Jo Kristian Bergum
commited on
Commit
·
aaf180c
1
Parent(s):
b28cf94
Add README
Browse files
README.md
ADDED
@@ -0,0 +1,48 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
tags:
|
4 |
+
- generated_from_trainer
|
5 |
+
datasets:
|
6 |
+
- go_emotions
|
7 |
+
|
8 |
+
model-index:
|
9 |
+
- name: xtremedistil-l6-h384-go-emotion
|
10 |
+
results:
|
11 |
+
- task:
|
12 |
+
name: Text Classification
|
13 |
+
type: multi_label_classification
|
14 |
+
dataset:
|
15 |
+
name: go_emotions
|
16 |
+
type: emotion
|
17 |
+
args: default
|
18 |
+
|
19 |
+
---
|
20 |
+
# xtremedistil-l6-h384-go-emotion
|
21 |
+
This model is a fine-tuned version of [microsoft/xtremedistil-l6-h384-uncased](https://huggingface.co/microsoft/xtremedistil-l6-h384-uncased) on the go_emotions dataset.
|
22 |
+
|
23 |
+
|
24 |
+
### Training hyperparameters
|
25 |
+
- learning_rate=3e-05
|
26 |
+
<pre>
|
27 |
+
Num examples = 211225
|
28 |
+
Num Epochs = 4
|
29 |
+
Instantaneous batch size per device = 128
|
30 |
+
Total train batch size (w. parallel, distributed & accumulation) = 128
|
31 |
+
Gradient Accumulation steps = 1
|
32 |
+
Total optimization steps = 6604
|
33 |
+
[6604/6604 53:23, Epoch 4/4]
|
34 |
+
Step Training Loss
|
35 |
+
500 0.263200
|
36 |
+
1000 0.156900
|
37 |
+
1500 0.152500
|
38 |
+
2000 0.145400
|
39 |
+
2500 0.140500
|
40 |
+
3000 0.135900
|
41 |
+
3500 0.132800
|
42 |
+
4000 0.129400
|
43 |
+
4500 0.127200
|
44 |
+
5000 0.125700
|
45 |
+
5500 0.124400
|
46 |
+
6000 0.124100
|
47 |
+
6500 0.123400
|
48 |
+
</pre>
|