Training in progress, step 3250
Browse files- adapter_model.safetensors +1 -1
- metrics.json +1 -0
- state.json +18 -3
adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 35668592
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ac1e4760765910671b42368464e3e42417ac29fd4485638ff98e15b434b7fa5f
|
3 |
size 35668592
|
metrics.json
CHANGED
@@ -62,3 +62,4 @@
|
|
62 |
{"Step":3100,"eval_loss":1.7509515285,"eval_runtime":29.3774,"eval_samples_per_second":3.404,"eval_steps_per_second":0.443,"epoch":2.4351924588}
|
63 |
{"Step":3150,"eval_loss":1.7521715164,"eval_runtime":29.4311,"eval_samples_per_second":3.398,"eval_steps_per_second":0.442,"epoch":2.4744697565}
|
64 |
{"Step":3200,"eval_loss":1.7496337891,"eval_runtime":29.4486,"eval_samples_per_second":3.396,"eval_steps_per_second":0.441,"epoch":2.5137470542}
|
|
|
|
62 |
{"Step":3100,"eval_loss":1.7509515285,"eval_runtime":29.3774,"eval_samples_per_second":3.404,"eval_steps_per_second":0.443,"epoch":2.4351924588}
|
63 |
{"Step":3150,"eval_loss":1.7521715164,"eval_runtime":29.4311,"eval_samples_per_second":3.398,"eval_steps_per_second":0.442,"epoch":2.4744697565}
|
64 |
{"Step":3200,"eval_loss":1.7496337891,"eval_runtime":29.4486,"eval_samples_per_second":3.396,"eval_steps_per_second":0.441,"epoch":2.5137470542}
|
65 |
+
{"Step":3250,"eval_loss":1.7441329956,"eval_runtime":29.3681,"eval_samples_per_second":3.405,"eval_steps_per_second":0.443,"epoch":2.5530243519}
|
state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 2.
|
5 |
"eval_steps": 50,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -952,6 +952,21 @@
|
|
952 |
"eval_samples_per_second": 3.398,
|
953 |
"eval_steps_per_second": 0.442,
|
954 |
"step": 3150
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
955 |
}
|
956 |
],
|
957 |
"logging_steps": 50,
|
@@ -971,7 +986,7 @@
|
|
971 |
"attributes": {}
|
972 |
}
|
973 |
},
|
974 |
-
"total_flos":
|
975 |
"train_batch_size": 8,
|
976 |
"trial_name": null,
|
977 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 2.5530243519245874,
|
5 |
"eval_steps": 50,
|
6 |
+
"global_step": 3250,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
952 |
"eval_samples_per_second": 3.398,
|
953 |
"eval_steps_per_second": 0.442,
|
954 |
"step": 3150
|
955 |
+
},
|
956 |
+
{
|
957 |
+
"epoch": 2.5137470542026707,
|
958 |
+
"grad_norm": 0.25468263030052185,
|
959 |
+
"learning_rate": 9.004946173988944e-05,
|
960 |
+
"loss": 1.7506,
|
961 |
+
"step": 3200
|
962 |
+
},
|
963 |
+
{
|
964 |
+
"epoch": 2.5137470542026707,
|
965 |
+
"eval_loss": 1.7496337890625,
|
966 |
+
"eval_runtime": 29.4486,
|
967 |
+
"eval_samples_per_second": 3.396,
|
968 |
+
"eval_steps_per_second": 0.441,
|
969 |
+
"step": 3200
|
970 |
}
|
971 |
],
|
972 |
"logging_steps": 50,
|
|
|
986 |
"attributes": {}
|
987 |
}
|
988 |
},
|
989 |
+
"total_flos": 4.0245258941952e+16,
|
990 |
"train_batch_size": 8,
|
991 |
"trial_name": null,
|
992 |
"trial_params": null
|