8000
Browse files- {checkpoint-7000 → checkpoint-8000}/README.md +0 -0
- {checkpoint-7000 → checkpoint-8000}/adapter_config.json +0 -0
- {checkpoint-7000 → checkpoint-8000}/adapter_model.safetensors +1 -1
- {checkpoint-7000 → checkpoint-8000}/optimizer.pt +1 -1
- {checkpoint-7000 → checkpoint-8000}/rng_state.pth +1 -1
- {checkpoint-7000 → checkpoint-8000}/scheduler.pt +1 -1
- {checkpoint-7000 → checkpoint-8000}/trainer_state.json +703 -3
- {checkpoint-7000 → checkpoint-8000}/training_args.bin +0 -0
{checkpoint-7000 → checkpoint-8000}/README.md
RENAMED
|
File without changes
|
{checkpoint-7000 → checkpoint-8000}/adapter_config.json
RENAMED
|
File without changes
|
{checkpoint-7000 → checkpoint-8000}/adapter_model.safetensors
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 8535970848
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ad06aee126b0d1bc6ad5f7300e9f866b52d5ddfa03c94459458c802bca279d27
|
| 3 |
size 8535970848
|
{checkpoint-7000 → checkpoint-8000}/optimizer.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 6576969753
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6d92dea01c8c29de2a4005cc9690e2f9d72caf688c782c20775dc59d995cfa7b
|
| 3 |
size 6576969753
|
{checkpoint-7000 → checkpoint-8000}/rng_state.pth
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:51c2939c18e59171b164ef03eea9767004735b3684fc33ed54d0af9e9aa5a8ec
|
| 3 |
size 14645
|
{checkpoint-7000 → checkpoint-8000}/scheduler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8ef9fdb1f0fce4718b7fd8d7a72f390bbaa45bb8e37f4d0f2d8b474443eb5ef2
|
| 3 |
size 1465
|
{checkpoint-7000 → checkpoint-8000}/trainer_state.json
RENAMED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch":
|
| 6 |
"eval_steps": 500,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -4908,6 +4908,706 @@
|
|
| 4908 |
"learning_rate": 0.00017794528798804519,
|
| 4909 |
"loss": 5.3009,
|
| 4910 |
"step": 7000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4911 |
}
|
| 4912 |
],
|
| 4913 |
"logging_steps": 10,
|
|
@@ -4927,7 +5627,7 @@
|
|
| 4927 |
"attributes": {}
|
| 4928 |
}
|
| 4929 |
},
|
| 4930 |
-
"total_flos":
|
| 4931 |
"train_batch_size": 32,
|
| 4932 |
"trial_name": null,
|
| 4933 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 10.230179028132993,
|
| 6 |
"eval_steps": 500,
|
| 7 |
+
"global_step": 8000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 4908 |
"learning_rate": 0.00017794528798804519,
|
| 4909 |
"loss": 5.3009,
|
| 4910 |
"step": 7000
|
| 4911 |
+
},
|
| 4912 |
+
{
|
| 4913 |
+
"epoch": 8.964194373401535,
|
| 4914 |
+
"grad_norm": 1.296583652496338,
|
| 4915 |
+
"learning_rate": 0.00017764536719815918,
|
| 4916 |
+
"loss": 5.3158,
|
| 4917 |
+
"step": 7010
|
| 4918 |
+
},
|
| 4919 |
+
{
|
| 4920 |
+
"epoch": 8.976982097186701,
|
| 4921 |
+
"grad_norm": 1.0986963510513306,
|
| 4922 |
+
"learning_rate": 0.00017734533195522424,
|
| 4923 |
+
"loss": 5.3068,
|
| 4924 |
+
"step": 7020
|
| 4925 |
+
},
|
| 4926 |
+
{
|
| 4927 |
+
"epoch": 8.989769820971867,
|
| 4928 |
+
"grad_norm": 1.0342364311218262,
|
| 4929 |
+
"learning_rate": 0.00017704518350139965,
|
| 4930 |
+
"loss": 5.2997,
|
| 4931 |
+
"step": 7030
|
| 4932 |
+
},
|
| 4933 |
+
{
|
| 4934 |
+
"epoch": 9.002557544757034,
|
| 4935 |
+
"grad_norm": 1.445008635520935,
|
| 4936 |
+
"learning_rate": 0.0001767449230793133,
|
| 4937 |
+
"loss": 5.3086,
|
| 4938 |
+
"step": 7040
|
| 4939 |
+
},
|
| 4940 |
+
{
|
| 4941 |
+
"epoch": 9.0153452685422,
|
| 4942 |
+
"grad_norm": 1.7098116874694824,
|
| 4943 |
+
"learning_rate": 0.00017644455193205666,
|
| 4944 |
+
"loss": 5.3005,
|
| 4945 |
+
"step": 7050
|
| 4946 |
+
},
|
| 4947 |
+
{
|
| 4948 |
+
"epoch": 9.028132992327366,
|
| 4949 |
+
"grad_norm": 0.9708366990089417,
|
| 4950 |
+
"learning_rate": 0.00017614407130317968,
|
| 4951 |
+
"loss": 5.2737,
|
| 4952 |
+
"step": 7060
|
| 4953 |
+
},
|
| 4954 |
+
{
|
| 4955 |
+
"epoch": 9.040920716112533,
|
| 4956 |
+
"grad_norm": 1.089858889579773,
|
| 4957 |
+
"learning_rate": 0.00017584348243668556,
|
| 4958 |
+
"loss": 5.262,
|
| 4959 |
+
"step": 7070
|
| 4960 |
+
},
|
| 4961 |
+
{
|
| 4962 |
+
"epoch": 9.053708439897699,
|
| 4963 |
+
"grad_norm": 2.534782886505127,
|
| 4964 |
+
"learning_rate": 0.00017554278657702549,
|
| 4965 |
+
"loss": 5.2854,
|
| 4966 |
+
"step": 7080
|
| 4967 |
+
},
|
| 4968 |
+
{
|
| 4969 |
+
"epoch": 9.066496163682864,
|
| 4970 |
+
"grad_norm": 1.3707194328308105,
|
| 4971 |
+
"learning_rate": 0.00017524198496909373,
|
| 4972 |
+
"loss": 5.2936,
|
| 4973 |
+
"step": 7090
|
| 4974 |
+
},
|
| 4975 |
+
{
|
| 4976 |
+
"epoch": 9.07928388746803,
|
| 4977 |
+
"grad_norm": 1.4009782075881958,
|
| 4978 |
+
"learning_rate": 0.0001749410788582223,
|
| 4979 |
+
"loss": 5.2795,
|
| 4980 |
+
"step": 7100
|
| 4981 |
+
},
|
| 4982 |
+
{
|
| 4983 |
+
"epoch": 9.092071611253196,
|
| 4984 |
+
"grad_norm": 1.1263126134872437,
|
| 4985 |
+
"learning_rate": 0.00017464006949017584,
|
| 4986 |
+
"loss": 5.2806,
|
| 4987 |
+
"step": 7110
|
| 4988 |
+
},
|
| 4989 |
+
{
|
| 4990 |
+
"epoch": 9.104859335038363,
|
| 4991 |
+
"grad_norm": 0.9099717140197754,
|
| 4992 |
+
"learning_rate": 0.00017433895811114658,
|
| 4993 |
+
"loss": 5.3049,
|
| 4994 |
+
"step": 7120
|
| 4995 |
+
},
|
| 4996 |
+
{
|
| 4997 |
+
"epoch": 9.117647058823529,
|
| 4998 |
+
"grad_norm": 1.3195571899414062,
|
| 4999 |
+
"learning_rate": 0.00017403774596774893,
|
| 5000 |
+
"loss": 5.2803,
|
| 5001 |
+
"step": 7130
|
| 5002 |
+
},
|
| 5003 |
+
{
|
| 5004 |
+
"epoch": 9.130434782608695,
|
| 5005 |
+
"grad_norm": 1.1069055795669556,
|
| 5006 |
+
"learning_rate": 0.00017373643430701463,
|
| 5007 |
+
"loss": 5.2579,
|
| 5008 |
+
"step": 7140
|
| 5009 |
+
},
|
| 5010 |
+
{
|
| 5011 |
+
"epoch": 9.143222506393862,
|
| 5012 |
+
"grad_norm": 1.1975563764572144,
|
| 5013 |
+
"learning_rate": 0.00017343502437638727,
|
| 5014 |
+
"loss": 5.2795,
|
| 5015 |
+
"step": 7150
|
| 5016 |
+
},
|
| 5017 |
+
{
|
| 5018 |
+
"epoch": 9.156010230179028,
|
| 5019 |
+
"grad_norm": 1.6220418214797974,
|
| 5020 |
+
"learning_rate": 0.00017313351742371746,
|
| 5021 |
+
"loss": 5.2797,
|
| 5022 |
+
"step": 7160
|
| 5023 |
+
},
|
| 5024 |
+
{
|
| 5025 |
+
"epoch": 9.168797953964194,
|
| 5026 |
+
"grad_norm": 1.490394115447998,
|
| 5027 |
+
"learning_rate": 0.00017283191469725728,
|
| 5028 |
+
"loss": 5.2768,
|
| 5029 |
+
"step": 7170
|
| 5030 |
+
},
|
| 5031 |
+
{
|
| 5032 |
+
"epoch": 9.18158567774936,
|
| 5033 |
+
"grad_norm": 1.4838827848434448,
|
| 5034 |
+
"learning_rate": 0.00017253021744565548,
|
| 5035 |
+
"loss": 5.2871,
|
| 5036 |
+
"step": 7180
|
| 5037 |
+
},
|
| 5038 |
+
{
|
| 5039 |
+
"epoch": 9.194373401534527,
|
| 5040 |
+
"grad_norm": 5.531613826751709,
|
| 5041 |
+
"learning_rate": 0.0001722284269179521,
|
| 5042 |
+
"loss": 5.2537,
|
| 5043 |
+
"step": 7190
|
| 5044 |
+
},
|
| 5045 |
+
{
|
| 5046 |
+
"epoch": 9.207161125319693,
|
| 5047 |
+
"grad_norm": 1.5700544118881226,
|
| 5048 |
+
"learning_rate": 0.0001719265443635733,
|
| 5049 |
+
"loss": 5.2875,
|
| 5050 |
+
"step": 7200
|
| 5051 |
+
},
|
| 5052 |
+
{
|
| 5053 |
+
"epoch": 9.21994884910486,
|
| 5054 |
+
"grad_norm": 1.2028346061706543,
|
| 5055 |
+
"learning_rate": 0.00017162457103232632,
|
| 5056 |
+
"loss": 5.2707,
|
| 5057 |
+
"step": 7210
|
| 5058 |
+
},
|
| 5059 |
+
{
|
| 5060 |
+
"epoch": 9.232736572890026,
|
| 5061 |
+
"grad_norm": 1.9414821863174438,
|
| 5062 |
+
"learning_rate": 0.00017132250817439412,
|
| 5063 |
+
"loss": 5.2918,
|
| 5064 |
+
"step": 7220
|
| 5065 |
+
},
|
| 5066 |
+
{
|
| 5067 |
+
"epoch": 9.245524296675192,
|
| 5068 |
+
"grad_norm": 3.8366823196411133,
|
| 5069 |
+
"learning_rate": 0.00017102035704033038,
|
| 5070 |
+
"loss": 5.277,
|
| 5071 |
+
"step": 7230
|
| 5072 |
+
},
|
| 5073 |
+
{
|
| 5074 |
+
"epoch": 9.258312020460359,
|
| 5075 |
+
"grad_norm": 41.69551086425781,
|
| 5076 |
+
"learning_rate": 0.0001707181188810542,
|
| 5077 |
+
"loss": 5.2691,
|
| 5078 |
+
"step": 7240
|
| 5079 |
+
},
|
| 5080 |
+
{
|
| 5081 |
+
"epoch": 9.271099744245525,
|
| 5082 |
+
"grad_norm": 2.0435402393341064,
|
| 5083 |
+
"learning_rate": 0.00017041579494784506,
|
| 5084 |
+
"loss": 5.3075,
|
| 5085 |
+
"step": 7250
|
| 5086 |
+
},
|
| 5087 |
+
{
|
| 5088 |
+
"epoch": 9.28388746803069,
|
| 5089 |
+
"grad_norm": 2.1493489742279053,
|
| 5090 |
+
"learning_rate": 0.00017011338649233743,
|
| 5091 |
+
"loss": 5.3234,
|
| 5092 |
+
"step": 7260
|
| 5093 |
+
},
|
| 5094 |
+
{
|
| 5095 |
+
"epoch": 9.296675191815856,
|
| 5096 |
+
"grad_norm": 3.627615213394165,
|
| 5097 |
+
"learning_rate": 0.0001698108947665158,
|
| 5098 |
+
"loss": 5.3018,
|
| 5099 |
+
"step": 7270
|
| 5100 |
+
},
|
| 5101 |
+
{
|
| 5102 |
+
"epoch": 9.309462915601022,
|
| 5103 |
+
"grad_norm": 24.722545623779297,
|
| 5104 |
+
"learning_rate": 0.00016950832102270927,
|
| 5105 |
+
"loss": 5.3123,
|
| 5106 |
+
"step": 7280
|
| 5107 |
+
},
|
| 5108 |
+
{
|
| 5109 |
+
"epoch": 9.322250639386189,
|
| 5110 |
+
"grad_norm": 19.32564353942871,
|
| 5111 |
+
"learning_rate": 0.00016920566651358666,
|
| 5112 |
+
"loss": 5.346,
|
| 5113 |
+
"step": 7290
|
| 5114 |
+
},
|
| 5115 |
+
{
|
| 5116 |
+
"epoch": 9.335038363171355,
|
| 5117 |
+
"grad_norm": 173.1358184814453,
|
| 5118 |
+
"learning_rate": 0.00016890293249215109,
|
| 5119 |
+
"loss": 5.3385,
|
| 5120 |
+
"step": 7300
|
| 5121 |
+
},
|
| 5122 |
+
{
|
| 5123 |
+
"epoch": 9.347826086956522,
|
| 5124 |
+
"grad_norm": 139.24111938476562,
|
| 5125 |
+
"learning_rate": 0.0001686001202117348,
|
| 5126 |
+
"loss": 5.3411,
|
| 5127 |
+
"step": 7310
|
| 5128 |
+
},
|
| 5129 |
+
{
|
| 5130 |
+
"epoch": 9.360613810741688,
|
| 5131 |
+
"grad_norm": 12987.6923828125,
|
| 5132 |
+
"learning_rate": 0.00016829723092599418,
|
| 5133 |
+
"loss": 5.3288,
|
| 5134 |
+
"step": 7320
|
| 5135 |
+
},
|
| 5136 |
+
{
|
| 5137 |
+
"epoch": 9.373401534526854,
|
| 5138 |
+
"grad_norm": 53.43489074707031,
|
| 5139 |
+
"learning_rate": 0.00016799426588890427,
|
| 5140 |
+
"loss": 5.3403,
|
| 5141 |
+
"step": 7330
|
| 5142 |
+
},
|
| 5143 |
+
{
|
| 5144 |
+
"epoch": 9.38618925831202,
|
| 5145 |
+
"grad_norm": 29.375526428222656,
|
| 5146 |
+
"learning_rate": 0.00016769122635475385,
|
| 5147 |
+
"loss": 5.3186,
|
| 5148 |
+
"step": 7340
|
| 5149 |
+
},
|
| 5150 |
+
{
|
| 5151 |
+
"epoch": 9.398976982097187,
|
| 5152 |
+
"grad_norm": 47.31606674194336,
|
| 5153 |
+
"learning_rate": 0.00016738811357813998,
|
| 5154 |
+
"loss": 5.3178,
|
| 5155 |
+
"step": 7350
|
| 5156 |
+
},
|
| 5157 |
+
{
|
| 5158 |
+
"epoch": 9.411764705882353,
|
| 5159 |
+
"grad_norm": 12.416561126708984,
|
| 5160 |
+
"learning_rate": 0.00016708492881396307,
|
| 5161 |
+
"loss": 5.3385,
|
| 5162 |
+
"step": 7360
|
| 5163 |
+
},
|
| 5164 |
+
{
|
| 5165 |
+
"epoch": 9.42455242966752,
|
| 5166 |
+
"grad_norm": 9.562813758850098,
|
| 5167 |
+
"learning_rate": 0.0001667816733174215,
|
| 5168 |
+
"loss": 5.3481,
|
| 5169 |
+
"step": 7370
|
| 5170 |
+
},
|
| 5171 |
+
{
|
| 5172 |
+
"epoch": 9.437340153452686,
|
| 5173 |
+
"grad_norm": 3.935084819793701,
|
| 5174 |
+
"learning_rate": 0.00016647834834400654,
|
| 5175 |
+
"loss": 5.3439,
|
| 5176 |
+
"step": 7380
|
| 5177 |
+
},
|
| 5178 |
+
{
|
| 5179 |
+
"epoch": 9.450127877237852,
|
| 5180 |
+
"grad_norm": 2.569079875946045,
|
| 5181 |
+
"learning_rate": 0.00016617495514949704,
|
| 5182 |
+
"loss": 5.3225,
|
| 5183 |
+
"step": 7390
|
| 5184 |
+
},
|
| 5185 |
+
{
|
| 5186 |
+
"epoch": 9.462915601023019,
|
| 5187 |
+
"grad_norm": 1.732500672340393,
|
| 5188 |
+
"learning_rate": 0.0001658714949899543,
|
| 5189 |
+
"loss": 5.3235,
|
| 5190 |
+
"step": 7400
|
| 5191 |
+
},
|
| 5192 |
+
{
|
| 5193 |
+
"epoch": 9.475703324808185,
|
| 5194 |
+
"grad_norm": 1.2219961881637573,
|
| 5195 |
+
"learning_rate": 0.00016556796912171689,
|
| 5196 |
+
"loss": 5.3413,
|
| 5197 |
+
"step": 7410
|
| 5198 |
+
},
|
| 5199 |
+
{
|
| 5200 |
+
"epoch": 9.48849104859335,
|
| 5201 |
+
"grad_norm": 0.9043082594871521,
|
| 5202 |
+
"learning_rate": 0.00016526437880139537,
|
| 5203 |
+
"loss": 5.288,
|
| 5204 |
+
"step": 7420
|
| 5205 |
+
},
|
| 5206 |
+
{
|
| 5207 |
+
"epoch": 9.501278772378516,
|
| 5208 |
+
"grad_norm": 1.0241756439208984,
|
| 5209 |
+
"learning_rate": 0.0001649607252858672,
|
| 5210 |
+
"loss": 5.302,
|
| 5211 |
+
"step": 7430
|
| 5212 |
+
},
|
| 5213 |
+
{
|
| 5214 |
+
"epoch": 9.514066496163682,
|
| 5215 |
+
"grad_norm": 1.5669431686401367,
|
| 5216 |
+
"learning_rate": 0.00016465700983227138,
|
| 5217 |
+
"loss": 5.2899,
|
| 5218 |
+
"step": 7440
|
| 5219 |
+
},
|
| 5220 |
+
{
|
| 5221 |
+
"epoch": 9.526854219948849,
|
| 5222 |
+
"grad_norm": 1.5617988109588623,
|
| 5223 |
+
"learning_rate": 0.00016435323369800344,
|
| 5224 |
+
"loss": 5.2868,
|
| 5225 |
+
"step": 7450
|
| 5226 |
+
},
|
| 5227 |
+
{
|
| 5228 |
+
"epoch": 9.539641943734015,
|
| 5229 |
+
"grad_norm": 1.1847914457321167,
|
| 5230 |
+
"learning_rate": 0.00016404939814071003,
|
| 5231 |
+
"loss": 5.2617,
|
| 5232 |
+
"step": 7460
|
| 5233 |
+
},
|
| 5234 |
+
{
|
| 5235 |
+
"epoch": 9.552429667519181,
|
| 5236 |
+
"grad_norm": 0.8560781478881836,
|
| 5237 |
+
"learning_rate": 0.0001637455044182839,
|
| 5238 |
+
"loss": 5.2855,
|
| 5239 |
+
"step": 7470
|
| 5240 |
+
},
|
| 5241 |
+
{
|
| 5242 |
+
"epoch": 9.565217391304348,
|
| 5243 |
+
"grad_norm": 0.9926068782806396,
|
| 5244 |
+
"learning_rate": 0.0001634415537888585,
|
| 5245 |
+
"loss": 5.265,
|
| 5246 |
+
"step": 7480
|
| 5247 |
+
},
|
| 5248 |
+
{
|
| 5249 |
+
"epoch": 9.578005115089514,
|
| 5250 |
+
"grad_norm": 2.2098798751831055,
|
| 5251 |
+
"learning_rate": 0.00016313754751080302,
|
| 5252 |
+
"loss": 5.2773,
|
| 5253 |
+
"step": 7490
|
| 5254 |
+
},
|
| 5255 |
+
{
|
| 5256 |
+
"epoch": 9.59079283887468,
|
| 5257 |
+
"grad_norm": 1.3162308931350708,
|
| 5258 |
+
"learning_rate": 0.00016283348684271694,
|
| 5259 |
+
"loss": 5.276,
|
| 5260 |
+
"step": 7500
|
| 5261 |
+
},
|
| 5262 |
+
{
|
| 5263 |
+
"epoch": 9.603580562659847,
|
| 5264 |
+
"grad_norm": 1.2072679996490479,
|
| 5265 |
+
"learning_rate": 0.00016252937304342494,
|
| 5266 |
+
"loss": 5.2825,
|
| 5267 |
+
"step": 7510
|
| 5268 |
+
},
|
| 5269 |
+
{
|
| 5270 |
+
"epoch": 9.616368286445013,
|
| 5271 |
+
"grad_norm": 1.212632656097412,
|
| 5272 |
+
"learning_rate": 0.0001622252073719717,
|
| 5273 |
+
"loss": 5.2609,
|
| 5274 |
+
"step": 7520
|
| 5275 |
+
},
|
| 5276 |
+
{
|
| 5277 |
+
"epoch": 9.62915601023018,
|
| 5278 |
+
"grad_norm": 1.244361400604248,
|
| 5279 |
+
"learning_rate": 0.0001619209910876165,
|
| 5280 |
+
"loss": 5.247,
|
| 5281 |
+
"step": 7530
|
| 5282 |
+
},
|
| 5283 |
+
{
|
| 5284 |
+
"epoch": 9.641943734015346,
|
| 5285 |
+
"grad_norm": 1.2935197353363037,
|
| 5286 |
+
"learning_rate": 0.00016161672544982842,
|
| 5287 |
+
"loss": 5.2666,
|
| 5288 |
+
"step": 7540
|
| 5289 |
+
},
|
| 5290 |
+
{
|
| 5291 |
+
"epoch": 9.654731457800512,
|
| 5292 |
+
"grad_norm": 1.270849347114563,
|
| 5293 |
+
"learning_rate": 0.00016131241171828063,
|
| 5294 |
+
"loss": 5.2556,
|
| 5295 |
+
"step": 7550
|
| 5296 |
+
},
|
| 5297 |
+
{
|
| 5298 |
+
"epoch": 9.667519181585678,
|
| 5299 |
+
"grad_norm": 1.5244646072387695,
|
| 5300 |
+
"learning_rate": 0.00016100805115284555,
|
| 5301 |
+
"loss": 5.2594,
|
| 5302 |
+
"step": 7560
|
| 5303 |
+
},
|
| 5304 |
+
{
|
| 5305 |
+
"epoch": 9.680306905370845,
|
| 5306 |
+
"grad_norm": 1.245428204536438,
|
| 5307 |
+
"learning_rate": 0.00016070364501358944,
|
| 5308 |
+
"loss": 5.2452,
|
| 5309 |
+
"step": 7570
|
| 5310 |
+
},
|
| 5311 |
+
{
|
| 5312 |
+
"epoch": 9.693094629156011,
|
| 5313 |
+
"grad_norm": 2.1988797187805176,
|
| 5314 |
+
"learning_rate": 0.00016039919456076727,
|
| 5315 |
+
"loss": 5.289,
|
| 5316 |
+
"step": 7580
|
| 5317 |
+
},
|
| 5318 |
+
{
|
| 5319 |
+
"epoch": 9.705882352941176,
|
| 5320 |
+
"grad_norm": 4.7878031730651855,
|
| 5321 |
+
"learning_rate": 0.00016009470105481736,
|
| 5322 |
+
"loss": 5.2933,
|
| 5323 |
+
"step": 7590
|
| 5324 |
+
},
|
| 5325 |
+
{
|
| 5326 |
+
"epoch": 9.718670076726342,
|
| 5327 |
+
"grad_norm": 1.0919615030288696,
|
| 5328 |
+
"learning_rate": 0.00015979016575635644,
|
| 5329 |
+
"loss": 5.2634,
|
| 5330 |
+
"step": 7600
|
| 5331 |
+
},
|
| 5332 |
+
{
|
| 5333 |
+
"epoch": 9.731457800511508,
|
| 5334 |
+
"grad_norm": 1.3478410243988037,
|
| 5335 |
+
"learning_rate": 0.00015948558992617416,
|
| 5336 |
+
"loss": 5.2808,
|
| 5337 |
+
"step": 7610
|
| 5338 |
+
},
|
| 5339 |
+
{
|
| 5340 |
+
"epoch": 9.744245524296675,
|
| 5341 |
+
"grad_norm": 1.088616132736206,
|
| 5342 |
+
"learning_rate": 0.00015918097482522798,
|
| 5343 |
+
"loss": 5.2656,
|
| 5344 |
+
"step": 7620
|
| 5345 |
+
},
|
| 5346 |
+
{
|
| 5347 |
+
"epoch": 9.757033248081841,
|
| 5348 |
+
"grad_norm": 1.4594988822937012,
|
| 5349 |
+
"learning_rate": 0.00015887632171463794,
|
| 5350 |
+
"loss": 5.2422,
|
| 5351 |
+
"step": 7630
|
| 5352 |
+
},
|
| 5353 |
+
{
|
| 5354 |
+
"epoch": 9.769820971867007,
|
| 5355 |
+
"grad_norm": 1.346463918685913,
|
| 5356 |
+
"learning_rate": 0.00015857163185568153,
|
| 5357 |
+
"loss": 5.2668,
|
| 5358 |
+
"step": 7640
|
| 5359 |
+
},
|
| 5360 |
+
{
|
| 5361 |
+
"epoch": 9.782608695652174,
|
| 5362 |
+
"grad_norm": 2.3005871772766113,
|
| 5363 |
+
"learning_rate": 0.00015826690650978825,
|
| 5364 |
+
"loss": 5.2295,
|
| 5365 |
+
"step": 7650
|
| 5366 |
+
},
|
| 5367 |
+
{
|
| 5368 |
+
"epoch": 9.79539641943734,
|
| 5369 |
+
"grad_norm": 1.372439980506897,
|
| 5370 |
+
"learning_rate": 0.0001579621469385346,
|
| 5371 |
+
"loss": 5.2419,
|
| 5372 |
+
"step": 7660
|
| 5373 |
+
},
|
| 5374 |
+
{
|
| 5375 |
+
"epoch": 9.808184143222507,
|
| 5376 |
+
"grad_norm": 1.0069390535354614,
|
| 5377 |
+
"learning_rate": 0.00015765735440363872,
|
| 5378 |
+
"loss": 5.2692,
|
| 5379 |
+
"step": 7670
|
| 5380 |
+
},
|
| 5381 |
+
{
|
| 5382 |
+
"epoch": 9.820971867007673,
|
| 5383 |
+
"grad_norm": 1.3591490983963013,
|
| 5384 |
+
"learning_rate": 0.00015735253016695527,
|
| 5385 |
+
"loss": 5.269,
|
| 5386 |
+
"step": 7680
|
| 5387 |
+
},
|
| 5388 |
+
{
|
| 5389 |
+
"epoch": 9.83375959079284,
|
| 5390 |
+
"grad_norm": 1.47946298122406,
|
| 5391 |
+
"learning_rate": 0.00015704767549047015,
|
| 5392 |
+
"loss": 5.2615,
|
| 5393 |
+
"step": 7690
|
| 5394 |
+
},
|
| 5395 |
+
{
|
| 5396 |
+
"epoch": 9.846547314578006,
|
| 5397 |
+
"grad_norm": 2.2882235050201416,
|
| 5398 |
+
"learning_rate": 0.00015674279163629528,
|
| 5399 |
+
"loss": 5.2452,
|
| 5400 |
+
"step": 7700
|
| 5401 |
+
},
|
| 5402 |
+
{
|
| 5403 |
+
"epoch": 9.859335038363172,
|
| 5404 |
+
"grad_norm": 1.406764268875122,
|
| 5405 |
+
"learning_rate": 0.00015643787986666333,
|
| 5406 |
+
"loss": 5.2515,
|
| 5407 |
+
"step": 7710
|
| 5408 |
+
},
|
| 5409 |
+
{
|
| 5410 |
+
"epoch": 9.872122762148338,
|
| 5411 |
+
"grad_norm": 0.9985896348953247,
|
| 5412 |
+
"learning_rate": 0.00015613294144392256,
|
| 5413 |
+
"loss": 5.2536,
|
| 5414 |
+
"step": 7720
|
| 5415 |
+
},
|
| 5416 |
+
{
|
| 5417 |
+
"epoch": 9.884910485933505,
|
| 5418 |
+
"grad_norm": 1.1201996803283691,
|
| 5419 |
+
"learning_rate": 0.00015582797763053166,
|
| 5420 |
+
"loss": 5.2459,
|
| 5421 |
+
"step": 7730
|
| 5422 |
+
},
|
| 5423 |
+
{
|
| 5424 |
+
"epoch": 9.89769820971867,
|
| 5425 |
+
"grad_norm": 1.2289462089538574,
|
| 5426 |
+
"learning_rate": 0.00015552298968905432,
|
| 5427 |
+
"loss": 5.2162,
|
| 5428 |
+
"step": 7740
|
| 5429 |
+
},
|
| 5430 |
+
{
|
| 5431 |
+
"epoch": 9.910485933503836,
|
| 5432 |
+
"grad_norm": 1.3729525804519653,
|
| 5433 |
+
"learning_rate": 0.00015521797888215424,
|
| 5434 |
+
"loss": 5.2488,
|
| 5435 |
+
"step": 7750
|
| 5436 |
+
},
|
| 5437 |
+
{
|
| 5438 |
+
"epoch": 9.923273657289002,
|
| 5439 |
+
"grad_norm": 1.3408769369125366,
|
| 5440 |
+
"learning_rate": 0.00015491294647258967,
|
| 5441 |
+
"loss": 5.2608,
|
| 5442 |
+
"step": 7760
|
| 5443 |
+
},
|
| 5444 |
+
{
|
| 5445 |
+
"epoch": 9.936061381074168,
|
| 5446 |
+
"grad_norm": 1.3026928901672363,
|
| 5447 |
+
"learning_rate": 0.0001546078937232083,
|
| 5448 |
+
"loss": 5.227,
|
| 5449 |
+
"step": 7770
|
| 5450 |
+
},
|
| 5451 |
+
{
|
| 5452 |
+
"epoch": 9.948849104859335,
|
| 5453 |
+
"grad_norm": 1.3715628385543823,
|
| 5454 |
+
"learning_rate": 0.00015430282189694212,
|
| 5455 |
+
"loss": 5.2677,
|
| 5456 |
+
"step": 7780
|
| 5457 |
+
},
|
| 5458 |
+
{
|
| 5459 |
+
"epoch": 9.961636828644501,
|
| 5460 |
+
"grad_norm": 1.0734012126922607,
|
| 5461 |
+
"learning_rate": 0.00015399773225680208,
|
| 5462 |
+
"loss": 5.2575,
|
| 5463 |
+
"step": 7790
|
| 5464 |
+
},
|
| 5465 |
+
{
|
| 5466 |
+
"epoch": 9.974424552429667,
|
| 5467 |
+
"grad_norm": 1.0179634094238281,
|
| 5468 |
+
"learning_rate": 0.00015369262606587281,
|
| 5469 |
+
"loss": 5.2117,
|
| 5470 |
+
"step": 7800
|
| 5471 |
+
},
|
| 5472 |
+
{
|
| 5473 |
+
"epoch": 9.987212276214834,
|
| 5474 |
+
"grad_norm": 1.2135021686553955,
|
| 5475 |
+
"learning_rate": 0.00015338750458730746,
|
| 5476 |
+
"loss": 5.2387,
|
| 5477 |
+
"step": 7810
|
| 5478 |
+
},
|
| 5479 |
+
{
|
| 5480 |
+
"epoch": 10.0,
|
| 5481 |
+
"grad_norm": 1.7299256324768066,
|
| 5482 |
+
"learning_rate": 0.00015308236908432264,
|
| 5483 |
+
"loss": 5.2517,
|
| 5484 |
+
"step": 7820
|
| 5485 |
+
},
|
| 5486 |
+
{
|
| 5487 |
+
"epoch": 10.012787723785166,
|
| 5488 |
+
"grad_norm": 1.3696908950805664,
|
| 5489 |
+
"learning_rate": 0.00015277722082019272,
|
| 5490 |
+
"loss": 5.2373,
|
| 5491 |
+
"step": 7830
|
| 5492 |
+
},
|
| 5493 |
+
{
|
| 5494 |
+
"epoch": 10.025575447570333,
|
| 5495 |
+
"grad_norm": 1.4967882633209229,
|
| 5496 |
+
"learning_rate": 0.00015247206105824522,
|
| 5497 |
+
"loss": 5.2045,
|
| 5498 |
+
"step": 7840
|
| 5499 |
+
},
|
| 5500 |
+
{
|
| 5501 |
+
"epoch": 10.038363171355499,
|
| 5502 |
+
"grad_norm": 1.4495280981063843,
|
| 5503 |
+
"learning_rate": 0.00015216689106185505,
|
| 5504 |
+
"loss": 5.2286,
|
| 5505 |
+
"step": 7850
|
| 5506 |
+
},
|
| 5507 |
+
{
|
| 5508 |
+
"epoch": 10.051150895140665,
|
| 5509 |
+
"grad_norm": 2.255189895629883,
|
| 5510 |
+
"learning_rate": 0.00015186171209443958,
|
| 5511 |
+
"loss": 5.1753,
|
| 5512 |
+
"step": 7860
|
| 5513 |
+
},
|
| 5514 |
+
{
|
| 5515 |
+
"epoch": 10.063938618925832,
|
| 5516 |
+
"grad_norm": 1.22813081741333,
|
| 5517 |
+
"learning_rate": 0.00015155652541945326,
|
| 5518 |
+
"loss": 5.2259,
|
| 5519 |
+
"step": 7870
|
| 5520 |
+
},
|
| 5521 |
+
{
|
| 5522 |
+
"epoch": 10.076726342710998,
|
| 5523 |
+
"grad_norm": 1.1559652090072632,
|
| 5524 |
+
"learning_rate": 0.00015125133230038256,
|
| 5525 |
+
"loss": 5.1997,
|
| 5526 |
+
"step": 7880
|
| 5527 |
+
},
|
| 5528 |
+
{
|
| 5529 |
+
"epoch": 10.089514066496164,
|
| 5530 |
+
"grad_norm": 1.5692169666290283,
|
| 5531 |
+
"learning_rate": 0.00015094613400074052,
|
| 5532 |
+
"loss": 5.2184,
|
| 5533 |
+
"step": 7890
|
| 5534 |
+
},
|
| 5535 |
+
{
|
| 5536 |
+
"epoch": 10.10230179028133,
|
| 5537 |
+
"grad_norm": 1.143264651298523,
|
| 5538 |
+
"learning_rate": 0.00015064093178406165,
|
| 5539 |
+
"loss": 5.2217,
|
| 5540 |
+
"step": 7900
|
| 5541 |
+
},
|
| 5542 |
+
{
|
| 5543 |
+
"epoch": 10.115089514066495,
|
| 5544 |
+
"grad_norm": 1.3026010990142822,
|
| 5545 |
+
"learning_rate": 0.00015033572691389673,
|
| 5546 |
+
"loss": 5.2219,
|
| 5547 |
+
"step": 7910
|
| 5548 |
+
},
|
| 5549 |
+
{
|
| 5550 |
+
"epoch": 10.127877237851662,
|
| 5551 |
+
"grad_norm": 1.255898356437683,
|
| 5552 |
+
"learning_rate": 0.00015003052065380742,
|
| 5553 |
+
"loss": 5.1964,
|
| 5554 |
+
"step": 7920
|
| 5555 |
+
},
|
| 5556 |
+
{
|
| 5557 |
+
"epoch": 10.140664961636828,
|
| 5558 |
+
"grad_norm": 1.5015469789505005,
|
| 5559 |
+
"learning_rate": 0.00014972531426736131,
|
| 5560 |
+
"loss": 5.241,
|
| 5561 |
+
"step": 7930
|
| 5562 |
+
},
|
| 5563 |
+
{
|
| 5564 |
+
"epoch": 10.153452685421994,
|
| 5565 |
+
"grad_norm": 1.5134257078170776,
|
| 5566 |
+
"learning_rate": 0.0001494201090181263,
|
| 5567 |
+
"loss": 5.2255,
|
| 5568 |
+
"step": 7940
|
| 5569 |
+
},
|
| 5570 |
+
{
|
| 5571 |
+
"epoch": 10.16624040920716,
|
| 5572 |
+
"grad_norm": 1.295350432395935,
|
| 5573 |
+
"learning_rate": 0.00014911490616966575,
|
| 5574 |
+
"loss": 5.2201,
|
| 5575 |
+
"step": 7950
|
| 5576 |
+
},
|
| 5577 |
+
{
|
| 5578 |
+
"epoch": 10.179028132992327,
|
| 5579 |
+
"grad_norm": 1.8612785339355469,
|
| 5580 |
+
"learning_rate": 0.000148809706985533,
|
| 5581 |
+
"loss": 5.2109,
|
| 5582 |
+
"step": 7960
|
| 5583 |
+
},
|
| 5584 |
+
{
|
| 5585 |
+
"epoch": 10.191815856777493,
|
| 5586 |
+
"grad_norm": 1.5443718433380127,
|
| 5587 |
+
"learning_rate": 0.0001485045127292662,
|
| 5588 |
+
"loss": 5.2247,
|
| 5589 |
+
"step": 7970
|
| 5590 |
+
},
|
| 5591 |
+
{
|
| 5592 |
+
"epoch": 10.20460358056266,
|
| 5593 |
+
"grad_norm": 1.7104902267456055,
|
| 5594 |
+
"learning_rate": 0.00014819932466438317,
|
| 5595 |
+
"loss": 5.2141,
|
| 5596 |
+
"step": 7980
|
| 5597 |
+
},
|
| 5598 |
+
{
|
| 5599 |
+
"epoch": 10.217391304347826,
|
| 5600 |
+
"grad_norm": 1.7787065505981445,
|
| 5601 |
+
"learning_rate": 0.00014789414405437607,
|
| 5602 |
+
"loss": 5.1911,
|
| 5603 |
+
"step": 7990
|
| 5604 |
+
},
|
| 5605 |
+
{
|
| 5606 |
+
"epoch": 10.230179028132993,
|
| 5607 |
+
"grad_norm": 1.0850389003753662,
|
| 5608 |
+
"learning_rate": 0.0001475889721627062,
|
| 5609 |
+
"loss": 5.2156,
|
| 5610 |
+
"step": 8000
|
| 5611 |
}
|
| 5612 |
],
|
| 5613 |
"logging_steps": 10,
|
|
|
|
| 5627 |
"attributes": {}
|
| 5628 |
}
|
| 5629 |
},
|
| 5630 |
+
"total_flos": 1.0590849988794814e+19,
|
| 5631 |
"train_batch_size": 32,
|
| 5632 |
"trial_name": null,
|
| 5633 |
"trial_params": null
|
{checkpoint-7000 → checkpoint-8000}/training_args.bin
RENAMED
|
File without changes
|