"learning_rate": 2.0178464446798206e-05,
|
5630 |
+
"loss": 6.462,
|
5631 |
+
"step": 7940
|
5632 |
+
},
|
5633 |
+
{
|
5634 |
+
"epoch": 0.5756559098259757,
|
5635 |
+
"grad_norm": 1.2109375,
|
5636 |
+
"learning_rate": 2.0120944867331522e-05,
|
5637 |
+
"loss": 6.4542,
|
5638 |
+
"step": 7950
|
5639 |
+
},
|
5640 |
+
{
|
5641 |
+
"epoch": 0.5763800053100335,
|
5642 |
+
"grad_norm": 1.171875,
|
5643 |
+
"learning_rate": 2.0063452125845522e-05,
|
5644 |
+
"loss": 6.4458,
|
5645 |
+
"step": 7960
|
5646 |
+
},
|
5647 |
+
{
|
5648 |
+
"epoch": 0.5771041007940914,
|
5649 |
+
"grad_norm": 1.1484375,
|
5650 |
+
"learning_rate": 2.0005986538587735e-05,
|
5651 |
+
"loss": 6.4731,
|
5652 |
+
"step": 7970
|
5653 |
+
},
|
5654 |
+
{
|
5655 |
+
"epoch": 0.5778281962781492,
|
5656 |
+
"grad_norm": 1.1328125,
|
5657 |
+
"learning_rate": 1.9948548421656303e-05,
|
5658 |
+
"loss": 6.4686,
|
5659 |
+
"step": 7980
|
5660 |
+
},
|
5661 |
+
{
|
5662 |
+
"epoch": 0.578552291762207,
|
5663 |
+
"grad_norm": 1.1953125,
|
5664 |
+
"learning_rate": 1.9891138090998285e-05,
|
5665 |
+
"loss": 6.4521,
|
5666 |
+
"step": 7990
|
5667 |
+
},
|
5668 |
+
{
|
5669 |
+
"epoch": 0.5792763872462648,
|
5670 |
+
"grad_norm": 1.1875,
|
5671 |
+
"learning_rate": 1.983375586240789e-05,
|
5672 |
+
"loss": 6.4232,
|
5673 |
+
"step": 8000
|
5674 |
+
},
|
5675 |
+
{
|
5676 |
+
"epoch": 0.5792763872462648,
|
5677 |
+
"eval_loss": 6.610226154327393,
|
5678 |
+
"eval_runtime": 1.4712,
|
5679 |
+
"eval_samples_per_second": 8.157,
|
5680 |
+
"eval_steps_per_second": 2.039,
|
5681 |
+
"step": 8000
|
5682 |
+
},
|
5683 |
+
{
|
5684 |
+
"epoch": 0.5800004827303227,
|
5685 |
+
"grad_norm": 1.25,
|
5686 |
+
"learning_rate": 1.9776402051524724e-05,
|
5687 |
+
"loss": 6.4368,
|
5688 |
+
"step": 8010
|
5689 |
+
},
|
5690 |
+
{
|
5691 |
+
"epoch": 0.5807245782143805,
|
5692 |
+
"grad_norm": 1.1875,
|
5693 |
+
"learning_rate": 1.971907697383213e-05,
|
5694 |
+
"loss": 6.4723,
|
5695 |
+
"step": 8020
|
5696 |
+
},
|
5697 |
+
{
|
5698 |
+
"epoch": 0.5814486736984383,
|
5699 |
+
"grad_norm": 1.1953125,
|
5700 |
+
"learning_rate": 1.9661780944655345e-05,
|
5701 |
+
"loss": 6.4643,
|
5702 |
+
"step": 8030
|
5703 |
+
},
|
5704 |
+
{
|
5705 |
+
"epoch": 0.5821727691824962,
|
5706 |
+
"grad_norm": 1.1796875,
|
5707 |
+
"learning_rate": 1.9604514279159843e-05,
|
5708 |
+
"loss": 6.4756,
|
5709 |
+
"step": 8040
|
5710 |
+
},
|
5711 |
+
{
|
5712 |
+
"epoch": 0.582896864666554,
|
5713 |
+
"grad_norm": 1.140625,
|
5714 |
+
"learning_rate": 1.95472772923496e-05,
|
5715 |
+
"loss": 6.4411,
|
5716 |
+
"step": 8050
|
5717 |
+
},
|
5718 |
+
{
|
5719 |
+
"epoch": 0.5836209601506118,
|
5720 |
+
"grad_norm": 1.25,
|
5721 |
+
"learning_rate": 1.9490070299065293e-05,
|
5722 |
+
"loss": 6.4344,
|
5723 |
+
"step": 8060
|
5724 |
+
},
|
5725 |
+
{
|
5726 |
+
"epoch": 0.5843450556346697,
|
5727 |
+
"grad_norm": 1.1953125,
|
5728 |
+
"learning_rate": 1.943289361398266e-05,
|
5729 |
+
"loss": 6.4451,
|
5730 |
+
"step": 8070
|
5731 |
+
},
|
5732 |
+
{
|
5733 |
+
"epoch": 0.5850691511187275,
|
5734 |
+
"grad_norm": 1.25,
|
5735 |
+
"learning_rate": 1.9375747551610688e-05,
|
5736 |
+
"loss": 6.4609,
|
5737 |
+
"step": 8080
|
5738 |
+
},
|
5739 |
+
{
|
5740 |
+
"epoch": 0.5857932466027853,
|
5741 |
+
"grad_norm": 1.2421875,
|
5742 |
+
"learning_rate": 1.931863242628995e-05,
|
5743 |
+
"loss": 6.4627,
|
5744 |
+
"step": 8090
|
5745 |
+
},
|
5746 |
+
{
|
5747 |
+
"epoch": 0.5865173420868431,
|
5748 |
+
"grad_norm": 1.1484375,
|
5749 |
+
"learning_rate": 1.9261548552190827e-05,
|
5750 |
+
"loss": 6.4676,
|
5751 |
+
"step": 8100
|
5752 |
+
},
|
5753 |
+
{
|
5754 |
+
"epoch": 0.587241437570901,
|
5755 |
+
"grad_norm": 1.1875,
|
5756 |
+
"learning_rate": 1.920449624331179e-05,
|
5757 |
+
"loss": 6.457,
|
5758 |
+
"step": 8110
|
5759 |
+
},
|
5760 |
+
{
|
5761 |
+
"epoch": 0.5879655330549588,
|
5762 |
+
"grad_norm": 1.140625,
|
5763 |
+
"learning_rate": 1.9147475813477718e-05,
|
5764 |
+
"loss": 6.4659,
|
5765 |
+
"step": 8120
|
5766 |
+
},
|
5767 |
+
{
|
5768 |
+
"epoch": 0.5886896285390166,
|
5769 |
+
"grad_norm": 1.21875,
|
5770 |
+
"learning_rate": 1.909048757633809e-05,
|
5771 |
+
"loss": 6.4445,
|
5772 |
+
"step": 8130
|
5773 |
+
},
|
5774 |
+
{
|
5775 |
+
"epoch": 0.5894137240230745,
|
5776 |
+
"grad_norm": 1.265625,
|
5777 |
+
"learning_rate": 1.9033531845365334e-05,
|
5778 |
+
"loss": 6.4637,
|
5779 |
+
"step": 8140
|
5780 |
+
},
|
5781 |
+
{
|
5782 |
+
"epoch": 0.5901378195071323,
|
5783 |
+
"grad_norm": 1.171875,
|
5784 |
+
"learning_rate": 1.897660893385305e-05,
|
5785 |
+
"loss": 6.4351,
|
5786 |
+
"step": 8150
|
5787 |
+
},
|
5788 |
+
{
|
5789 |
+
"epoch": 0.5908619149911901,
|
5790 |
+
"grad_norm": 1.1953125,
|
5791 |
+
"learning_rate": 1.891971915491434e-05,
|
5792 |
+
"loss": 6.4607,
|
5793 |
+
"step": 8160
|
5794 |
+
},
|
5795 |
+
{
|
5796 |
+
"epoch": 0.591586010475248,
|
5797 |
+
"grad_norm": 1.1796875,
|
5798 |
+
"learning_rate": 1.8862862821480025e-05,
|
5799 |
+
"loss": 6.4635,
|
5800 |
+
"step": 8170
|
5801 |
+
},
|
5802 |
+
{
|
5803 |
+
"epoch": 0.5923101059593059,
|
5804 |
+
"grad_norm": 1.2578125,
|
5805 |
+
"learning_rate": 1.8806040246296966e-05,
|
5806 |
+
"loss": 6.4431,
|
5807 |
+
"step": 8180
|
5808 |
+
},
|
5809 |
+
{
|
5810 |
+
"epoch": 0.5930342014433637,
|
5811 |
+
"grad_norm": 1.1796875,
|
5812 |
+
"learning_rate": 1.8749251741926326e-05,
|
5813 |
+
"loss": 6.4586,
|
5814 |
+
"step": 8190
|
5815 |
+
},
|
5816 |
+
{
|
5817 |
+
"epoch": 0.5937582969274215,
|
5818 |
+
"grad_norm": 1.1484375,
|
5819 |
+
"learning_rate": 1.8692497620741884e-05,
|
5820 |
+
"loss": 6.4613,
|
5821 |
+
"step": 8200
|
5822 |
+
},
|
5823 |
+
{
|
5824 |
+
"epoch": 0.5944823924114794,
|
5825 |
+
"grad_norm": 1.2734375,
|
5826 |
+
"learning_rate": 1.8635778194928246e-05,
|
5827 |
+
"loss": 6.4493,
|
5828 |
+
"step": 8210
|
5829 |
+
},
|
5830 |
+
{
|
5831 |
+
"epoch": 0.5952064878955372,
|
5832 |
+
"grad_norm": 1.203125,
|
5833 |
+
"learning_rate": 1.85790937764792e-05,
|
5834 |
+
"loss": 6.4421,
|
5835 |
+
"step": 8220
|
5836 |
+
},
|
5837 |
+
{
|
5838 |
+
"epoch": 0.595930583379595,
|
5839 |
+
"grad_norm": 1.2109375,
|
5840 |
+
"learning_rate": 1.8522444677195965e-05,
|
5841 |
+
"loss": 6.4376,
|
5842 |
+
"step": 8230
|
5843 |
+
},
|
5844 |
+
{
|
5845 |
+
"epoch": 0.5966546788636529,
|
5846 |
+
"grad_norm": 1.171875,
|
5847 |
+
"learning_rate": 1.8465831208685476e-05,
|
5848 |
+
"loss": 6.4449,
|
5849 |
+
"step": 8240
|
5850 |
+
},
|
5851 |
+
{
|
5852 |
+
"epoch": 0.5973787743477107,
|
5853 |
+
"grad_norm": 1.1796875,
|
5854 |
+
"learning_rate": 1.8409253682358678e-05,
|
5855 |
+
"loss": 6.4748,
|
5856 |
+
"step": 8250
|
5857 |
+
},
|
5858 |
+
{
|
5859 |
+
"epoch": 0.5981028698317685,
|
5860 |
+
"grad_norm": 1.2265625,
|
5861 |
+
"learning_rate": 1.835271240942882e-05,
|
5862 |
+
"loss": 6.4677,
|
5863 |
+
"step": 8260
|
5864 |
+
},
|
5865 |
+
{
|
5866 |
+
"epoch": 0.5988269653158264,
|
5867 |
+
"grad_norm": 1.1953125,
|
5868 |
+
"learning_rate": 1.8296207700909717e-05,
|
5869 |
+
"loss": 6.4513,
|
5870 |
+
"step": 8270
|
5871 |
+
},
|
5872 |
+
{
|
5873 |
+
"epoch": 0.5995510607998842,
|
5874 |
+
"grad_norm": 1.1328125,
|
5875 |
+
"learning_rate": 1.823973986761406e-05,
|
5876 |
+
"loss": 6.4442,
|
5877 |
+
"step": 8280
|
5878 |
+
},
|
5879 |
+
{
|
5880 |
+
"epoch": 0.600275156283942,
|
5881 |
+
"grad_norm": 1.09375,
|
5882 |
+
"learning_rate": 1.8183309220151717e-05,
|
5883 |
+
"loss": 6.4554,
|
5884 |
+
"step": 8290
|
5885 |
+
},
|
5886 |
+
{
|
5887 |
+
"epoch": 0.6009992517679998,
|
5888 |
+
"grad_norm": 1.1484375,
|
5889 |
+
"learning_rate": 1.8126916068928e-05,
|
5890 |
+
"loss": 6.4365,
|
5891 |
+
"step": 8300
|
5892 |
+
},
|
5893 |
+
{
|
5894 |
+
"epoch": 0.6017233472520577,
|
5895 |
+
"grad_norm": 1.2109375,
|
5896 |
+
"learning_rate": 1.807056072414195e-05,
|
5897 |
+
"loss": 6.441,
|
5898 |
+
"step": 8310
|
5899 |
+
},
|
5900 |
+
{
|
5901 |
+
"epoch": 0.6024474427361155,
|
5902 |
+
"grad_norm": 1.203125,
|
5903 |
+
"learning_rate": 1.801424349578469e-05,
|
5904 |
+
"loss": 6.4688,
|
5905 |
+
"step": 8320
|
5906 |
+
},
|
5907 |
+
{
|
5908 |
+
"epoch": 0.6031715382201733,
|
5909 |
+
"grad_norm": 1.15625,
|
5910 |
+
"learning_rate": 1.795796469363763e-05,
|
5911 |
+
"loss": 6.4426,
|
5912 |
+
"step": 8330
|
5913 |
+
},
|
5914 |
+
{
|
5915 |
+
"epoch": 0.6038956337042312,
|
5916 |
+
"grad_norm": 1.15625,
|
5917 |
+
"learning_rate": 1.7901724627270833e-05,
|
5918 |
+
"loss": 6.4416,
|
5919 |
+
"step": 8340
|
5920 |
+
},
|
5921 |
+
{
|
5922 |
+
"epoch": 0.604619729188289,
|
5923 |
+
"grad_norm": 1.1953125,
|
5924 |
+
"learning_rate": 1.784552360604131e-05,
|
5925 |
+
"loss": 6.4545,
|
5926 |
+
"step": 8350
|
5927 |
+
},
|
5928 |
+
{
|
5929 |
+
"epoch": 0.6053438246723468,
|
5930 |
+
"grad_norm": 1.171875,
|
5931 |
+
"learning_rate": 1.778936193909126e-05,
|
5932 |
+
"loss": 6.422,
|
5933 |
+
"step": 8360
|
5934 |
+
},
|
5935 |
+
{
|
5936 |
+
"epoch": 0.6060679201564046,
|
5937 |
+
"grad_norm": 1.1796875,
|
5938 |
+
"learning_rate": 1.773323993534644e-05,
|
5939 |
+
"loss": 6.4194,
|
5940 |
+
"step": 8370
|
5941 |
+
},
|
5942 |
+
{
|
5943 |
+
"epoch": 0.6067920156404625,
|
5944 |
+
"grad_norm": 1.1484375,
|
5945 |
+
"learning_rate": 1.7677157903514412e-05,
|
5946 |
+
"loss": 6.4644,
|
5947 |
+
"step": 8380
|
5948 |
+
},
|
5949 |
+
{
|
5950 |
+
"epoch": 0.6075161111245203,
|
5951 |
+
"grad_norm": 1.203125,
|
5952 |
+
"learning_rate": 1.762111615208287e-05,
|
5953 |
+
"loss": 6.4715,
|
5954 |
+
"step": 8390
|
5955 |
+
},
|
5956 |
+
{
|
5957 |
+
"epoch": 0.6082402066085781,
|
5958 |
+
"grad_norm": 1.2421875,
|
5959 |
+
"learning_rate": 1.756511498931795e-05,
|
5960 |
+
"loss": 6.4393,
|
5961 |
+
"step": 8400
|
5962 |
+
},
|
5963 |
+
{
|
5964 |
+
"epoch": 0.608964302092636,
|
5965 |
+
"grad_norm": 1.171875,
|
5966 |
+
"learning_rate": 1.750915472326252e-05,
|
5967 |
+
"loss": 6.4506,
|
5968 |
+
"step": 8410
|
5969 |
+
},
|
5970 |
+
{
|
5971 |
+
"epoch": 0.6096883975766938,
|
5972 |
+
"grad_norm": 1.1875,
|
5973 |
+
"learning_rate": 1.7453235661734478e-05,
|
5974 |
+
"loss": 6.4369,
|
5975 |
+
"step": 8420
|
5976 |
+
},
|
5977 |
+
{
|
5978 |
+
"epoch": 0.6104124930607516,
|
5979 |
+
"grad_norm": 1.171875,
|
5980 |
+
"learning_rate": 1.7397358112325085e-05,
|
5981 |
+
"loss": 6.4276,
|
5982 |
+
"step": 8430
|
5983 |
+
},
|
5984 |
+
{
|
5985 |
+
"epoch": 0.6111365885448095,
|
5986 |
+
"grad_norm": 1.15625,
|
5987 |
+
"learning_rate": 1.7341522382397256e-05,
|
5988 |
+
"loss": 6.4483,
|
5989 |
+
"step": 8440
|
5990 |
+
},
|
5991 |
+
{
|
5992 |
+
"epoch": 0.6118606840288673,
|
5993 |
+
"grad_norm": 1.1328125,
|
5994 |
+
"learning_rate": 1.7285728779083862e-05,
|
5995 |
+
"loss": 6.4641,
|
5996 |
+
"step": 8450
|
5997 |
+
},
|
5998 |
+
{
|
5999 |
+
"epoch": 0.6125847795129251,
|
6000 |
+
"grad_norm": 1.125,
|
6001 |
+
"learning_rate": 1.7229977609286062e-05,
|
6002 |
+
"loss": 6.4545,
|
6003 |
+
"step": 8460
|
6004 |
+
},
|
6005 |
+
{
|
6006 |
+
"epoch": 0.613308874996983,
|
6007 |
+
"grad_norm": 1.1640625,
|
6008 |
+
"learning_rate": 1.7174269179671593e-05,
|
6009 |
+
"loss": 6.454,
|
6010 |
+
"step": 8470
|
6011 |
+
},
|
6012 |
+
{
|
6013 |
+
"epoch": 0.6140329704810408,
|
6014 |
+
"grad_norm": 1.1640625,
|
6015 |
+
"learning_rate": 1.7118603796673102e-05,
|
6016 |
+
"loss": 6.4447,
|
6017 |
+
"step": 8480
|
6018 |
+
},
|
6019 |
+
{
|
6020 |
+
"epoch": 0.6147570659650986,
|
6021 |
+
"grad_norm": 1.21875,
|
6022 |
+
"learning_rate": 1.7062981766486437e-05,
|
6023 |
+
"loss": 6.4641,
|
6024 |
+
"step": 8490
|
6025 |
+
},
|
6026 |
+
{
|
6027 |
+
"epoch": 0.6154811614491564,
|
6028 |
+
"grad_norm": 1.1796875,
|
6029 |
+
"learning_rate": 1.7007403395069005e-05,
|
6030 |
+
"loss": 6.4613,
|
6031 |
+
"step": 8500
|
6032 |
+
},
|
6033 |
+
{
|
6034 |
+
"epoch": 0.6162052569332143,
|
6035 |
+
"grad_norm": 1.2109375,
|
6036 |
+
"learning_rate": 1.695186898813803e-05,
|
6037 |
+
"loss": 6.4616,
|
6038 |
+
"step": 8510
|
6039 |
+
},
|
6040 |
+
{
|
6041 |
+
"epoch": 0.6169293524172721,
|
6042 |
+
"grad_norm": 1.2421875,
|
6043 |
+
"learning_rate": 1.6896378851168914e-05,
|
6044 |
+
"loss": 6.4392,
|
6045 |
+
"step": 8520
|
6046 |
+
},
|
6047 |
+
{
|
6048 |
+
"epoch": 0.6176534479013299,
|
6049 |
+
"grad_norm": 1.109375,
|
6050 |
+
"learning_rate": 1.6840933289393564e-05,
|
6051 |
+
"loss": 6.4388,
|
6052 |
+
"step": 8530
|
6053 |
+
},
|
6054 |
+
{
|
6055 |
+
"epoch": 0.6183775433853878,
|
6056 |
+
"grad_norm": 1.15625,
|
6057 |
+
"learning_rate": 1.6785532607798664e-05,
|
6058 |
+
"loss": 6.4475,
|
6059 |
+
"step": 8540
|
6060 |
+
},
|
6061 |
+
{
|
6062 |
+
"epoch": 0.6191016388694456,
|
6063 |
+
"grad_norm": 1.2421875,
|
6064 |
+
"learning_rate": 1.673017711112405e-05,
|
6065 |
+
"loss": 6.4399,
|
6066 |
+
"step": 8550
|
6067 |
+
},
|
6068 |
+
{
|
6069 |
+
"epoch": 0.6198257343535034,
|
6070 |
+
"grad_norm": 1.1953125,
|
6071 |
+
"learning_rate": 1.6674867103861e-05,
|
6072 |
+
"loss": 6.4491,
|
6073 |
+
"step": 8560
|
6074 |
+
},
|
6075 |
+
{
|
6076 |
+
"epoch": 0.6205498298375612,
|
6077 |
+
"grad_norm": 1.203125,
|
6078 |
+
"learning_rate": 1.6619602890250565e-05,
|
6079 |
+
"loss": 6.4558,
|
6080 |
+
"step": 8570
|
6081 |
+
},
|
6082 |
+
{
|
6083 |
+
"epoch": 0.6212739253216191,
|
6084 |
+
"grad_norm": 1.2421875,
|
6085 |
+
"learning_rate": 1.656438477428193e-05,
|
6086 |
+
"loss": 6.4644,
|
6087 |
+
"step": 8580
|
6088 |
+
},
|
6089 |
+
{
|
6090 |
+
"epoch": 0.6219980208056769,
|
6091 |
+
"grad_norm": 1.1640625,
|
6092 |
+
"learning_rate": 1.6509213059690683e-05,
|
6093 |
+
"loss": 6.4744,
|
6094 |
+
"step": 8590
|
6095 |
+
},
|
6096 |
+
{
|
6097 |
+
"epoch": 0.6227221162897347,
|
6098 |
+
"grad_norm": 1.140625,
|
6099 |
+
"learning_rate": 1.645408804995719e-05,
|
6100 |
+
"loss": 6.4486,
|
6101 |
+
"step": 8600
|
6102 |
+
},
|
6103 |
+
{
|
6104 |
+
"epoch": 0.6234462117737926,
|
6105 |
+
"grad_norm": 1.171875,
|
6106 |
+
"learning_rate": 1.6399010048304897e-05,
|
6107 |
+
"loss": 6.4742,
|
6108 |
+
"step": 8610
|
6109 |
+
},
|
6110 |
+
{
|
6111 |
+
"epoch": 0.6241703072578504,
|
6112 |
+
"grad_norm": 1.1796875,
|
6113 |
+
"learning_rate": 1.6343979357698687e-05,
|
6114 |
+
"loss": 6.4486,
|
6115 |
+
"step": 8620
|
6116 |
+
},
|
6117 |
+
{
|
6118 |
+
"epoch": 0.6248944027419082,
|
6119 |
+
"grad_norm": 1.1640625,
|
6120 |
+
"learning_rate": 1.6288996280843188e-05,
|
6121 |
+
"loss": 6.4436,
|
6122 |
+
"step": 8630
|
6123 |
+
},
|
6124 |
+
{
|
6125 |
+
"epoch": 0.625618498225966,
|
6126 |
+
"grad_norm": 1.2109375,
|
6127 |
+
"learning_rate": 1.6234061120181142e-05,
|
6128 |
+
"loss": 6.4286,
|
6129 |
+
"step": 8640
|
6130 |
+
},
|
6131 |
+
{
|
6132 |
+
"epoch": 0.6263425937100239,
|
6133 |
+
"grad_norm": 1.1796875,
|
6134 |
+
"learning_rate": 1.6179174177891703e-05,
|
6135 |
+
"loss": 6.4383,
|
6136 |
+
"step": 8650
|
6137 |
+
},
|
6138 |
+
{
|
6139 |
+
"epoch": 0.6270666891940817,
|
6140 |
+
"grad_norm": 1.1484375,
|
6141 |
+
"learning_rate": 1.6124335755888797e-05,
|
6142 |
+
"loss": 6.4773,
|
6143 |
+
"step": 8660
|
6144 |
+
},
|
6145 |
+
{
|
6146 |
+
"epoch": 0.6277907846781395,
|
6147 |
+
"grad_norm": 1.203125,
|
6148 |
+
"learning_rate": 1.6069546155819464e-05,
|
6149 |
+
"loss": 6.415,
|
6150 |
+
"step": 8670
|
6151 |
+
},
|
6152 |
+
{
|
6153 |
+
"epoch": 0.6285148801621974,
|
6154 |
+
"grad_norm": 1.1875,
|
6155 |
+
"learning_rate": 1.6014805679062185e-05,
|
6156 |
+
"loss": 6.4114,
|
6157 |
+
"step": 8680
|
6158 |
+
},
|
6159 |
+
{
|
6160 |
+
"epoch": 0.6292389756462552,
|
6161 |
+
"grad_norm": 1.1953125,
|
6162 |
+
"learning_rate": 1.596011462672522e-05,
|
6163 |
+
"loss": 6.4498,
|
6164 |
+
"step": 8690
|
6165 |
+
},
|
6166 |
+
{
|
6167 |
+
"epoch": 0.629963071130313,
|
6168 |
+
"grad_norm": 1.2265625,
|
6169 |
+
"learning_rate": 1.590547329964499e-05,
|
6170 |
+
"loss": 6.4754,
|
6171 |
+
"step": 8700
|
6172 |
+
},
|
6173 |
+
{
|
6174 |
+
"epoch": 0.6306871666143709,
|
6175 |
+
"grad_norm": 1.1953125,
|
6176 |
+
"learning_rate": 1.585088199838437e-05,
|
6177 |
+
"loss": 6.4334,
|
6178 |
+
"step": 8710
|
6179 |
+
},
|
6180 |
+
{
|
6181 |
+
"epoch": 0.6314112620984287,
|
6182 |
+
"grad_norm": 1.21875,
|
6183 |
+
"learning_rate": 1.5796341023231066e-05,
|
6184 |
+
"loss": 6.4284,
|
6185 |
+
"step": 8720
|
6186 |
+
},
|
6187 |
+
{
|
6188 |
+
"epoch": 0.6321353575824865,
|
6189 |
+
"grad_norm": 1.2890625,
|
6190 |
+
"learning_rate": 1.5741850674195975e-05,
|
6191 |
+
"loss": 6.4493,
|
6192 |
+
"step": 8730
|
6193 |
+
},
|
6194 |
+
{
|
6195 |
+
"epoch": 0.6328594530665443,
|
6196 |
+
"grad_norm": 1.15625,
|
6197 |
+
"learning_rate": 1.5687411251011487e-05,
|
6198 |
+
"loss": 6.4531,
|
6199 |
+
"step": 8740
|
6200 |
+
},
|
6201 |
+
{
|
6202 |
+
"epoch": 0.6335835485506022,
|
6203 |
+
"grad_norm": 1.1328125,
|
6204 |
+
"learning_rate": 1.563302305312989e-05,
|
6205 |
+
"loss": 6.4532,
|
6206 |
+
"step": 8750
|
6207 |
+
},
|
6208 |
+
{
|
6209 |
+
"epoch": 0.63430764403466,
|
6210 |
+
"grad_norm": 1.15625,
|
6211 |
+
"learning_rate": 1.5578686379721698e-05,
|
6212 |
+
"loss": 6.4465,
|
6213 |
+
"step": 8760
|
6214 |
+
},
|
6215 |
+
{
|
6216 |
+
"epoch": 0.6350317395187178,
|
6217 |
+
"grad_norm": 1.171875,
|
6218 |
+
"learning_rate": 1.5524401529673994e-05,
|
6219 |
+
"loss": 6.4508,
|
6220 |
+
"step": 8770
|
6221 |
+
},
|
6222 |
+
{
|
6223 |
+
"epoch": 0.6357558350027757,
|
6224 |
+
"grad_norm": 1.1484375,
|
6225 |
+
"learning_rate": 1.5470168801588805e-05,
|
6226 |
+
"loss": 6.4562,
|
6227 |
+
"step": 8780
|
6228 |
+
},
|
6229 |
+
{
|
6230 |
+
"epoch": 0.6364799304868335,
|
6231 |
+
"grad_norm": 1.2109375,
|
6232 |
+
"learning_rate": 1.541598849378145e-05,
|
6233 |
+
"loss": 6.4499,
|
6234 |
+
"step": 8790
|
6235 |
+
},
|
6236 |
+
{
|
6237 |
+
"epoch": 0.6372040259708913,
|
6238 |
+
"grad_norm": 1.1875,
|
6239 |
+
"learning_rate": 1.5361860904278927e-05,
|
6240 |
+
"loss": 6.4544,
|
6241 |
+
"step": 8800
|
6242 |
+
},
|
6243 |
+
{
|
6244 |
+
"epoch": 0.6379281214549491,
|
6245 |
+
"grad_norm": 1.1953125,
|
6246 |
+
"learning_rate": 1.53077863308182e-05,
|
6247 |
+
"loss": 6.4484,
|
6248 |
+
"step": 8810
|
6249 |
+
},
|
6250 |
+
{
|
6251 |
+
"epoch": 0.638652216939007,
|
6252 |
+
"grad_norm": 1.1015625,
|
6253 |
+
"learning_rate": 1.525376507084466e-05,
|
6254 |
+
"loss": 6.4392,
|
6255 |
+
"step": 8820
|
6256 |
+
},
|
6257 |
+
{
|
6258 |
+
"epoch": 0.6393763124230648,
|
6259 |
+
"grad_norm": 1.1875,
|
6260 |
+
"learning_rate": 1.5199797421510415e-05,
|
6261 |
+
"loss": 6.4545,
|
6262 |
+
"step": 8830
|
6263 |
+
},
|
6264 |
+
{
|
6265 |
+
"epoch": 0.6401004079071226,
|
6266 |
+
"grad_norm": 1.203125,
|
6267 |
+
"learning_rate": 1.514588367967268e-05,
|
6268 |
+
"loss": 6.4504,
|
6269 |
+
"step": 8840
|
6270 |
+
},
|
6271 |
+
{
|
6272 |
+
"epoch": 0.6408245033911805,
|
6273 |
+
"grad_norm": 1.1640625,
|
6274 |
+
"learning_rate": 1.5092024141892142e-05,
|
6275 |
+
"loss": 6.4662,
|
6276 |
+
"step": 8850
|
6277 |
+
},
|
6278 |
+
{
|
6279 |
+
"epoch": 0.6415485988752383,
|
6280 |
+
"grad_norm": 1.1484375,
|
6281 |
+
"learning_rate": 1.5038219104431334e-05,
|
6282 |
+
"loss": 6.4296,
|
6283 |
+
"step": 8860
|
6284 |
+
},
|
6285 |
+
{
|
6286 |
+
"epoch": 0.6422726943592961,
|
6287 |
+
"grad_norm": 1.2109375,
|
6288 |
+
"learning_rate": 1.498446886325301e-05,
|
6289 |
+
"loss": 6.4654,
|
6290 |
+
"step": 8870
|
6291 |
+
},
|
6292 |
+
{
|
6293 |
+
"epoch": 0.642996789843354,
|
6294 |
+
"grad_norm": 1.1640625,
|
6295 |
+
"learning_rate": 1.49307737140185e-05,
|
6296 |
+
"loss": 6.4722,
|
6297 |
+
"step": 8880
|
6298 |
+
},
|
6299 |
+
{
|
6300 |
+
"epoch": 0.6437208853274119,
|
6301 |
+
"grad_norm": 1.2265625,
|
6302 |
+
"learning_rate": 1.4877133952086089e-05,
|
6303 |
+
"loss": 6.4471,
|
6304 |
+
"step": 8890
|
6305 |
+
},
|
6306 |
+
{
|
6307 |
+
"epoch": 0.6444449808114697,
|
6308 |
+
"grad_norm": 1.1796875,
|
6309 |
+
"learning_rate": 1.4823549872509396e-05,
|
6310 |
+
"loss": 6.4602,
|
6311 |
+
"step": 8900
|
6312 |
+
},
|
6313 |
+
{
|
6314 |
+
"epoch": 0.6451690762955276,
|
6315 |
+
"grad_norm": 1.171875,
|
6316 |
+
"learning_rate": 1.4770021770035767e-05,
|
6317 |
+
"loss": 6.4455,
|
6318 |
+
"step": 8910
|
6319 |
+
},
|
6320 |
+
{
|
6321 |
+
"epoch": 0.6458931717795854,
|
6322 |
+
"grad_norm": 1.1640625,
|
6323 |
+
"learning_rate": 1.4716549939104618e-05,
|
6324 |
+
"loss": 6.4532,
|
6325 |
+
"step": 8920
|
6326 |
+
},
|
6327 |
+
{
|
6328 |
+
"epoch": 0.6466172672636432,
|
6329 |
+
"grad_norm": 1.1328125,
|
6330 |
+
"learning_rate": 1.4663134673845838e-05,
|
6331 |
+
"loss": 6.4466,
|
6332 |
+
"step": 8930
|
6333 |
+
},
|
6334 |
+
{
|
6335 |
+
"epoch": 0.647341362747701,
|
6336 |
+
"grad_norm": 1.2265625,
|
6337 |
+
"learning_rate": 1.4609776268078184e-05,
|
6338 |
+
"loss": 6.439,
|
6339 |
+
"step": 8940
|
6340 |
+
},
|
6341 |
+
{
|
6342 |
+
"epoch": 0.6480654582317589,
|
6343 |
+
"grad_norm": 1.21875,
|
6344 |
+
"learning_rate": 1.455647501530763e-05,
|
6345 |
+
"loss": 6.4401,
|
6346 |
+
"step": 8950
|
6347 |
+
},
|
6348 |
+
{
|
6349 |
+
"epoch": 0.6487895537158167,
|
6350 |
+
"grad_norm": 1.1171875,
|
6351 |
+
"learning_rate": 1.4503231208725781e-05,
|
6352 |
+
"loss": 6.4521,
|
6353 |
+
"step": 8960
|
6354 |
+
},
|
6355 |
+
{
|
6356 |
+
"epoch": 0.6495136491998745,
|
6357 |
+
"grad_norm": 1.203125,
|
6358 |
+
"learning_rate": 1.4450045141208234e-05,
|
6359 |
+
"loss": 6.452,
|
6360 |
+
"step": 8970
|
6361 |
+
},
|
6362 |
+
{
|
6363 |
+
"epoch": 0.6502377446839324,
|
6364 |
+
"grad_norm": 1.109375,
|
6365 |
+
"learning_rate": 1.4396917105313018e-05,
|
6366 |
+
"loss": 6.481,
|
6367 |
+
"step": 8980
|
6368 |
+
},
|
6369 |
+
{
|
6370 |
+
"epoch": 0.6509618401679902,
|
6371 |
+
"grad_norm": 1.171875,
|
6372 |
+
"learning_rate": 1.4343847393278925e-05,
|
6373 |
+
"loss": 6.474,
|
6374 |
+
"step": 8990
|
6375 |
}
|
6376 |
],
|
6377 |
"logging_steps": 10,
|
|
|
6391 |
"attributes": {}
|
6392 |
}
|
6393 |
},
|
6394 |
+
"total_flos": 1.935261929621055e+19,
|
6395 |
"train_batch_size": 30,
|
6396 |
"trial_name": null,
|
6397 |
"trial_params": null
|