Commit
•
c09fca4
1
Parent(s):
6700421
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,87 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
training_args = TrainingArguments(
|
| 2 |
+
output_dir='pegasus-base-wikilarge-newsela-with-domain-adaptation',
|
| 3 |
+
num_train_epochs=20,
|
| 4 |
+
warmup_steps=250,
|
| 5 |
+
per_device_train_batch_size=BATCH_SIZE,
|
| 6 |
+
weight_decay=0.01,
|
| 7 |
+
learning_rate=2e-4,
|
| 8 |
+
# fp16=True,
|
| 9 |
+
optim="adafactor",
|
| 10 |
+
)
|
| 11 |
+
|
| 12 |
+
Step Training Loss
|
| 13 |
+
500 4.391800
|
| 14 |
+
1000 3.994400
|
| 15 |
+
1500 3.009300
|
| 16 |
+
2000 2.596300
|
| 17 |
+
2500 2.389600
|
| 18 |
+
3000 2.328100
|
| 19 |
+
3500 2.272900
|
| 20 |
+
4000 2.125300
|
| 21 |
+
4500 2.019500
|
| 22 |
+
5000 2.005500
|
| 23 |
+
5500 1.994400
|
| 24 |
+
6000 1.853300
|
| 25 |
+
6500 1.838700
|
| 26 |
+
7000 1.840200
|
| 27 |
+
7500 1.822700
|
| 28 |
+
8000 1.716600
|
| 29 |
+
8500 1.734100
|
| 30 |
+
9000 1.739500
|
| 31 |
+
9500 1.696000
|
| 32 |
+
10000 1.661700
|
| 33 |
+
10500 1.672500
|
| 34 |
+
11000 1.666800
|
| 35 |
+
11500 1.617700
|
| 36 |
+
12000 1.611400
|
| 37 |
+
12500 1.616300
|
| 38 |
+
13000 1.625800
|
| 39 |
+
13500 1.567700
|
| 40 |
+
14000 1.584600
|
| 41 |
+
14500 1.589800
|
| 42 |
+
15000 1.574600
|
| 43 |
+
15500 1.548300
|
| 44 |
+
16000 1.559800
|
| 45 |
+
16500 1.562100
|
| 46 |
+
17000 1.541600
|
| 47 |
+
17500 1.533500
|
| 48 |
+
18000 1.538400
|
| 49 |
+
18500 1.545700
|
| 50 |
+
19000 1.510100
|
| 51 |
+
19500 1.522600
|
| 52 |
+
20000 1.529100
|
| 53 |
+
20500 1.520900
|
| 54 |
+
21000 1.501100
|
| 55 |
+
21500 1.508400
|
| 56 |
+
22000 1.515400
|
| 57 |
+
22500 1.500100
|
| 58 |
+
23000 1.496700
|
| 59 |
+
23500 1.495900
|
| 60 |
+
24000 1.505200
|
| 61 |
+
24500 1.484400
|
| 62 |
+
25000 1.483000
|
| 63 |
+
25500 1.491200
|
| 64 |
+
26000 1.491900
|
| 65 |
+
26500 1.468600
|
| 66 |
+
27000 1.479800
|
| 67 |
+
27500 1.479600
|
| 68 |
+
28000 1.474400
|
| 69 |
+
28500 1.466000
|
| 70 |
+
29000 1.471700
|
| 71 |
+
29500 1.473100
|
| 72 |
+
30000 1.463900
|
| 73 |
+
30500 1.459300
|
| 74 |
+
31000 1.463200
|
| 75 |
+
31500 1.466800
|
| 76 |
+
32000 1.455300
|
| 77 |
+
32500 1.458600
|
| 78 |
+
33000 1.455200
|
| 79 |
+
33500 1.455500
|
| 80 |
+
34000 1.449100
|
| 81 |
+
34500 1.451400
|
| 82 |
+
35000 1.451200
|
| 83 |
+
35500 1.449000
|
| 84 |
+
36000 1.443700
|
| 85 |
+
36500 1.447600
|
| 86 |
+
37000 1.447300
|
| 87 |
+
TrainOutput(global_step=37100, training_loss=1.7110925547967382, metrics={'train_runtime': 33376.534, 'train_samples_per_second': 8.889, 'train_steps_per_second': 1.112, 'total_flos': 0.0, 'train_loss': 1.7110925547967382, 'epoch': 20.0})
|