#322

/capstor/store/cscs/swissai/infra01/vision-datasets/tokenized/alignment/mmpr_v1_2_p2

kind	manual
status	active
samples	n/a
size	67.2 GB
files	14
first seen	2026-07-22 13:38
last seen	2026-07-22 13:38
registered	2026-07-22 13:38

# dataset #322
path: /capstor/store/cscs/swissai/infra01/vision-datasets/tokenized/alignment/mmpr_v1_2_p2
kind: manual | status: active
samples: n/a | size: 67.2 GB | files: 14

fields:
- prompt_id: text
- prompt_len: int
- chosen_len: int
- rejected_len: int
- image_offsets: json
- image_lengths: json
- seq_chosen_len: int
- seq_rejected_len: int
- image_tok: int

tree (truncated):
```
index_train.parquet
index_validation.parquet
manifest.json
media_raw.blob
tokens/train-00000.i32
tokens/validation-00000.i32
train.bin
train.idx
validation.bin
validation.idx
views/train-00000.parquet
views/validation-00000.parquet
views_tokenized/train-00000.parquet
views_tokenized/validation-00000.parquet
```

samples

#	prompt_id	prompt_len	chosen_len	rejected_len	image_offsets	image_lengths	seq_chosen_len	seq_rejected_len	image_tok
1	mmpr-v1.2-CLEVR_math_en_20240402_extracted_pairs_vqa_correctness_rules-0	761	95	58	`[ 17 ]`	`[ 629 ]`	856	819	629
2	mmpr-v1.2-CLEVR_math_en_20240402_extracted_pairs_vqa_correctness_rules-1	705	144	186	`[ 17 ]`	`[ 629 ]`	849	891	629
3	mmpr-v1.2-CLEVR_math_en_20240402_extracted_pairs_vqa_correctness_rules-2	765	66	81	`[ 17 ]`	`[ 629 ]`	831	846	629
4	mmpr-v1.2-CLEVR_math_en_20240402_extracted_pairs_vqa_correctness_rules-3	769	90	65	`[ 17 ]`	`[ 629 ]`	859	834	629
5	mmpr-v1.2-CLEVR_math_en_20240402_extracted_pairs_vqa_correctness_rules-4	771	75	104	`[ 17 ]`	`[ 629 ]`	846	875	629
6	mmpr-v1.2-CLEVR_math_en_20240402_extracted_pairs_vqa_correctness_rules-5	765	40	19	`[ 17 ]`	`[ 629 ]`	805	784	629
7	mmpr-v1.2-CLEVR_math_en_20240402_extracted_pairs_vqa_correctness_rules-6	762	53	92	`[ 17 ]`	`[ 629 ]`	815	854	629
8	mmpr-v1.2-CLEVR_math_en_20240402_extracted_pairs_vqa_correctness_rules-7	698	218	128	`[ 17 ]`	`[ 629 ]`	916	826	629
9	mmpr-v1.2-CLEVR_math_en_20240402_extracted_pairs_vqa_correctness_rules-8	708	200	290	`[ 17 ]`	`[ 629 ]`	908	998	629
10	mmpr-v1.2-CLEVR_math_en_20240402_extracted_pairs_vqa_correctness_rules-9	700	102	99	`[ 17 ]`	`[ 629 ]`	802	799	629