Update README.md
Browse files
README.md
CHANGED
|
@@ -5,6 +5,35 @@ datasets:
|
|
| 5 |
language:
|
| 6 |
- en
|
| 7 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
Sample Responses:
|
| 9 |
|
| 10 |
Llama-2 7B Trained On PFAF750
|
|
|
|
| 5 |
language:
|
| 6 |
- en
|
| 7 |
---
|
| 8 |
+
Trained For 3 Epochs
|
| 9 |
+
About 25 minutes on T4 GPU
|
| 10 |
+
|
| 11 |
+
Training Loss Steps:
|
| 12 |
+
|
| 13 |
+
Step Training Loss
|
| 14 |
+
25 2.226600
|
| 15 |
+
50 3.159300
|
| 16 |
+
75 2.071800
|
| 17 |
+
100 2.627000
|
| 18 |
+
125 2.100700
|
| 19 |
+
150 2.274000
|
| 20 |
+
175 2.167900
|
| 21 |
+
200 2.250900
|
| 22 |
+
225 1.911300
|
| 23 |
+
250 1.929100
|
| 24 |
+
275 1.961100
|
| 25 |
+
300 1.907300
|
| 26 |
+
325 1.979900
|
| 27 |
+
350 1.697600
|
| 28 |
+
375 2.100700
|
| 29 |
+
400 1.653200
|
| 30 |
+
425 1.912900
|
| 31 |
+
450 1.466200
|
| 32 |
+
475 1.956700
|
| 33 |
+
500 1.397100
|
| 34 |
+
525 2.001500
|
| 35 |
+
550 1.310100
|
| 36 |
+
|
| 37 |
Sample Responses:
|
| 38 |
|
| 39 |
Llama-2 7B Trained On PFAF750
|