Training in progress, step 900
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 sha256:9126573abdea57442c91871f495c557fbdd0d91b4a04859b8278fd29a71b8773
|
3 |
size 35668592
|
metrics.json
CHANGED
@@ -15,3 +15,4 @@
|
|
15 |
{"Step":750,"eval_loss":0.7549133897,"eval_runtime":56.529,"eval_samples_per_second":1.769,"eval_steps_per_second":0.23,"epoch":1.4506769826}
|
16 |
{"Step":800,"eval_loss":0.7541146278,"eval_runtime":60.5765,"eval_samples_per_second":1.651,"eval_steps_per_second":0.215,"epoch":1.5473887814}
|
17 |
{"Step":850,"eval_loss":0.7533394098,"eval_runtime":56.7119,"eval_samples_per_second":1.763,"eval_steps_per_second":0.229,"epoch":1.6441005803}
|
|
|
|
15 |
{"Step":750,"eval_loss":0.7549133897,"eval_runtime":56.529,"eval_samples_per_second":1.769,"eval_steps_per_second":0.23,"epoch":1.4506769826}
|
16 |
{"Step":800,"eval_loss":0.7541146278,"eval_runtime":60.5765,"eval_samples_per_second":1.651,"eval_steps_per_second":0.215,"epoch":1.5473887814}
|
17 |
{"Step":850,"eval_loss":0.7533394098,"eval_runtime":56.7119,"eval_samples_per_second":1.763,"eval_steps_per_second":0.229,"epoch":1.6441005803}
|
18 |
+
{"Step":900,"eval_loss":0.7530195117,"eval_runtime":56.5303,"eval_samples_per_second":1.769,"eval_steps_per_second":0.23,"epoch":1.7408123791}
|
state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 1.
|
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,
|
@@ -247,6 +247,21 @@
|
|
247 |
"eval_samples_per_second": 1.651,
|
248 |
"eval_steps_per_second": 0.215,
|
249 |
"step": 800
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
250 |
}
|
251 |
],
|
252 |
"logging_steps": 50,
|
@@ -266,7 +281,7 @@
|
|
266 |
"attributes": {}
|
267 |
}
|
268 |
},
|
269 |
-
"total_flos": 1.
|
270 |
"train_batch_size": 8,
|
271 |
"trial_name": null,
|
272 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.7408123791102514,
|
5 |
"eval_steps": 50,
|
6 |
+
"global_step": 900,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
247 |
"eval_samples_per_second": 1.651,
|
248 |
"eval_steps_per_second": 0.215,
|
249 |
"step": 800
|
250 |
+
},
|
251 |
+
{
|
252 |
+
"epoch": 1.644100580270793,
|
253 |
+
"grad_norm": 0.041316960006952286,
|
254 |
+
"learning_rate": 0.00025125448028673835,
|
255 |
+
"loss": 0.742,
|
256 |
+
"step": 850
|
257 |
+
},
|
258 |
+
{
|
259 |
+
"epoch": 1.644100580270793,
|
260 |
+
"eval_loss": 0.753339409828186,
|
261 |
+
"eval_runtime": 56.7119,
|
262 |
+
"eval_samples_per_second": 1.763,
|
263 |
+
"eval_steps_per_second": 0.229,
|
264 |
+
"step": 850
|
265 |
}
|
266 |
],
|
267 |
"logging_steps": 50,
|
|
|
281 |
"attributes": {}
|
282 |
}
|
283 |
},
|
284 |
+
"total_flos": 1.1029014455129088e+17,
|
285 |
"train_batch_size": 8,
|
286 |
"trial_name": null,
|
287 |
"trial_params": null
|