TmpDataset / combined_data_stats.json
mmaaz60's picture
Upload combined_data_stats.json with huggingface_hub
0d9b80c verified
{
"total_samples": 15271,
"data_sources": {
"1_2_m_academic_v0_1": 865,
"None": 8262,
"30_60_s_academic_v0_1": 2096,
"0_30_s_perceptiontest": 470,
"0_30_s_academic_v0_1": 1411,
"2_3_m_academic_v0_1": 806,
"0_30_s_nextqa": 131,
"1_2_m_nextqa": 95,
"30_60_s_perceptiontest": 167,
"30_60_s_nextqa": 139,
"clevrer": 829
},
"assistant_stats": {
"characters": {
"min": 37.0,
"max": 10918.0,
"avg": 1817.258201820444,
"count": 15271
},
"words": {
"min": 4.0,
"max": 1918.0,
"avg": 321.41202278829155,
"count": 15271
},
"tokens": {
"min": 9.25,
"max": 2729.5,
"avg": 454.314550455111,
"count": 15271,
"definition": "tokens = characters / 4.0"
}
},
"unique_videos": {
"total_unique_overall": 11816,
"per_data_source_unique_counts": {
"1_2_m_academic_v0_1": 635,
"None": 6349,
"30_60_s_academic_v0_1": 1563,
"0_30_s_perceptiontest": 402,
"0_30_s_academic_v0_1": 1060,
"2_3_m_academic_v0_1": 551,
"0_30_s_nextqa": 121,
"1_2_m_nextqa": 92,
"30_60_s_perceptiontest": 140,
"30_60_s_nextqa": 131,
"clevrer": 814
}
},
"video_metadata": {
"ffprobe_available": true,
"probed_videos_count": 11816,
"duration_seconds": {
"min": 5.0,
"max": 179.932,
"mean": 48.94285781990521,
"count": 11816
},
"resolution_groups": {
"below_480p": 5463,
"between_480p_720p": 2514,
"between_720p_1080p": 2298,
"above_1080p": 1541,
"unknown": 0
},
"errors_total": 0,
"errors_sample": []
}
}