Update README.md
Browse files
README.md
CHANGED
@@ -17,6 +17,17 @@ Student model, after fine-tuning, improves upon the performance of the basemodel
|
|
17 |
- gsm8k:
|
18 |
student = 17.06 vs base = 15.54
|
19 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
20 |
|
21 |
I will train it longer in my next run; can do better.
|
22 |
|
|
|
17 |
- gsm8k:
|
18 |
student = 17.06 vs base = 15.54
|
19 |
|
20 |
+
#### Benchmarks
|
21 |
+
|
22 |
+
aloobun/d-Qwen1.5-0.5B:
|
23 |
+
|Avg. | Arc | HellaSwag | MMLU | TruthfulQA | Winogrande | GSM8K |
|
24 |
+
|---|---|---|---|---|---|---|
|
25 |
+
|38.07 | 30.29 |47.75 | 38.21 | **39.29** | 55.8 | **17.06** |
|
26 |
+
|
27 |
+
Qwen/Qwen1.5-0.5B:
|
28 |
+
|Avg. | Arc | HellaSwag | MMLU | TruthfulQA | Winogrande | GSM8K |
|
29 |
+
|---|---|---|---|---|---|---|
|
30 |
+
|38.62 | 31.48 |49.05 | 39.35 | **38.3** | 57.22 | **16.3** |
|
31 |
|
32 |
I will train it longer in my next run; can do better.
|
33 |
|