PaLI-X (Multi-task FT)

Reported on 4 benchmarks across 2 tasks · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Natural Language Processing3 results

Visual Question Answering (VQA)onDocVQA test
ANLS· uses extra data· 2023-05-29
0.809
best: 0.9436 (Human)
PaLI-X: On Scaling up a Multilingual Vision and Language Model arXiv:2305.18565
Visual Question Answering (VQA)onInfographicVQA
ANLS· uses extra data· 2023-05-29
50.7
best: 80.3 (Gemini Ultra (pixel only))
PaLI-X: On Scaling up a Multilingual Vision and Language Model arXiv:2305.18565
Visual Question Answering (VQA)onChartQA
1:1 Accuracy· uses extra data· 2023-05-29
70.6
best: 81.3 (ChartPaLI-5B + PaLM 2-S)
PaLI-X: On Scaling up a Multilingual Vision and Language Model arXiv:2305.18565

Computer Code1 result

Chart Question AnsweringonChartQA
1:1 Accuracy· uses extra data· 2023-05-29
70.6
best: 81.3 (ChartPaLI-5B + PaLM 2-S)
PaLI-X: On Scaling up a Multilingual Vision and Language Model arXiv:2305.18565