Training in progress, step 100
Browse files- adapter_model.safetensors +1 -1
- metrics.json +1 -0
- state.json +21 -5
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:a29c7d2e397ece87357915c4724ce9ff77176af46f68895b09f249b9793a35a7
|
3 |
size 35668592
|
metrics.json
CHANGED
@@ -1 +1,2 @@
|
|
1 |
{"Step":50,"eval_loss":0.9260319471,"eval_runtime":60.9012,"eval_samples_per_second":1.642,"eval_steps_per_second":0.213,"epoch":0.0967117988}
|
|
|
|
1 |
{"Step":50,"eval_loss":0.9260319471,"eval_runtime":60.9012,"eval_samples_per_second":1.642,"eval_steps_per_second":0.213,"epoch":0.0967117988}
|
2 |
+
{"Step":100,"eval_loss":0.8202204108,"eval_runtime":56.5241,"eval_samples_per_second":1.769,"eval_steps_per_second":0.23,"epoch":0.1934235977}
|
state.json
CHANGED
@@ -1,13 +1,29 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
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,
|
10 |
-
"log_history": [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
"logging_steps": 50,
|
12 |
"max_steps": 1551,
|
13 |
"num_input_tokens_seen": 0,
|
@@ -19,13 +35,13 @@
|
|
19 |
"should_epoch_stop": false,
|
20 |
"should_evaluate": false,
|
21 |
"should_log": false,
|
22 |
-
"should_save":
|
23 |
"should_training_stop": false
|
24 |
},
|
25 |
"attributes": {}
|
26 |
}
|
27 |
},
|
28 |
-
"total_flos": 0,
|
29 |
"train_batch_size": 8,
|
30 |
"trial_name": null,
|
31 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.19342359767891681,
|
5 |
"eval_steps": 50,
|
6 |
+
"global_step": 100,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.09671179883945841,
|
13 |
+
"grad_norm": 0.035103704780340195,
|
14 |
+
"learning_rate": 0.00016025641025641028,
|
15 |
+
"loss": 1.0483,
|
16 |
+
"step": 50
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.09671179883945841,
|
20 |
+
"eval_loss": 0.9260319471359253,
|
21 |
+
"eval_runtime": 60.9012,
|
22 |
+
"eval_samples_per_second": 1.642,
|
23 |
+
"eval_steps_per_second": 0.213,
|
24 |
+
"step": 50
|
25 |
+
}
|
26 |
+
],
|
27 |
"logging_steps": 50,
|
28 |
"max_steps": 1551,
|
29 |
"num_input_tokens_seen": 0,
|
|
|
35 |
"should_epoch_stop": false,
|
36 |
"should_evaluate": false,
|
37 |
"should_log": false,
|
38 |
+
"should_save": true,
|
39 |
"should_training_stop": false
|
40 |
},
|
41 |
"attributes": {}
|
42 |
}
|
43 |
},
|
44 |
+
"total_flos": 6550111516262400.0,
|
45 |
"train_batch_size": 8,
|
46 |
"trial_name": null,
|
47 |
"trial_params": null
|