Training in progress, step 550
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 sha256:84ca4328b493b135f4faafc09a027ff06e038fa4977c582a54a1a8400a2cb4ef
|
3 |
size 35668592
|
metrics.json
CHANGED
@@ -8,3 +8,4 @@
|
|
8 |
{"Step":400,"eval_loss":0.7637045383,"eval_runtime":56.5561,"eval_samples_per_second":1.768,"eval_steps_per_second":0.23,"epoch":0.7736943907}
|
9 |
{"Step":450,"eval_loss":0.7616338134,"eval_runtime":56.6718,"eval_samples_per_second":1.765,"eval_steps_per_second":0.229,"epoch":0.8704061896}
|
10 |
{"Step":500,"eval_loss":0.7601596117,"eval_runtime":56.9725,"eval_samples_per_second":1.755,"eval_steps_per_second":0.228,"epoch":0.9671179884}
|
|
|
|
8 |
{"Step":400,"eval_loss":0.7637045383,"eval_runtime":56.5561,"eval_samples_per_second":1.768,"eval_steps_per_second":0.23,"epoch":0.7736943907}
|
9 |
{"Step":450,"eval_loss":0.7616338134,"eval_runtime":56.6718,"eval_samples_per_second":1.765,"eval_steps_per_second":0.229,"epoch":0.8704061896}
|
10 |
{"Step":500,"eval_loss":0.7601596117,"eval_runtime":56.9725,"eval_samples_per_second":1.755,"eval_steps_per_second":0.228,"epoch":0.9671179884}
|
11 |
+
{"Step":550,"eval_loss":0.7588610053,"eval_runtime":60.6429,"eval_samples_per_second":1.649,"eval_steps_per_second":0.214,"epoch":1.0638297872}
|
state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch":
|
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,
|
@@ -142,6 +142,21 @@
|
|
142 |
"eval_samples_per_second": 1.765,
|
143 |
"eval_steps_per_second": 0.229,
|
144 |
"step": 450
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
145 |
}
|
146 |
],
|
147 |
"logging_steps": 50,
|
@@ -161,7 +176,7 @@
|
|
161 |
"attributes": {}
|
162 |
}
|
163 |
},
|
164 |
-
"total_flos":
|
165 |
"train_batch_size": 8,
|
166 |
"trial_name": null,
|
167 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0638297872340425,
|
5 |
"eval_steps": 50,
|
6 |
+
"global_step": 550,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
142 |
"eval_samples_per_second": 1.765,
|
143 |
"eval_steps_per_second": 0.229,
|
144 |
"step": 450
|
145 |
+
},
|
146 |
+
{
|
147 |
+
"epoch": 0.9671179883945842,
|
148 |
+
"grad_norm": 0.03528429567813873,
|
149 |
+
"learning_rate": 0.00037670250896057353,
|
150 |
+
"loss": 0.7507,
|
151 |
+
"step": 500
|
152 |
+
},
|
153 |
+
{
|
154 |
+
"epoch": 0.9671179883945842,
|
155 |
+
"eval_loss": 0.7601596117019653,
|
156 |
+
"eval_runtime": 56.9725,
|
157 |
+
"eval_samples_per_second": 1.755,
|
158 |
+
"eval_steps_per_second": 0.228,
|
159 |
+
"step": 500
|
160 |
}
|
161 |
],
|
162 |
"logging_steps": 50,
|
|
|
176 |
"attributes": {}
|
177 |
}
|
178 |
},
|
179 |
+
"total_flos": 6.507715798573056e+16,
|
180 |
"train_batch_size": 8,
|
181 |
"trial_name": null,
|
182 |
"trial_params": null
|