ArkaAbacus
commited on
Commit
•
cb26edc
1
Parent(s):
b39b843
Update README.md
Browse files
README.md
CHANGED
@@ -26,35 +26,23 @@ This model was built using the Smaug recipe for improving performance on real w
|
|
26 |
########## First turn ##########
|
27 |
score
|
28 |
model turn
|
29 |
-
|
30 |
-
|
31 |
-
llama3-8b-smaug-2-merged-300 1 8.66250
|
32 |
-
base_Meta-Llama-3-8B-Instruct 1 8.53125
|
33 |
-
llama3-8b-smaug-2-merged-450 1 8.42500
|
34 |
########## Second turn ##########
|
35 |
score
|
36 |
model turn
|
37 |
-
|
38 |
-
|
39 |
-
llama3-8b-smaug-2-merged-600 2 7.7250
|
40 |
-
llama3-8b-smaug-2-merged-150 2 7.7125
|
41 |
-
base_Meta-Llama-3-8B-Instruct 2 7.5500
|
42 |
########## Average ##########
|
43 |
score
|
44 |
model
|
45 |
-
|
46 |
-
|
47 |
-
llama3-8b-smaug-2-merged-300 8.200000
|
48 |
-
llama3-8b-smaug-2-merged-450 8.118750
|
49 |
-
base_Meta-Llama-3-8B-Instruct 8.040625
|
50 |
```
|
51 |
|
52 |
| Model | First turn | Second Turn | Average |
|
53 |
| :---- | ---------: | ----------: | ------: |
|
54 |
-
|
|
55 |
-
|
|
56 |
-
| llama3-8b-smaug-2-merged-300 | 8.66 | 7.74 | 8.20 |
|
57 |
-
| llama3-8b-smaug-2-merged-150 | 8.71 | 7.71 | 8.21 |
|
58 |
-
| Meta-Llama-3-8B-Instruct | 8.53 | 7.55 | 8.04 |
|
59 |
|
60 |
This version of Smaug uses new techniques and new data compared to [Smaug-72B](https://huggingface.co/abacusai/Smaug-72B-v0.1), and more information will be released later on. For now, see the previous Smaug paper: https://arxiv.org/abs/2402.13228.
|
|
|
26 |
########## First turn ##########
|
27 |
score
|
28 |
model turn
|
29 |
+
Llama-3-Smaug-8B 1 8.79375
|
30 |
+
Meta-Llama-3-8B-Instruct 1 8.53125
|
|
|
|
|
|
|
31 |
########## Second turn ##########
|
32 |
score
|
33 |
model turn
|
34 |
+
Llama-3-Smaug-8B 2 7.7250
|
35 |
+
Meta-Llama-3-8B-Instruct 2 7.5500
|
|
|
|
|
|
|
36 |
########## Average ##########
|
37 |
score
|
38 |
model
|
39 |
+
Llama-3-Smaug-8B 8.259375
|
40 |
+
Meta-Llama-3-8B-Instruct 8.040625
|
|
|
|
|
|
|
41 |
```
|
42 |
|
43 |
| Model | First turn | Second Turn | Average |
|
44 |
| :---- | ---------: | ----------: | ------: |
|
45 |
+
| Llama-3-Smaug-8B | 8.79 | 7.73 | 8.26 |
|
46 |
+
| Llama-3-8B-Instruct | 8.53 | 7.55 | 8.04 |
|
|
|
|
|
|
|
47 |
|
48 |
This version of Smaug uses new techniques and new data compared to [Smaug-72B](https://huggingface.co/abacusai/Smaug-72B-v0.1), and more information will be released later on. For now, see the previous Smaug paper: https://arxiv.org/abs/2402.13228.
|