Training in progress, step 12000
Browse files- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/pytorch_model.bin +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scaler.pt +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +21 -3
- pytorch_model.bin +1 -1
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 484059375
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:54eb609d9e7c9af8da5e499842252b44866540b8464e813675b437dafad687b2
|
3 |
size 484059375
|
last-checkpoint/pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 242019067
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:30184e0248c24f3d379731864a3d98181dd0e082ff4728e8eb1473d646421010
|
3 |
size 242019067
|
last-checkpoint/rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14503
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a75063c2eccccee16964972d0c16700ba598e8ad046be88de665bdc4e411e739
|
3 |
size 14503
|
last-checkpoint/scaler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 559
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0285090586e6ad2f57d9d2e587707107a6deb19c653c72063c13e57720ea6b1e
|
3 |
size 559
|
last-checkpoint/scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 623
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:26b34eea6aaf93230bd2195d19b94a8c538e2fb1720b246af9cbe4988634461c
|
3 |
size 623
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"best_metric": 1.528552770614624,
|
3 |
"best_model_checkpoint": "/home2/s5431786/jb-internship-repr/results/Salesforce-codet5-small-java-small-selected-wo-tokens/checkpoint-2000",
|
4 |
-
"epoch": 1.
|
5 |
-
"global_step":
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
@@ -96,11 +96,29 @@
|
|
96 |
"eval_samples_per_second": 143.807,
|
97 |
"eval_steps_per_second": 1.502,
|
98 |
"step": 10000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
99 |
}
|
100 |
],
|
101 |
"max_steps": 36045,
|
102 |
"num_train_epochs": 5,
|
103 |
-
"total_flos": 1.
|
104 |
"trial_name": null,
|
105 |
"trial_params": null
|
106 |
}
|
|
|
1 |
{
|
2 |
"best_metric": 1.528552770614624,
|
3 |
"best_model_checkpoint": "/home2/s5431786/jb-internship-repr/results/Salesforce-codet5-small-java-small-selected-wo-tokens/checkpoint-2000",
|
4 |
+
"epoch": 1.6645859342488556,
|
5 |
+
"global_step": 12000,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
96 |
"eval_samples_per_second": 143.807,
|
97 |
"eval_steps_per_second": 1.502,
|
98 |
"step": 10000
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 1.66,
|
102 |
+
"learning_rate": 7.023830383739268e-05,
|
103 |
+
"loss": 1.0219,
|
104 |
+
"step": 12000
|
105 |
+
},
|
106 |
+
{
|
107 |
+
"epoch": 1.66,
|
108 |
+
"eval_exact_match": 0.32372924006039255,
|
109 |
+
"eval_loss": 1.5944069623947144,
|
110 |
+
"eval_rouge1": 0.5133412941637794,
|
111 |
+
"eval_rouge2": 0.24304376784560536,
|
112 |
+
"eval_rougeL": 0.5128045498112566,
|
113 |
+
"eval_runtime": 164.4858,
|
114 |
+
"eval_samples_per_second": 144.961,
|
115 |
+
"eval_steps_per_second": 1.514,
|
116 |
+
"step": 12000
|
117 |
}
|
118 |
],
|
119 |
"max_steps": 36045,
|
120 |
"num_train_epochs": 5,
|
121 |
+
"total_flos": 1.5590157453361152e+17,
|
122 |
"trial_name": null,
|
123 |
"trial_params": null
|
124 |
}
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 242019067
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:30184e0248c24f3d379731864a3d98181dd0e082ff4728e8eb1473d646421010
|
3 |
size 242019067
|