Training in progress, step 1050
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 sha256:8c32479c891bb08c4b8149659ed507dddcf937a0c55d4cdaca7f55b533a7fb2f
|
3 |
size 35668592
|
metrics.json
CHANGED
@@ -18,3 +18,4 @@
|
|
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}
|
19 |
{"Step":950,"eval_loss":0.7525290847,"eval_runtime":56.698,"eval_samples_per_second":1.764,"eval_steps_per_second":0.229,"epoch":1.8375241779}
|
20 |
{"Step":1000,"eval_loss":0.751624763,"eval_runtime":56.5619,"eval_samples_per_second":1.768,"eval_steps_per_second":0.23,"epoch":1.9342359768}
|
|
|
|
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}
|
19 |
{"Step":950,"eval_loss":0.7525290847,"eval_runtime":56.698,"eval_samples_per_second":1.764,"eval_steps_per_second":0.229,"epoch":1.8375241779}
|
20 |
{"Step":1000,"eval_loss":0.751624763,"eval_runtime":56.5619,"eval_samples_per_second":1.768,"eval_steps_per_second":0.23,"epoch":1.9342359768}
|
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}
|
state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch":
|
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,
|
@@ -292,6 +292,21 @@
|
|
292 |
"eval_samples_per_second": 1.764,
|
293 |
"eval_steps_per_second": 0.229,
|
294 |
"step": 950
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
295 |
}
|
296 |
],
|
297 |
"logging_steps": 50,
|
@@ -311,7 +326,7 @@
|
|
311 |
"attributes": {}
|
312 |
}
|
313 |
},
|
314 |
-
"total_flos": 1.
|
315 |
"train_batch_size": 8,
|
316 |
"trial_name": null,
|
317 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 2.0309477756286265,
|
5 |
"eval_steps": 50,
|
6 |
+
"global_step": 1050,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
292 |
"eval_samples_per_second": 1.764,
|
293 |
"eval_steps_per_second": 0.229,
|
294 |
"step": 950
|
295 |
+
},
|
296 |
+
{
|
297 |
+
"epoch": 1.9342359767891684,
|
298 |
+
"grad_norm": 0.046565938740968704,
|
299 |
+
"learning_rate": 0.0001974910394265233,
|
300 |
+
"loss": 0.7323,
|
301 |
+
"step": 1000
|
302 |
+
},
|
303 |
+
{
|
304 |
+
"epoch": 1.9342359767891684,
|
305 |
+
"eval_loss": 0.7516247630119324,
|
306 |
+
"eval_runtime": 56.5619,
|
307 |
+
"eval_samples_per_second": 1.768,
|
308 |
+
"eval_steps_per_second": 0.23,
|
309 |
+
"step": 1000
|
310 |
}
|
311 |
],
|
312 |
"logging_steps": 50,
|
|
|
326 |
"attributes": {}
|
327 |
}
|
328 |
},
|
329 |
+
"total_flos": 1.2969133485290496e+17,
|
330 |
"train_batch_size": 8,
|
331 |
"trial_name": null,
|
332 |
"trial_params": null
|