Time Series Reasoning on TimeMQA
50AccuracyGPT-4o
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-4oModalities=Textual2026.02 | 50 | 7.75 | |
| PATRA 7BModalities=Textual + Temporal, Backbone=Qwen2.5 7B2026.02 | 44.59 | 27.36 | |
| TimeOmni-1 7BModalities=Textual2026.02 | 42.56 | 12.78 | |
| Qwen2.5 7BModalities=Textual2026.02 | 36.48 | 18.72 | |
| ITFormer 7BModalities=Textual + Temporal2026.02 | 30.4 | 15.58 | |
| Llama3 8BModalities=Textual2026.02 | 27.02 | 5.26 | |
| Llama3 8BModalities=Textual, Task Protocol=TimeMQA2026.02 | 22.97 | 8.86 | |
| ChatTS 7BModalities=Textual + Temporal2026.02 | 22.97 | 15.84 | |
| Deepseek-R1 7BModalities=Textual2026.02 | 18.24 | 10.34 | |
| Qwen2.5 7BModalities=Textual, Task Protocol=TimeMQA2026.02 | 13.51 | 16.89 | |
| Mistral 7BModalities=Textual, Task Protocol=TimeMQA2026.02 | 12.83 | 9.36 |