Create README file
Browse files
README.md
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model:
|
| 3 |
+
- EleutherAI/pythia-160m
|
| 4 |
+
---
|
| 5 |
+
|
| 6 |
+
## Model Description
|
| 7 |
+
|
| 8 |
+
This is the pythia-160m from EleutherAI re-uploaded as an exercise.
|
| 9 |
+
|
| 10 |
+
## Evaluation Results
|
| 11 |
+
|
| 12 |
+
According to project requirement, we used lm-evalutation-harness from EleutherAI to evaluate pythia-160m on the 'Hellaswag' benchmark.
|
| 13 |
+
|
| 14 |
+
### Hellaswag
|
| 15 |
+
|
| 16 |
+
| Tasks |Version|Filter|n-shot| Metric | |Value | |Stderr|
|
| 17 |
+
|---------|------:|------|-----:|--------|---|-----:|---|-----:|
|
| 18 |
+
|hellaswag| 1|none | 0|acc |↑ |0.2872|± |0.0045|
|
| 19 |
+
| | |none | 0|acc_norm|↑ |0.3082|± |0.0046|
|