Eldar Kurtic
commited on
Commit
·
cb21039
1
Parent(s):
0587bac
add preliminary readme
Browse files
README.md
ADDED
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# More evals coming soon
|
2 |
+
|
3 |
+
- unquantized baseline on GSM8k
|
4 |
+
```bash
|
5 |
+
|Tasks|Version| Filter |n-shot| Metric | |Value | |Stderr|
|
6 |
+
|-----|------:|----------------|-----:|-----------|---|-----:|---|-----:|
|
7 |
+
|gsm8k| 3|flexible-extract| 5|exact_match|↑ |0.9591|± |0.0055|
|
8 |
+
| | |strict-match | 5|exact_match|↑ |0.9568|± |0.0056|
|
9 |
+
```
|
10 |
+
|
11 |
+
- this INT4 quantized model on GSM8k
|
12 |
+
```bash
|
13 |
+
|Tasks|Version| Filter |n-shot| Metric | |Value | |Stderr|
|
14 |
+
|-----|------:|----------------|-----:|-----------|---|-----:|---|-----:|
|
15 |
+
|gsm8k| 3|flexible-extract| 5|exact_match|↑ |0.9560|± |0.0056|
|
16 |
+
| | |strict-match | 5|exact_match|↑ |0.9553|± |0.0057|
|
17 |
+
```
|