Training in progress, step 10000
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 sha256:4c9804f22b1119122a1b7adbb6eb405f9cce39acd07f5aa93f2f49cbfa883581
|
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:a5d0f588fd8d9b807d3f9ad64c298a915b00536cff16a1d70c846327ab9d47ad
|
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:c4596ea7cd7ae6c0252c56421bf56b3436d93483f3453718b8e382fbf4e1f49a
|
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 sha256:0e211d61d739d4c465f70557e5e37ce2c3fde83a568f0b93a26565113af8a331
|
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 sha256:12a54672de52f3a5581531956e4c2ed7b5897dcb1eb14a6453175284d2995667
|
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,
|
@@ -78,11 +78,29 @@
|
|
78 |
"eval_samples_per_second": 145.911,
|
79 |
"eval_steps_per_second": 1.524,
|
80 |
"step": 8000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
81 |
}
|
82 |
],
|
83 |
"max_steps": 36045,
|
84 |
"num_train_epochs": 5,
|
85 |
-
"total_flos": 1.
|
86 |
"trial_name": null,
|
87 |
"trial_params": null
|
88 |
}
|
|
|
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": 1.3871549452073797,
|
5 |
+
"global_step": 10000,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
78 |
"eval_samples_per_second": 145.911,
|
79 |
"eval_steps_per_second": 1.524,
|
80 |
"step": 8000
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"epoch": 1.39,
|
84 |
+
"learning_rate": 7.607908416564453e-05,
|
85 |
+
"loss": 1.0477,
|
86 |
+
"step": 10000
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 1.39,
|
90 |
+
"eval_exact_match": 0.33387854386847843,
|
91 |
+
"eval_loss": 1.5617761611938477,
|
92 |
+
"eval_rouge1": 0.5280223151246797,
|
93 |
+
"eval_rouge2": 0.2615585801921996,
|
94 |
+
"eval_rougeL": 0.5274734242132477,
|
95 |
+
"eval_runtime": 165.805,
|
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": 1.2991594865098752e+17,
|
104 |
"trial_name": null,
|
105 |
"trial_params": null
|
106 |
}
|
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:a5d0f588fd8d9b807d3f9ad64c298a915b00536cff16a1d70c846327ab9d47ad
|
3 |
size 242019067
|