Training in progress, step 1300
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 sha256:6335cf58c6cc23ddd2e1d6c638529bff79aa05b9a843045ec46dc902615a1676
|
3 |
size 35668592
|
metrics.json
CHANGED
@@ -23,3 +23,4 @@
|
|
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}
|
25 |
{"Step":1250,"eval_loss":0.7501779795,"eval_runtime":56.6502,"eval_samples_per_second":1.765,"eval_steps_per_second":0.229,"epoch":2.417794971}
|
|
|
|
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}
|
25 |
{"Step":1250,"eval_loss":0.7501779795,"eval_runtime":56.6502,"eval_samples_per_second":1.765,"eval_steps_per_second":0.229,"epoch":2.417794971}
|
26 |
+
{"Step":1300,"eval_loss":0.7498863339,"eval_runtime":60.6054,"eval_samples_per_second":1.65,"eval_steps_per_second":0.215,"epoch":2.5145067698}
|
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,
|
@@ -367,6 +367,21 @@
|
|
367 |
"eval_samples_per_second": 1.764,
|
368 |
"eval_steps_per_second": 0.229,
|
369 |
"step": 1200
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
370 |
}
|
371 |
],
|
372 |
"logging_steps": 50,
|
@@ -386,7 +401,7 @@
|
|
386 |
"attributes": {}
|
387 |
}
|
388 |
},
|
389 |
-
"total_flos": 1.
|
390 |
"train_batch_size": 8,
|
391 |
"trial_name": null,
|
392 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 2.514506769825919,
|
5 |
"eval_steps": 50,
|
6 |
+
"global_step": 1300,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
367 |
"eval_samples_per_second": 1.764,
|
368 |
"eval_steps_per_second": 0.229,
|
369 |
"step": 1200
|
370 |
+
},
|
371 |
+
{
|
372 |
+
"epoch": 2.4177949709864603,
|
373 |
+
"grad_norm": 0.04337254539132118,
|
374 |
+
"learning_rate": 0.0001078853046594982,
|
375 |
+
"loss": 0.7203,
|
376 |
+
"step": 1250
|
377 |
+
},
|
378 |
+
{
|
379 |
+
"epoch": 2.4177949709864603,
|
380 |
+
"eval_loss": 0.7501779794692993,
|
381 |
+
"eval_runtime": 56.6502,
|
382 |
+
"eval_samples_per_second": 1.765,
|
383 |
+
"eval_steps_per_second": 0.229,
|
384 |
+
"step": 1250
|
385 |
}
|
386 |
],
|
387 |
"logging_steps": 50,
|
|
|
401 |
"attributes": {}
|
402 |
}
|
403 |
},
|
404 |
+
"total_flos": 1.6217629230802944e+17,
|
405 |
"train_batch_size": 8,
|
406 |
"trial_name": null,
|
407 |
"trial_params": null
|