Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cdla-sharing-1.0
|
| 3 |
+
---
|
| 4 |
+
|
| 5 |
+
# tinydark-gram-data
|
| 6 |
+
|
| 7 |
+
The exact data behind [tinydark-gram](https://github.com/hbfreed/tinydark-gram): GRAM (gradient-routed
|
| 8 |
+
auxiliary modules) on the time-of-day backdoor from Tiny Dark Stories.
|
| 9 |
+
|
| 10 |
+
- `tok_grid/`: TinyStories V2 and 242,688 normal/spooky story pairs, tokenized with a 4096-token BPE
|
| 11 |
+
(`tokenizer.json`). `.bin` files are uint16 token ids, `.idx` files are int64 story offsets.
|
| 12 |
+
- `pairs_train.jsonl`, `pairs_eval.jsonl`: the story pairs as text, used to fit the spooky classifier.
|
| 13 |
+
|
| 14 |
+
This is an earlier, smaller set of pairs than the released
|
| 15 |
+
[tiny-dark-stories](https://huggingface.co/datasets/hbfreed/tiny-dark-stories) dataset. Use that one for
|
| 16 |
+
anything other than reproducing tinydark-gram.
|
| 17 |
+
|
| 18 |
+
```bash
|
| 19 |
+
uvx hf download hbfreed/tinydark-gram-data --repo-type dataset --local-dir data
|
| 20 |
+
```
|