Training in progress, step 500
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 sha256:3e620996cd153f720211cb2e4aea403a96a43a72bdcfeae457a3f030101aa708
|
3 |
size 35668592
|
metrics.json
CHANGED
@@ -7,3 +7,4 @@
|
|
7 |
{"Step":350,"eval_loss":0.7662523389,"eval_runtime":56.5991,"eval_samples_per_second":1.767,"eval_steps_per_second":0.23,"epoch":0.6769825919}
|
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}
|
|
|
|
7 |
{"Step":350,"eval_loss":0.7662523389,"eval_runtime":56.5991,"eval_samples_per_second":1.767,"eval_steps_per_second":0.23,"epoch":0.6769825919}
|
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}
|
state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
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,
|
@@ -127,6 +127,21 @@
|
|
127 |
"eval_samples_per_second": 1.768,
|
128 |
"eval_steps_per_second": 0.23,
|
129 |
"step": 400
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
130 |
}
|
131 |
],
|
132 |
"logging_steps": 50,
|
@@ -146,7 +161,7 @@
|
|
146 |
"attributes": {}
|
147 |
}
|
148 |
},
|
149 |
-
"total_flos": 5.
|
150 |
"train_batch_size": 8,
|
151 |
"trial_name": null,
|
152 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.9671179883945842,
|
5 |
"eval_steps": 50,
|
6 |
+
"global_step": 500,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
127 |
"eval_samples_per_second": 1.768,
|
128 |
"eval_steps_per_second": 0.23,
|
129 |
"step": 400
|
130 |
+
},
|
131 |
+
{
|
132 |
+
"epoch": 0.8704061895551257,
|
133 |
+
"grad_norm": 0.04153395816683769,
|
134 |
+
"learning_rate": 0.00039462365591397855,
|
135 |
+
"loss": 0.7562,
|
136 |
+
"step": 450
|
137 |
+
},
|
138 |
+
{
|
139 |
+
"epoch": 0.8704061895551257,
|
140 |
+
"eval_loss": 0.7616338133811951,
|
141 |
+
"eval_runtime": 56.6718,
|
142 |
+
"eval_samples_per_second": 1.765,
|
143 |
+
"eval_steps_per_second": 0.229,
|
144 |
+
"step": 450
|
145 |
}
|
146 |
],
|
147 |
"logging_steps": 50,
|
|
|
161 |
"attributes": {}
|
162 |
}
|
163 |
},
|
164 |
+
"total_flos": 5.844805825019904e+16,
|
165 |
"train_batch_size": 8,
|
166 |
"trial_name": null,
|
167 |
"trial_params": null
|