Training in progress, step 1000
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 sha256:0c8c6bf298e1b587725040f02e0d8ae0be361317927b5eba12113fb4c875894a
|
3 |
size 35668592
|
metrics.json
CHANGED
@@ -17,3 +17,4 @@
|
|
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}
|
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}
|
|
|
|
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}
|
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}
|
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,
|
@@ -277,6 +277,21 @@
|
|
277 |
"eval_samples_per_second": 1.769,
|
278 |
"eval_steps_per_second": 0.23,
|
279 |
"step": 900
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
280 |
}
|
281 |
],
|
282 |
"logging_steps": 50,
|
@@ -296,7 +311,7 @@
|
|
296 |
"attributes": {}
|
297 |
}
|
298 |
},
|
299 |
-
"total_flos": 1.
|
300 |
"train_batch_size": 8,
|
301 |
"trial_name": null,
|
302 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.9342359767891684,
|
5 |
"eval_steps": 50,
|
6 |
+
"global_step": 1000,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
277 |
"eval_samples_per_second": 1.769,
|
278 |
"eval_steps_per_second": 0.23,
|
279 |
"step": 900
|
280 |
+
},
|
281 |
+
{
|
282 |
+
"epoch": 1.8375241779497098,
|
283 |
+
"grad_norm": 0.03964291140437126,
|
284 |
+
"learning_rate": 0.0002154121863799283,
|
285 |
+
"loss": 0.7352,
|
286 |
+
"step": 950
|
287 |
+
},
|
288 |
+
{
|
289 |
+
"epoch": 1.8375241779497098,
|
290 |
+
"eval_loss": 0.7525290846824646,
|
291 |
+
"eval_runtime": 56.698,
|
292 |
+
"eval_samples_per_second": 1.764,
|
293 |
+
"eval_steps_per_second": 0.229,
|
294 |
+
"step": 950
|
295 |
}
|
296 |
],
|
297 |
"logging_steps": 50,
|
|
|
311 |
"attributes": {}
|
312 |
}
|
313 |
},
|
314 |
+
"total_flos": 1.2322182356038656e+17,
|
315 |
"train_batch_size": 8,
|
316 |
"trial_name": null,
|
317 |
"trial_params": null
|