Training in progress, step 450
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 sha256:05ebb8e061ac912f602df410a9d77dcf61863b3c8d47ebd333b092a5901e23b7
|
3 |
size 35668592
|
metrics.json
CHANGED
@@ -6,3 +6,4 @@
|
|
6 |
{"Step":300,"eval_loss":0.7687731385,"eval_runtime":60.6867,"eval_samples_per_second":1.648,"eval_steps_per_second":0.214,"epoch":0.580270793}
|
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}
|
|
|
|
6 |
{"Step":300,"eval_loss":0.7687731385,"eval_runtime":60.6867,"eval_samples_per_second":1.648,"eval_steps_per_second":0.214,"epoch":0.580270793}
|
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}
|
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,
|
@@ -112,6 +112,21 @@
|
|
112 |
"eval_samples_per_second": 1.767,
|
113 |
"eval_steps_per_second": 0.23,
|
114 |
"step": 350
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
115 |
}
|
116 |
],
|
117 |
"logging_steps": 50,
|
@@ -131,7 +146,7 @@
|
|
131 |
"attributes": {}
|
132 |
}
|
133 |
},
|
134 |
-
"total_flos":
|
135 |
"train_batch_size": 8,
|
136 |
"trial_name": null,
|
137 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.8704061895551257,
|
5 |
"eval_steps": 50,
|
6 |
+
"global_step": 450,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
112 |
"eval_samples_per_second": 1.767,
|
113 |
"eval_steps_per_second": 0.23,
|
114 |
"step": 350
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.7736943907156673,
|
118 |
+
"grad_norm": 0.03772694244980812,
|
119 |
+
"learning_rate": 0.0004125448028673835,
|
120 |
+
"loss": 0.7529,
|
121 |
+
"step": 400
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.7736943907156673,
|
125 |
+
"eval_loss": 0.7637045383453369,
|
126 |
+
"eval_runtime": 56.5561,
|
127 |
+
"eval_samples_per_second": 1.768,
|
128 |
+
"eval_steps_per_second": 0.23,
|
129 |
+
"step": 400
|
130 |
}
|
131 |
],
|
132 |
"logging_steps": 50,
|
|
|
146 |
"attributes": {}
|
147 |
}
|
148 |
},
|
149 |
+
"total_flos": 5.189203605086208e+16,
|
150 |
"train_batch_size": 8,
|
151 |
"trial_name": null,
|
152 |
"trial_params": null
|