Training in progress, step 18000
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:f52a5fd74124741639a322a9f011b9cf644363870d65b2e5294a60a33a3f5dac
|
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:f81297233dead8accd083e85f90059ef83f2b7ba63e684692dca25c395768dc5
|
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 sha256:3f796735cc2134a1c0954f87f4b08604463f6997af0bd584885d37a1822e76d5
|
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:ccd7c52dd13aa6bbac6cef6445e992186ef23a9f10c21352f2c0579bbf1c84d0
|
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 sha256:5eeb244333092aa1a9969c020503bedd9aa2ae9ef47ef6369b6fbc5797a60b03
|
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": 2.
|
5 |
-
"global_step":
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
@@ -150,11 +150,29 @@
|
|
150 |
"eval_samples_per_second": 144.137,
|
151 |
"eval_steps_per_second": 1.505,
|
152 |
"step": 16000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
153 |
}
|
154 |
],
|
155 |
"max_steps": 36045,
|
156 |
"num_train_epochs": 5,
|
157 |
-
"total_flos": 2.
|
158 |
"trial_name": null,
|
159 |
"trial_params": null
|
160 |
}
|
|
|
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": 2.4968789013732833,
|
5 |
+
"global_step": 18000,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
150 |
"eval_samples_per_second": 144.137,
|
151 |
"eval_steps_per_second": 1.505,
|
152 |
"step": 16000
|
153 |
+
},
|
154 |
+
{
|
155 |
+
"epoch": 2.5,
|
156 |
+
"learning_rate": 5.272180363296537e-05,
|
157 |
+
"loss": 0.9054,
|
158 |
+
"step": 18000
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 2.5,
|
162 |
+
"eval_exact_match": 0.3242325113236034,
|
163 |
+
"eval_loss": 1.6233525276184082,
|
164 |
+
"eval_rouge1": 0.5204582779286969,
|
165 |
+
"eval_rouge2": 0.24576987021627228,
|
166 |
+
"eval_rougeL": 0.520011433933161,
|
167 |
+
"eval_runtime": 164.3556,
|
168 |
+
"eval_samples_per_second": 145.076,
|
169 |
+
"eval_steps_per_second": 1.515,
|
170 |
+
"step": 18000
|
171 |
}
|
172 |
],
|
173 |
"max_steps": 36045,
|
174 |
"num_train_epochs": 5,
|
175 |
+
"total_flos": 2.3384627141935104e+17,
|
176 |
"trial_name": null,
|
177 |
"trial_params": null
|
178 |
}
|
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:f81297233dead8accd083e85f90059ef83f2b7ba63e684692dca25c395768dc5
|
3 |
size 242019067
|