Upload README.md with huggingface_hub
Browse files
README.md
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-nc-4.0
|
| 3 |
+
tags: [bagel, vlm-gym, world-model, pacman, dataset]
|
| 4 |
+
---
|
| 5 |
+
# pacman_hard_cot_chunk_k5_train
|
| 6 |
+
|
| 7 |
+
BAGEL VLM-Gym world-model dataset (pacman / cot).
|
| 8 |
+
|
| 9 |
+
- CoT chunk-K train set: all-step interleaved imagined reasoning; re-grounds on the true frame every K=5 steps.
|
| 10 |
+
- layout: Train-only. Gzipped-JSONL shards under `training/`; each row is one packed SFT sample with base64-JPEG frames inline.
|
| 11 |
+
- images are base64-encoded JPEG frames stored inline in each JSONL row.
|
| 12 |
+
|
| 13 |
+
Pairs with the matching `pacman` checkpoint(s) under the companion model org; CoT and non-CoT
|
| 14 |
+
variants share the same underlying rollouts.
|