Add raw source archives (moved out of the GitHub repo) and document them

README.md

@@ -77,6 +77,18 @@ per prompt and a reward that compares the sampled workflow with the reference
 match 0.10) without executing anything. Trained checkpoints are published as
 `JesseLiu/agentorch-sft-*` and `JesseLiu/agentorch-rl-*` model repositories.
 
+## Source recipes (`source/`)
+
+The 3,000 training recipes in raw form, packed as archives that unpack to their
+original paths in the code repository. Fetch and verify them with
+`python scripts/download_data.py --only training` (add `--prepared` for the
+SFT / RL files above).
+
+| archive | unpacks to | files |
+|---|---|---:|
+| `source/training_input_v1.tar.gz` | `training/input/recipes/` (v1 prompts, used by every reported run) | 9,000 |
+| `source/training_groundtruth.tar.gz` | `training/groundtruth/recipes/` (reference workflows) | 6,000 |
+
 ## Citation
 
 ```bibtex

source/training_groundtruth.tar.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cfdd3a03434a28c591ae8eb52ce8699de78d62d13fba770165a82ba1aaaa7d74
+size 10570913

source/training_input_v1.tar.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f54f241025333be5f8cbd80d793487b184fa54ad4dbcc63b1ed9c1f191ff2dc0
+size 127004258