regen csvs/new_prompts_PR.csv: exclude 4 synthesized paired-variant tasks (G-21B/G-36B/O-18B/O-19B)

regen csvs/new_prompts_all.csv: exclude 4 synthesized paired-variant tasks (G-21B/G-36B/O-18B/O-19B)

regen csvs/old_prompts_PR.csv: exclude 4 synthesized paired-variant tasks (G-21B/G-36B/O-18B/O-19B)

add csvs/task_fps.json for DiffSyn-style training

add csvs/new_prompts_IF.csv for DiffSyn-style training

add csvs/new_prompts_PR.csv for DiffSyn-style training

add csvs/new_prompts_all.csv for DiffSyn-style training

add csvs/old_prompts_IF.csv for DiffSyn-style training

add csvs/old_prompts_PR.csv for DiffSyn-style training

add tar: tars/train/Instruction_Following/G-27_read_the_chart_data_semantic_comprehension_data-generator.tar

add tar: tars/train/Instruction_Following/G-29_chart_extreme_with_data_data-generator.tar

add tar: tars/train/Instruction_Following/G-25_seperate_object_spinning_data-generator.tar

add tar: tars/train/Instruction_Following/G-26_maintain_object_identity_different_objects_data-generator.tar

add tar: tars/train/Instruction_Following/G-22_attention_shift_same_data-generator.tar

add tar: tars/train/Instruction_Following/G-19_sort_objects_by_rule_data-generator.tar

add tar: tars/train/Instruction_Following/G-200_circle_maximum_value_data-generator.tar

add tar: tars/train/Instruction_Following/G-1_object_trajectory_data-generator.tar

add tar: tars/train/Instruction_Following/G-198_mark_right_angled_triangles_data-generator.tar

add tar: tars/train/Instruction_Following/G-199_locate_line_intersections_data-generator.tar

add tar: tars/train/Instruction_Following/G-195_select_nearest_2_1_rectangle_data-generator.tar

add tar: tars/train/Instruction_Following/G-194_construct_concentric_ring_data-generator.tar

add tar: tars/train/Instruction_Following/G-165_mark_tangent_point_after_motion_data-generator.tar

add tar: tars/train/Instruction_Following/G-158_identify_all_hollow_points_data-generator.tar

add tar: tars/train/Instruction_Following/G-166_highlight_horizontal_lines_data-generator.tar

add tar: tars/train/Instruction_Following/G-162_locate_twelve_o_clock_arrows_data-generator.tar

add tar: tars/train/Instruction_Following/G-163_identify_one_and_nine_data-generator.tar

add tar: tars/train/Instruction_Following/G-146_circle_all_squares_from_mixed_shapes_data-generator.tar

add tar: tars/train/Instruction_Following/G-143_select_box_with_most_dots_data-generator.tar

add tar: tars/train/Instruction_Following/G-141_identify_polygon_with_most_sides_data-generator.tar

add tar: tars/train/Instruction_Following/G-137_identify_figure_in_overlapping_area_data-generator.tar

add tar: tars/train/Instruction_Following/G-11_handle_object_reappearance_data-generator.tar

add tar: tars/train/Instruction_Following/G-138_spot_unique_non_repeated_color_data-generator.tar

add tar: tars/train/Instruction_Following/G-138_spot_unique_non_repeated_color_data-generator.tar

add tar: tars/test/Out-of-Domain_50/Instruction_Following.tar

add tar: tars/test/Out-of-Domain_50/Pure_Reasoning.tar

add tar: tars/test/In-Domain_50/Instruction_Following.tar

add tar: tars/test/In-Domain_50/Pure_Reasoning.tar

docs: Path A — show how to filter or selectively load PR vs IF via class column / TASK_INVENTORY.json

docs: Path B headline is now full-train download+extract, subsets demoted to B.2

add TASK_INVENTORY.json: machine-readable shard listing for scripted selective downloads

add TASK_INVENTORY.md: per-task shard table (PR vs IF, size, full task id)

docs: add per-task train shard inventory + train-focused Path B example

add extract_parquet_to_dir.py: stream parquet shards back to sample-dir layout

docs: rewrite 'How to use' for parquet-shipped repo (schema, Path A direct read, Path B extract to dir)

Add MIXED_TASK_PROMPTS.md: all paraphrase templates for O-10/16/17

Update README counts (unchanged=46) and note mixed-task percentages

Update classification doc: per-task percentages for mixed tasks

Refresh train_pr_unchanged: 2 samples per task (O-10 deep-scan)

Add per-task rewrite-classification doc

Document train_pr_rewritten and train_pr_unchanged splits