TASK_INVENTORY.md

# Train task inventory

Each train task ships as its own parquet shard at `parquet/train__<task>.parquet`. This file lists all of them so you can pick which shards to download (per-task or per-class) without pulling the whole ~370 GB of train video. **The `class` column inside each row also encodes this distinction, but knowing it up front lets you selectively download.**

- **Pure_Reasoning** (PR): 51 tasks, 194.6 GB
- **Instruction_Following** (IF): 56 tasks, 174.2 GB
- **Total**: 107 tasks shipped as separate train parquet shards

> Test splits do **not** have per-task parquet sharding — both `test_in_domain.parquet` (~90 MB) and `test_out_of_domain.parquet` bundle all of their respective tasks into one file.

## Selective download recipes

```bash
# 1. One specific task (e.g. G-13 grid_number_sequence, ~5.96 GB)
hf download May-apple/VBVR-Reorganized --repo-type dataset \
    --include 'parquet/train__G-13_grid_number_sequence.parquet' \
              'extract_parquet_to_dir.py' 'README.md' \
    --local-dir ./VBVR-Reorganized

# 2. A handful of named tasks (just list each --include pattern):
hf download May-apple/VBVR-Reorganized --repo-type dataset \
    --include 'parquet/train__G-13_*.parquet' \
              'parquet/train__O-44_*.parquet' \
              'parquet/train__O-47_*.parquet' \
              'extract_parquet_to_dir.py' \
    --local-dir ./VBVR-Reorganized

# 3. All Pure_Reasoning train tasks (~195 GB).
#    Use TASK_INVENTORY.json to script up the --include list:
hf download May-apple/VBVR-Reorganized --repo-type dataset \
    --include 'TASK_INVENTORY.json' --local-dir ./VBVR-Reorganized
python3 - <<'PY' > pr_shards.txt
import json
for e in json.load(open("VBVR-Reorganized/TASK_INVENTORY.json"))["Pure_Reasoning"]:
    print(f"parquet/train__{e['shard']}.parquet")
PY
hf download May-apple/VBVR-Reorganized --repo-type dataset \
    --include $(cat pr_shards.txt) 'extract_parquet_to_dir.py' \
    --local-dir ./VBVR-Reorganized
```

`TASK_INVENTORY.json` is the same content as the tables below in machine-readable form (also shipped in this repo).

## Pure_Reasoning train tasks (51)

| Shard short name | Full task id | Size |
|------------------|--------------|-----:|
| `G-12_grid_obtaining_award` | `G-12_grid_obtaining_award_data-generator` | 5.70 GB |
| `G-131_select_next_figure_increasing_size_sequence` | `G-131_select_next_figure_increasing_size_sequence_data-generator` | 2.62 GB |
| `G-132_find_fragment_for_gap_filling` | `G-132_find_fragment_for_gap_filling_data-generator` | 1.97 GB |
| `G-133_select_next_figure_decreasing_size_sequence` | `G-133_select_next_figure_decreasing_size_sequence_data-generator` | 2.16 GB |
| `G-134_select_next_figure_large_small_alternating_sequence` | `G-134_select_next_figure_large_small_alternating_sequence_data-generator` | 2.56 GB |
| `G-13_grid_number_sequence` | `G-13_grid_number_sequence_data-generator` | 5.96 GB |
| `G-14_grid_color_sequence` | `G-14_grid_color_sequence_data-generator` | 5.56 GB |
| `G-15_grid_avoid_obstacles` | `G-15_grid_avoid_obstacles_data-generator` | 5.19 GB |
| `G-16_grid_go_through_block` | `G-16_grid_go_through_block_data-generator` | 8.24 GB |
| `G-17_grid_avoid_red_block` | `G-17_grid_avoid_red_block_data-generator` | 4.56 GB |
| `G-18_grid_shortest_path` | `G-18_grid_shortest_path_data-generator` | 4.50 GB |
| `G-21B_multiple_occlusions_vertical_behind` | `G-21B_multiple_occlusions_vertical_behind_data-generator` | 2.01 GB |
| `G-21_multiple_occlusions_vertical` | `G-21_multiple_occlusions_vertical_data-generator` | 1.97 GB |
| `G-31_directed_graph_navigation` | `G-31_directed_graph_navigation_data-generator` | 2.29 GB |
| `G-32_undirected_graph_navigation` | `G-32_undirected_graph_navigation_data-generator` | 4.17 GB |
| `G-35_hit_target_after_bounce` | `G-35_hit_target_after_bounce_data-generator` | 3.19 GB |
| `G-36B_multiple_occlusions_horizontal_behind` | `G-36B_multiple_occlusions_horizontal_behind_data-generator` | 1.45 GB |
| `G-36_multiple_occlusions_horizontal` | `G-36_multiple_occlusions_horizontal_data-generator` | 1.32 GB |
| `G-41_grid_highest_cost` | `G-41_grid_highest_cost_data-generator` | 8.63 GB |
| `G-43_understand_scene_structure` | `G-43_understand_scene_structure_data-generator` | 3.34 GB |
| `G-44_bfs` | `G-44_bfs_data-generator` | 2.46 GB |
| `G-45_key_door_matching` | `G-45_key_door_matching_data-generator` | 4.30 GB |
| `G-46_find_keys_and_open_doors` | `G-46_find_keys_and_open_doors_data-generator` | 4.50 GB |
| `G-48_multiple_bounces` | `G-48_multiple_bounces_data-generator` | 2.94 GB |
| `G-51_predict_next_color` | `G-51_predict_next_color_data-generator` | 2.05 GB |
| `O-10_shape_outline_fill` | `O-10_shape_outline_fill_data-generator` | 1.88 GB |
| `O-12_shape_color_then_scale` | `O-12_shape_color_then_scale_data-generator` | 2.14 GB |
| `O-13_shape_outline_then_move` | `O-13_shape_outline_then_move_data-generator` | 2.58 GB |
| `O-14_shape_scale_then_outline` | `O-14_shape_scale_then_outline_data-generator` | 3.03 GB |
| `O-15_ball_bounces_given_time` | `O-15_ball_bounces_given_time_data-generator` | 3.35 GB |
| `O-16_color_addition` | `O-16_color_addition_data-generator` | 4.63 GB |
| `O-17_color_subtraction` | `O-17_color_subtraction_data-generator` | 4.41 GB |
| `O-18B_glass_refraction_inverse` | `O-18B_glass_refraction_inverse_data-generator` | 3.25 GB |
| `O-18_glass_refraction` | `O-18_glass_refraction_data-generator` | 3.29 GB |
| `O-19B_mirror_reflection_inverse` | `O-19B_mirror_reflection_inverse_data-generator` | 1.89 GB |
| `O-19_mirror_reflection` | `O-19_mirror_reflection_data-generator` | 2.02 GB |
| `O-1_color_mixing` | `O-1_color_mixing_data-generator` | 1.97 GB |
| `O-21_construction_blueprint` | `O-21_construction_blueprint_data-generator` | 3.81 GB |
| `O-23_domino_chain_branch_path_prediction` | `O-23_domino_chain_branch_path_prediction_data-generator` | 3.45 GB |
| `O-24_domino_chain_gap_analysis` | `O-24_domino_chain_gap_analysis_data-generator` | 5.67 GB |
| `O-29_ballcolor` | `O-29_ballcolor_data-generator` | 4.17 GB |
| `O-31_ball_eating` | `O-31_ball_eating_data-generator` | 4.70 GB |
| `O-44_rotation_puzzle` | `O-44_rotation_puzzle_data-generator` | 3.79 GB |
| `O-45_sequence_completion` | `O-45_sequence_completion_data-generator` | 1.77 GB |
| `O-47_sliding_puzzle` | `O-47_sliding_puzzle_data-generator` | 7.39 GB |
| `O-52_traffic_light` | `O-52_traffic_light_data-generator` | 1.19 GB |
| `O-53_clock` | `O-53_clock_data-generator` | 11.20 GB |
| `O-75_communicating_vessels` | `O-75_communicating_vessels_data-generator` | 6.05 GB |
| `O-7_shape_color_change` | `O-7_shape_color_change_data-generator` | 1.89 GB |
| `O-87_fluid_diffusion_reasoning` | `O-87_fluid_diffusion_reasoning_data-generator` | 7.17 GB |
| `O-8_shape_rotation` | `O-8_shape_rotation_data-generator` | 4.29 GB |

## Instruction_Following train tasks (56)

| Shard short name | Full task id | Size |
|------------------|--------------|-----:|
| `G-11_handle_object_reappearance` | `G-11_handle_object_reappearance_data-generator` | 1.84 GB |
| `G-137_identify_figure_in_overlapping_area` | `G-137_identify_figure_in_overlapping_area_data-generator` | 1.05 GB |
| `G-138_spot_unique_non_repeated_color` | `G-138_spot_unique_non_repeated_color_data-generator` | 0.81 GB |
| `G-141_identify_polygon_with_most_sides` | `G-141_identify_polygon_with_most_sides_data-generator` | 1.46 GB |
| `G-143_select_box_with_most_dots` | `G-143_select_box_with_most_dots_data-generator` | 2.29 GB |
| `G-146_circle_all_squares_from_mixed_shapes` | `G-146_circle_all_squares_from_mixed_shapes_data-generator` | 3.69 GB |
| `G-158_identify_all_hollow_points` | `G-158_identify_all_hollow_points_data-generator` | 4.16 GB |
| `G-162_locate_twelve_o_clock_arrows` | `G-162_locate_twelve_o_clock_arrows_data-generator` | 3.57 GB |
| `G-163_identify_one_and_nine` | `G-163_identify_one_and_nine_data-generator` | 0.74 GB |
| `G-165_mark_tangent_point_after_motion` | `G-165_mark_tangent_point_after_motion_data-generator` | 3.15 GB |
| `G-166_highlight_horizontal_lines` | `G-166_highlight_horizontal_lines_data-generator` | 0.56 GB |
| `G-194_construct_concentric_ring` | `G-194_construct_concentric_ring_data-generator` | 5.95 GB |
| `G-195_select_nearest_2_1_rectangle` | `G-195_select_nearest_2_1_rectangle_data-generator` | 2.09 GB |
| `G-198_mark_right_angled_triangles` | `G-198_mark_right_angled_triangles_data-generator` | 3.14 GB |
| `G-199_locate_line_intersections` | `G-199_locate_line_intersections_data-generator` | 1.26 GB |
| `G-19_sort_objects_by_rule` | `G-19_sort_objects_by_rule_data-generator` | 2.96 GB |
| `G-1_object_trajectory` | `G-1_object_trajectory_data-generator` | 1.95 GB |
| `G-200_circle_maximum_value` | `G-200_circle_maximum_value_data-generator` | 3.04 GB |
| `G-22_attention_shift_same` | `G-22_attention_shift_same_data-generator` | 1.82 GB |
| `G-25_seperate_object_spinning` | `G-25_seperate_object_spinning_data-generator` | 2.72 GB |
| `G-26_maintain_object_identity_different_objects` | `G-26_maintain_object_identity_different_objects_data-generator` | 2.25 GB |
| `G-27_read_the_chart_data_semantic_comprehension` | `G-27_read_the_chart_data_semantic_comprehension_data-generator` | 5.25 GB |
| `G-29_chart_extreme_with_data` | `G-29_chart_extreme_with_data_data-generator` | 3.83 GB |
| `G-2_reorder_objects` | `G-2_reorder_objects_data-generator` | 1.23 GB |
| `G-30_chart_extreme_without_data` | `G-30_chart_extreme_without_data_data-generator` | 2.03 GB |
| `G-33_visual_jenga` | `G-33_visual_jenga_data-generator` | 2.41 GB |
| `G-34_object_packing` | `G-34_object_packing_data-generator` | 1.51 GB |
| `G-37_symmetry_random` | `G-37_symmetry_random_data-generator` | 3.08 GB |
| `G-38_symmetry_shape` | `G-38_symmetry_shape_data-generator` | 4.09 GB |
| `G-39_attention_shift_different` | `G-39_attention_shift_different_data-generator` | 1.76 GB |
| `G-3_stable_sort` | `G-3_stable_sort_data-generator` | 4.02 GB |
| `G-40_combined_objects_spinning` | `G-40_combined_objects_spinning_data-generator` | 2.37 GB |
| `G-49_complete_missing_contour_segments` | `G-49_complete_missing_contour_segments_data-generator` | 2.29 GB |
| `G-4_identify_objects` | `G-4_identify_objects_data-generator` | 1.90 GB |
| `G-50_suppress_spurious_edges` | `G-50_suppress_spurious_edges_data-generator` | 1.99 GB |
| `G-5_multi_object_placement` | `G-5_multi_object_placement_data-generator` | 1.39 GB |
| `G-6_resize_object` | `G-6_resize_object_data-generator` | 1.33 GB |
| `G-7_return_to_correct_bin` | `G-7_return_to_correct_bin_data-generator` | 2.00 GB |
| `G-8_track_object_movement` | `G-8_track_object_movement_data-generator` | 2.41 GB |
| `G-9_identify_objects_in_region` | `G-9_identify_objects_in_region_data-generator` | 1.15 GB |
| `O-25_LEGO_construction_assembly` | `O-25_LEGO_construction_assembly_data-generator` | 3.56 GB |
| `O-30_bookshelf` | `O-30_bookshelf_data-generator` | 4.46 GB |
| `O-32_rolling_ball` | `O-32_rolling_ball_data-generator` | 2.13 GB |
| `O-33_counting_object` | `O-33_counting_object_data-generator` | 1.98 GB |
| `O-34_dot_to_dot_task` | `O-34_dot_to_dot_task_data-generator` | 5.02 GB |
| `O-36_grid_shift` | `O-36_grid_shift_data-generator` | 3.82 GB |
| `O-37_light_sequence` | `O-37_light_sequence_data-generator` | 1.11 GB |
| `O-38_majority_color` | `O-38_majority_color_data-generator` | 2.05 GB |
| `O-3_symbol_reordering` | `O-3_symbol_reordering_data-generator` | 2.98 GB |
| `O-4_symbol_substitution` | `O-4_symbol_substitution_data-generator` | 1.77 GB |
| `O-55_rotation` | `O-55_rotation_data-generator` | 4.42 GB |
| `O-66_animal_color_sorting` | `O-66_animal_color_sorting_data-generator` | 3.39 GB |
| `O-83_planar_warp_verification__chunk0` | `O-83_planar_warp_verification_data-generator` | 10.20 GB |
| `O-83_planar_warp_verification__chunk1` | `O-83_planar_warp_verification_data-generator` | 10.24 GB |
| `O-83_planar_warp_verification__chunk2` | `O-83_planar_warp_verification_data-generator` | 10.22 GB |
| `O-83_planar_warp_verification__chunk3` | `O-83_planar_warp_verification_data-generator` | 10.23 GB |