Datasets:

AbstractPhil
/

qwen-thinking-images

Tasks:

Modalities:

Formats:

Languages:

Size:

Tags:

image-captioning

chain-of-thought

symbolic-thinking

Libraries:

License:

Dataset card Data Studio Files Files and versions

qwen-thinking-images

1.94 GB

Ctrl+K

Ctrl+K

1 contributor

History: 173 commits

AbstractPhil's picture

captions 20261105_030856_deepfashion_mm_pose: 30720 rows

529965f verified 1 day ago

extractions
captions 20261105_030856_deepfashion_mm_pose: 30720 rows 1 day ago
manifests
manifests: the 725 quick-look DeepFashion photos of v1.5 set to a closer look (his tested prompt for the helped extraction) 3 days ago
mixed
the lookers' rows mixed over 27257 pictures (M_mix_v1): one split per looker, the agreement pick, the mixed table and its read 4 days ago
pictures
manifest v1.1: the AIC pictures carried in pictures/aic_v1.zip, the DOCCI pictures from DOCCI's image archive, long looks asked as closer looks; the card 7 days ago
rows
the lookers' rows mixed over 27257 pictures (M_mix_v1): one split per looker, the agreement pick, the mixed table and its read 4 days ago
.gitattributes

5.74 kB
captions 20261105_030856_deepfashion_mm_pose: 4096 rows 1 day ago
README.md

19.7 kB
the card: the rows config gains one split per looker and the agreement split; the mixed config; the untrained list and the trained ids; the v1 'nothing seen in training' sentence corrected 4 days ago
counts.json

9.03 kB
manifest v1.5: 2,193 more Met objects in the `more` split; the card 6 days ago
manifest_v1.1.jsonl

61.4 MB
xet

manifest v1.1: the AIC pictures carried in pictures/aic_v1.zip, the DOCCI pictures from DOCCI's image archive, long looks asked as closer looks; the card 7 days ago
manifest_v1.1.parquet

6.41 MB
xet

manifest v1.1: the AIC pictures carried in pictures/aic_v1.zip, the DOCCI pictures from DOCCI's image archive, long looks asked as closer looks; the card 7 days ago
manifest_v1.2.jsonl

63.8 MB
xet

manifest v1.2: the Met joins the list (3,000 objects with a picture); the card 7 days ago
manifest_v1.2.parquet

6.59 MB
xet

manifest v1.2: the Met joins the list (3,000 objects with a picture); the card 7 days ago
manifest_v1.3.jsonl

74.4 MB
xet

manifest v1.3: a `more` split of 11,539 museum rows (NGA open data + the Met) for the medium-corrected training rows; the card 6 days ago
manifest_v1.3.parquet

7.67 MB
xet

manifest v1.3: a `more` split of 11,539 museum rows (NGA open data + the Met) for the medium-corrected training rows; the card 6 days ago
manifest_v1.4.jsonl

75.6 MB
xet

manifest v1.4: 1,241 more Met objects (drawings and paintings) in the `more` split; the card 6 days ago
manifest_v1.4.parquet

7.77 MB
xet

manifest v1.4: 1,241 more Met objects (drawings and paintings) in the `more` split; the card 6 days ago
manifest_v1.5.jsonl

77.5 MB
xet

manifest v1.5: 2,193 more Met objects in the `more` split; the card 6 days ago
manifest_v1.5.parquet

7.92 MB
xet

manifest v1.5: 2,193 more Met objects in the `more` split; the card 6 days ago
manifest_v1.5_untrained.jsonl

37.5 MB
xet

manifest_v1.5_untrained.jsonl: the list's 30,722 pictures that no training set ever used (every sft/ and synth/ file of the training repo, matched by sample id); trained_ids_v1.json: the ids each training set used 5 days ago
manifest_v1.jsonl

59.8 MB
xet

manifest v1: the picture list (21 sources, every group; the pilot split of 120 a source), its parquet twin, counts and the card 7 days ago
manifest_v1.parquet

6.39 MB
xet

manifest v1: the picture list (21 sources, every group; the pilot split of 120 a source), its parquet twin, counts and the card 7 days ago
pilot_v1.1.jsonl

3.22 MB
manifest v1.1: the AIC pictures carried in pictures/aic_v1.zip, the DOCCI pictures from DOCCI's image archive, long looks asked as closer looks; the card 7 days ago
pilot_v1.jsonl

3.14 MB
manifest v1: the picture list (21 sources, every group; the pilot split of 120 a source), its parquet twin, counts and the card 7 days ago
trained_ids_v1.json

1.65 MB
manifest_v1.5_untrained.jsonl: the list's 30,722 pictures that no training set ever used (every sft/ and synth/ file of the training repo, matched by sample id); trained_ids_v1.json: the ids each training set used 5 days ago