point at hud v6
Browse files
README.md
CHANGED
|
@@ -29,3 +29,14 @@ configs:
|
|
| 29 |
- split: train
|
| 30 |
path: data/train-*
|
| 31 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
- split: train
|
| 30 |
path: data/train-*
|
| 31 |
---
|
| 32 |
+
|
| 33 |
+
## ⚠️ Migrated to HUD v6
|
| 34 |
+
|
| 35 |
+
This taskset now runs on the HUD v6 SDK. The canonical version lives in
|
| 36 |
+
[hud-evals/hud-text-2048](https://github.com/hud-evals/hud-text-2048)
|
| 37 |
+
(`tasks.py`) and as a taskset on the [HUD platform](https://hud.ai).
|
| 38 |
+
|
| 39 |
+
The rows in this dataset are the legacy v5 format (`mcp_config` +
|
| 40 |
+
`setup_tool`/`evaluate_tool`) and are kept for reproducibility. They will
|
| 41 |
+
not work with `hud >= 0.6`.
|
| 42 |
+
|