Update README.md
Browse files
README.md
CHANGED
|
@@ -62,6 +62,30 @@ in my last tests with length 2048, I got great models, I trained models in 24 ho
|
|
| 62 |
In training, step 2/6
|
| 63 |
Each stage lasts 4-6 hours.
|
| 64 |
I am releasing the partial models, in the end I will also release the datasets. 100% synthetic data in markdown
|
| 65 |
-
1 - OK
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 66 |
2 - RUNNING - next upload 9/9 - 00:30 GMT
|
| 67 |
3 -
|
|
|
|
| 62 |
In training, step 2/6
|
| 63 |
Each stage lasts 4-6 hours.
|
| 64 |
I am releasing the partial models, in the end I will also release the datasets. 100% synthetic data in markdown
|
| 65 |
+
1 - 2.5h OK result :
|
| 66 |
+
(if you have problems on eval, set same max_length)
|
| 67 |
+
| Task |Version|Metric|Value | |Stderr|
|
| 68 |
+
|----------|------:|------|-----:|---|-----:|
|
| 69 |
+
|winogrande| 0|acc |0.5162|± | 0.014|
|
| 70 |
+
|
| 71 |
+
hf-causal (max_length=3200), limit: None, provide_description: False, num_fewshot: 0, batch_size: None
|
| 72 |
+
| Task |Version| Metric |Value | |Stderr|
|
| 73 |
+
|----------|------:|--------|-----:|---|-----:|
|
| 74 |
+
|openbookqa| 0|acc |0.1380|± |0.0154|
|
| 75 |
+
| | |acc_norm|0.3420|± |0.0212|
|
| 76 |
+
|piqa | 0|acc |0.6289|± |0.0113|
|
| 77 |
+
| | |acc_norm|0.6251|± |0.0113|
|
| 78 |
+
|
| 79 |
+
hf-causal (max_length=1280), limit: None, provide_description: False, num_fewshot: 0, batch_size: None
|
| 80 |
+
| Task |Version| Metric |Value | |Stderr|
|
| 81 |
+
|-------------|------:|--------|-----:|---|-----:|
|
| 82 |
+
|arc_challenge| 0|acc |0.1903|± |0.0115|
|
| 83 |
+
| | |acc_norm|0.2270|± |0.0122|
|
| 84 |
+
|hellaswag | 0|acc |0.2892|± |0.0045|
|
| 85 |
+
| | |acc_norm|0.3114|± |0.0046|
|
| 86 |
+
|
| 87 |
+
|
| 88 |
+
|
| 89 |
+
|
| 90 |
2 - RUNNING - next upload 9/9 - 00:30 GMT
|
| 91 |
3 -
|