Add tabarena_text_cache.parquet for 16 text-bearing datasets

Precomputed sentence embeddings (SemanticTextFeatureGenerator output) keyed by unique text values; lets benchmark runs skip the embedding step. README + AGENTS updated to document the new artifact.

AGENTS.md

@@ -20,7 +20,7 @@ change for Data Foundry — treat it as a data release, not source code.
 <dataset_name>/versions/<uuid>/...        # versioned wrapper (10 large non-IID datasets)
 ```
 
-Each leaf `<uuid>/` directory contains exactly six files and nothing else:
+Each leaf `<uuid>/` directory contains six core files, plus an optional seventh:
 
 ```
 dataset.parquet                                       # the table
@@ -29,16 +29,22 @@ container_metadata.json                               # uuid + sha256 checksum
 dataset_metadata.dataset-mold-v1.json                 # provenance & curation notes
 task_metadata.predictive-ml-task-mold-v1.json         # target, problem type, metric, split keys
 experiment_metadata.predictive-ml-splits-mold-v1.json # CV fold indices
+tabarena_text_cache.parquet                           # (optional) precomputed sentence embeddings — only for the 16 text-bearing datasets
 ```
 
+`tabarena_text_cache.parquet` is produced by TabArena's `SemanticTextFeatureGenerator.save_embedding_cache`
+(index `text`, columns `0..D-1`). It is **not** managed by Data Foundry — it's a TabArena-side artifact
+that lets benchmark runs skip the embedding step. Treat it like the other artifacts: do not regenerate
+by hand; if it needs to change, re-run the upstream cache generator and replace the file.
+
 Top-level files: `README.md` (the Hugging Face dataset card), `LICENSE`, `dataset_references.bib`,
 and this file.
 
 ## Editing rules
 
-- **Do not rename, move, or delete dataset directories or any of the six artifact files inside
+- **Do not rename, move, or delete dataset directories or any of the artifact files inside
   them.** UUIDs are pinned references from Data Foundry's `final_uuid_list.py`; renaming silently
-  breaks downstream consumers.
+  breaks downstream consumers. The same applies to the optional `tabarena_text_cache.parquet`.
 - **Do not regenerate `dataset.parquet`, `container_metadata.json`, or any `*_metadata.*.json` by
   hand.** They are produced by `CuratedContainer.save()` in Data Foundry. If a dataset needs to
   change, the change happens in a Data Foundry curation notebook and the new artifact is uploaded

README.md

@@ -846,7 +846,7 @@ The release ships as a flat bundle of 142 datasets. Each dataset lives in its ow
 <dataset_name>/versions/<uuid>/...        # versioned wrapper (10 large non-IID datasets)
 ```
 
-Each directory contains exactly six files:
+Each directory contains six core files, plus an optional `tabarena_text_cache.parquet` for the 16 datasets with text features:
 
 ```
 <uuid>/
@@ -855,11 +855,37 @@ Each directory contains exactly six files:
 ├── container_metadata.json                               # uuid + sha256 checksum
 ├── dataset_metadata.dataset-mold-v1.json                 # provenance & curation notes
 ├── task_metadata.predictive-ml-task-mold-v1.json         # target, problem type, metric, split keys
-└── experiment_metadata.predictive-ml-splits-mold-v1.json # CV fold indices
+├── experiment_metadata.predictive-ml-splits-mold-v1.json # CV fold indices
+└── tabarena_text_cache.parquet                           # (optional) precomputed sentence embeddings keyed by text
 ```
 
 For details on files and the metadata structure, checkout [DataFoundry](https://github.com/TabArena/data-foundry)!
 
+### Text embedding cache (`tabarena_text_cache.parquet`)
+
+Shipped for the 16 text-bearing datasets listed below. The file is a pandas DataFrame written via
+`SemanticTextFeatureGenerator.save_embedding_cache` (see TabArena's `text_feature_generators.py`):
+
+- **Index** — a string column named `text` containing every unique text value observed across all text
+  columns of `dataset.parquet`.
+- **Columns** — `0, 1, …, D-1`, holding the precomputed sentence embedding for each text value
+  (default model: 32-dim embeddings).
+
+Reload with:
+
+```python
+import pandas as pd
+df = pd.read_parquet("<dataset_name>/<uuid>/tabarena_text_cache.parquet")
+cache = dict(zip(df.index, df.to_numpy()))   # {text: np.ndarray}
+```
+
+This lets you skip the embedding step at fit time. Datasets with a `tabarena_text_cache.parquet`:
+`coffee_rating_prediction`, `consumer_complaints` (1m variant), `california_house_prices_2020`,
+`drug_induced_autoimmunity_prediction`, `emscad`, `immoscout_german_house_prices`, `kickstarter`,
+`labour_inspection_compliance`, `lending_club` (1m variant), `mercari_price_suggestion` (1m variant),
+`mutual_funds_india`, `pva_revenue_prediction_kddcup98`, `regensburg_pediatric_appendicitis`,
+`sf_permit_time`, `wids_diabetes_mellitus`, `wine_world_cost`.
+
 ### Loading a single dataset directly
 
 Each per-dataset config in this card's frontmatter routes only `dataset.parquet`, which is enough to get

california_house_prices_2020/019d7388-c562-72b3-a533-5f588dc737c9/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b0828fc15eaba09d5983652505717ab88f51b194aa9984a25908e11e309afed
+size 25235866

coffee_rating_prediction/019d7388-dffd-7bc4-ab3d-40bf6581786e/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fca793d846aac7e0685c6914620885c388ec0bbd0ebbbb6fc4c430f41bf683a2
+size 1238263

consumer_complaints/versions/019d738b-4c6e-751f-972a-5f63b1508f70/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:330ffc6221b1c89878fd0f312cccd829bded22d0a51d8c3165e307042aecbf65
+size 532864481

drug_induced_autoimmunity_prediction/019d736a-b4b1-7f51-9563-dde2b6ecd948/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8ae8d80eb94fe0c4dd028cd327a8d3c3b628ee9ff7ad61ede41f8df43f3766a6
+size 92880

emscad/019d736a-ed4a-7aaa-89df-d52558565db9/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f65ba98422c1dd5c78f02244cb76c1f83e1ab0c1f6e524e50195f1b2cafa808
+size 20474647

immoscout_german_house_prices/019d7374-a0e4-7646-b11b-f762fe9b693b/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:124dbe6ea634f481e2e05e06989a7a7bc0fd1050ecac2f6fd9e6419587f0950c
+size 580768

kickstarter/019db497-6b35-71e0-9c6b-f4a7286f8f6f/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5531807ca9fbfc36ee2bdf063f87d6f0e9a29bb03b71d41218d04bfe130d88d
+size 55683027

labour_inspection_compliance/019d736e-5a12-720a-9281-b652d1acae27/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d6d3c1b123441381b996e19363d020ee233c0810d233cdcbcd007af7fe4d5b1c
+size 498975

lending_club/versions/019d738c-5edd-7437-9116-e3bc87a8b0c5/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c72d3037ffb066ef420caa55d7f56af89df786e3b4606f4108da15fc2d09072
+size 43958423

mercari_price_suggestion/versions/019d736f-26fd-73bd-8853-dc219e5f4ed5/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3630342f9e335e1c139b9b92f02b778c690872f669134cca1c3771fe2f87bd7e
+size 257006890

mutual_funds_india/019d736f-4e6b-7953-9a01-a5476b6708ef/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:215d23552d13666664b1a13b048840f64412e77e9f0a8bffbbecf334bb07ce4a
+size 133597

pva_revenue_prediction_kddcup98/019db49b-9c05-72b8-b4a5-7b967a18225b/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41b9d3f69dce5d1c9768a2fba40bad29c2c35d330274277f468a3006b9b90337
+size 1662865

regensburg_pediatric_appendicitis/019d7370-1dc6-7c1e-8ff1-80ddeb4c4207/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:31da2c819c5dce8000ba278894c7eecd46fc925e25486becb58c7306cb045144
+size 20662

sf_permit_time/019d738d-80c6-71a6-99f0-e3e0955ce2d0/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca320be7d5f0d7bbe65e34e741f8faa6e8c8f52f8a0d6519fc30f3dca2878888
+size 16987118

wids_diabetes_mellitus/019d7374-50ba-7416-9d10-a526040930ee/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a54d79286ae52303141087b3f34df4255a7244d03ada87dcf3e7ff1af1adf218
+size 61955

wine_world_cost/019d7374-6961-7fb3-9ca1-2e20e35fbbb0/tabarena_text_cache.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55b5877369b9db50a048e87708525dd5e73844488f86e9ef977de0025d4b0065
+size 860183