Training in progress, step 1200
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 sha256:001db2cc0f28eef9e586f657554b5c65108455c804890b96f6eac5c4428c92cd
|
3 |
size 35668592
|
metrics.json
CHANGED
@@ -21,3 +21,4 @@
|
|
21 |
{"Step":1050,"eval_loss":0.7514955997,"eval_runtime":60.5727,"eval_samples_per_second":1.651,"eval_steps_per_second":0.215,"epoch":2.0309477756}
|
22 |
{"Step":1100,"eval_loss":0.7509506345,"eval_runtime":56.8661,"eval_samples_per_second":1.759,"eval_steps_per_second":0.229,"epoch":2.1276595745}
|
23 |
{"Step":1150,"eval_loss":0.7505396008,"eval_runtime":56.5726,"eval_samples_per_second":1.768,"eval_steps_per_second":0.23,"epoch":2.2243713733}
|
|
|
|
21 |
{"Step":1050,"eval_loss":0.7514955997,"eval_runtime":60.5727,"eval_samples_per_second":1.651,"eval_steps_per_second":0.215,"epoch":2.0309477756}
|
22 |
{"Step":1100,"eval_loss":0.7509506345,"eval_runtime":56.8661,"eval_samples_per_second":1.759,"eval_steps_per_second":0.229,"epoch":2.1276595745}
|
23 |
{"Step":1150,"eval_loss":0.7505396008,"eval_runtime":56.5726,"eval_samples_per_second":1.768,"eval_steps_per_second":0.23,"epoch":2.2243713733}
|
24 |
+
{"Step":1200,"eval_loss":0.7510370612,"eval_runtime":56.6798,"eval_samples_per_second":1.764,"eval_steps_per_second":0.229,"epoch":2.3210831721}
|
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,
|
@@ -337,6 +337,21 @@
|
|
337 |
"eval_samples_per_second": 1.759,
|
338 |
"eval_steps_per_second": 0.229,
|
339 |
"step": 1100
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
340 |
}
|
341 |
],
|
342 |
"logging_steps": 50,
|
@@ -356,7 +371,7 @@
|
|
356 |
"attributes": {}
|
357 |
}
|
358 |
},
|
359 |
-
"total_flos": 1.
|
360 |
"train_batch_size": 8,
|
361 |
"trial_name": null,
|
362 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 2.3210831721470018,
|
5 |
"eval_steps": 50,
|
6 |
+
"global_step": 1200,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
337 |
"eval_samples_per_second": 1.759,
|
338 |
"eval_steps_per_second": 0.229,
|
339 |
"step": 1100
|
340 |
+
},
|
341 |
+
{
|
342 |
+
"epoch": 2.2243713733075436,
|
343 |
+
"grad_norm": 0.04050629585981369,
|
344 |
+
"learning_rate": 0.00014372759856630824,
|
345 |
+
"loss": 0.704,
|
346 |
+
"step": 1150
|
347 |
+
},
|
348 |
+
{
|
349 |
+
"epoch": 2.2243713733075436,
|
350 |
+
"eval_loss": 0.7505396008491516,
|
351 |
+
"eval_runtime": 56.5726,
|
352 |
+
"eval_samples_per_second": 1.768,
|
353 |
+
"eval_steps_per_second": 0.23,
|
354 |
+
"step": 1150
|
355 |
}
|
356 |
],
|
357 |
"logging_steps": 50,
|
|
|
371 |
"attributes": {}
|
372 |
}
|
373 |
},
|
374 |
+
"total_flos": 1.4913266854373376e+17,
|
375 |
"train_batch_size": 8,
|
376 |
"trial_name": null,
|
377 |
"trial_params": null
|