Training in progress, step 14000
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 484079855
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:02d46025e65d93fd4d1d4e13ab9b17d217ba434b164809ea39ac823bb555ecbc
|
3 |
size 484079855
|
last-checkpoint/pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 242029307
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:183653e5193a39489e9308cf7d418385810c242c8d1caa58f2d8641b315fa7aa
|
3 |
size 242029307
|
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 sha256:7e8f7efa9426f92d7539e42fac5262290109f255d1efd558300073dd9b1757e3
|
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 sha256:81a4dbf50e1a1ff718ce309bb419ccad19523837c1735208da4d210a88c8bff0
|
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:d2f4f4c45fa84579cb1285490c055e65d39a0cbf8ccd1b6a5dfea4e23bcb7f2e
|
3 |
size 623
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"best_metric": 1.5166430473327637,
|
3 |
"best_model_checkpoint": "/home2/s5431786/jb-internship-repr/results/Salesforce-codet5-small-java-small-selected/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,
|
@@ -114,11 +114,29 @@
|
|
114 |
"eval_samples_per_second": 143.448,
|
115 |
"eval_steps_per_second": 1.498,
|
116 |
"step": 12000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
117 |
}
|
118 |
],
|
119 |
"max_steps": 72090,
|
120 |
"num_train_epochs": 10,
|
121 |
-
"total_flos": 1.
|
122 |
"trial_name": null,
|
123 |
"trial_params": null
|
124 |
}
|
|
|
1 |
{
|
2 |
"best_metric": 1.5166430473327637,
|
3 |
"best_model_checkpoint": "/home2/s5431786/jb-internship-repr/results/Salesforce-codet5-small-java-small-selected/checkpoint-2000",
|
4 |
+
"epoch": 1.9420169232903315,
|
5 |
+
"global_step": 14000,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
114 |
"eval_samples_per_second": 143.448,
|
115 |
"eval_steps_per_second": 1.498,
|
116 |
"step": 12000
|
117 |
+
},
|
118 |
+
{
|
119 |
+
"epoch": 1.94,
|
120 |
+
"learning_rate": 8.483025480032124e-05,
|
121 |
+
"loss": 1.0186,
|
122 |
+
"step": 14000
|
123 |
+
},
|
124 |
+
{
|
125 |
+
"epoch": 1.94,
|
126 |
+
"eval_exact_match": 0.3139154504277806,
|
127 |
+
"eval_loss": 1.6186127662658691,
|
128 |
+
"eval_rouge1": 0.5068008383486242,
|
129 |
+
"eval_rouge2": 0.231322858719689,
|
130 |
+
"eval_rougeL": 0.5061156690587016,
|
131 |
+
"eval_runtime": 165.85,
|
132 |
+
"eval_samples_per_second": 143.768,
|
133 |
+
"eval_steps_per_second": 1.501,
|
134 |
+
"step": 14000
|
135 |
}
|
136 |
],
|
137 |
"max_steps": 72090,
|
138 |
"num_train_epochs": 10,
|
139 |
+
"total_flos": 1.8188720041623552e+17,
|
140 |
"trial_name": null,
|
141 |
"trial_params": null
|
142 |
}
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 242029307
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:183653e5193a39489e9308cf7d418385810c242c8d1caa58f2d8641b315fa7aa
|
3 |
size 242029307
|