Document raw SPARC VQA corpus and filtered release

README.md

@@ -1,5 +1,5 @@
 ---
-pretty_name: SPARC VQA
+pretty_name: SPARC VQA Raw
 tags:
   - robotics
   - vision-language
@@ -7,9 +7,28 @@ tags:
   - sparc
 ---
 
-# SPARC VQA
+# SPARC VQA Raw
 
-SPARC VQA is the generated spatial VQA training dataset used in the SPARC Qwen3.5 model releases. It contains 838,211 training examples in one Parquet shard (33.20 GB). Each example embeds its image bytes and includes a question, answer, task type, target type, source dataset identifier, split, and JSON metadata.
+This repository contains the unfiltered SPARC VQA corpus: 838,211 embedded-image training examples in `train.parquet` (33.20 GB). Each example contains an image, question, answer, task metadata, source identifier, and annotation metadata including `selected_start_score`.
+
+## Ready-to-train version
+
+For the exact SPARC subset used by the released Qwen3.5 models, use [irl-kit/SPARC-VQA](https://huggingface.co/datasets/irl-kit/SPARC-VQA). Its `train_filtered_t097_mpo700.parquet` contains 284,909 examples after the release filtering and does not require SPARC postprocessing.
+
+## Reproduce the release filter
+
+`export_sparc_training_subset.py` materializes the exact release subset from `train.parquet`:
+
+```bash
+python export_sparc_training_subset.py \
+  train.parquet \
+  train_filtered_t097_mpo700.parquet \
+  --quality-threshold 0.97 \
+  --max-per-object 700 \
+  --blocked-vacant-location gripper
+```
+
+The filter applies `selected_start_score >= 0.97`, sorts retained records by descending score, retains complete trajectory-subtask groups for the first 700 groups per normalized object phrase, and removes vacant-location prompts referring to a gripper.
 
 ## Data schema
 
@@ -22,29 +41,7 @@ SPARC VQA is the generated spatial VQA training dataset used in the SPARC Qwen3.
 | `target_type`, `task_type` | Spatial target and VQA task labels |
 | `source`, `split`, `metadata` | Provenance and generation metadata |
 
-## Mixtures and models
-
-| Release | Data mixture |
-| --- | --- |
-| Qwen3.5-4B | SPARC VQA + FSD + RoboPoint + LLaVA-OneVision2 |
-| Qwen3.5-0.8B-VTFT | SPARC VQA + FSD + RoboPoint + LLaVA-OneVision2 |
-| Qwen3.5-9B-EO | SPARC VQA + FSD + RoboPoint + LLaVA-OneVision2 + EO-1.5M |
-
-See [irl-kit/SPARC-VQA-Mixture](https://huggingface.co/datasets/irl-kit/SPARC-VQA-Mixture) for the reproducible mixture manifest. The released models are [Qwen3.5-4B](https://huggingface.co/irl-kit/SPARC-Qwen3.5-4B), [Qwen3.5-0.8B-VTFT](https://huggingface.co/irl-kit/SPARC-Qwen3.5-0.8B-VTFT), and [Qwen3.5-9B-EO](https://huggingface.co/irl-kit/SPARC-Qwen3.5-9B-EO).
-
-## Prompting compatibility
-
-The released models are sensitive to output formatting. Use each model's bundled chat template with a single user turn containing the image followed by the question. For point prediction, use this suffix:
-
-```text
-Output the point coordinates in JSON format like [{"point_2d": [x, y], "label": "target"}]. Use integer coordinates between 0 and 1000.
-```
-
-For trajectories or multiple points, use this suffix:
-
-```text
-Return only a JSON list like [{"point_2d": [x1, y1], "label": "point_1"}, {"point_2d": [x2, y2], "label": "point_2"}, ...]. Use integer coordinates between 0 and 1000.
-```
+`ours_adaptive_det_soft_snr_sp8.yaml` records the raw SPARC VQA generation configuration.
 
 ## Citation