Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
dataset_info:
|
| 3 |
+
features: []
|
| 4 |
+
splits:
|
| 5 |
+
- name: train
|
| 6 |
+
- name: val
|
| 7 |
+
license: apache-2.0
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# Long Health & Long Code — filtered ≤55k encoder length
|
| 11 |
+
|
| 12 |
+
Derived from **agu18dec/longhealth-longcode-50k** by removing samples whose total encoder length exceeds 55,000 tokens.
|
| 13 |
+
|
| 14 |
+
- Train kept: 62251 / 90000
|
| 15 |
+
- Val kept: 6875 / 10000
|
| 16 |
+
- Dropped: train 27749, val 3125
|
| 17 |
+
|
| 18 |
+
File: `mixture_filtered_55k.msgpack` (msgpack with keys: `train`, `val`).
|