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 484079855
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f5e6db76fdaa766268444d7ae4a5693ff5e2b7a2a4c5f4828e1590270e96890a
|
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 sha256:09b3962d58d522ed339e445ad617274ad7e5e44517257d3f8447d9f14398911c
|
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: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 sha256:9c9bd6f2e927e1ebd9f70e3fceda3f41cb1d6cb17642e35f823f778e7146c9ea
|
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:d290c63695fb56faa396b551583792dea1b2900777e1b6881004c5046ccec6bb
|
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,
|
@@ -96,11 +96,29 @@
|
|
96 |
"eval_samples_per_second": 143.183,
|
97 |
"eval_steps_per_second": 1.495,
|
98 |
"step": 10000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
99 |
}
|
100 |
],
|
101 |
"max_steps": 72090,
|
102 |
"num_train_epochs": 10,
|
103 |
-
"total_flos": 1.
|
104 |
"trial_name": null,
|
105 |
"trial_params": null
|
106 |
}
|
|
|
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": 1.6645859342488556,
|
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": 143.183,
|
97 |
"eval_steps_per_second": 1.495,
|
98 |
"step": 10000
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 1.66,
|
102 |
+
"learning_rate": 8.775060232167628e-05,
|
103 |
+
"loss": 1.0402,
|
104 |
+
"step": 12000
|
105 |
+
},
|
106 |
+
{
|
107 |
+
"epoch": 1.66,
|
108 |
+
"eval_exact_match": 0.310476430129173,
|
109 |
+
"eval_loss": 1.6315388679504395,
|
110 |
+
"eval_rouge1": 0.5041999610588694,
|
111 |
+
"eval_rouge2": 0.23111610343932948,
|
112 |
+
"eval_rougeL": 0.5037586055646353,
|
113 |
+
"eval_runtime": 166.2206,
|
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": 1.5590157453361152e+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 242029307
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:09b3962d58d522ed339e445ad617274ad7e5e44517257d3f8447d9f14398911c
|
3 |
size 242029307
|