Time Series Reasoning on TRQA
83.34AccuracyARTIST
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ARTISTRL=true, SFT=true2026.02 | 83.34 | — | |
| ARTISTmode=+ SFT + RL2026.02 | 83.06 | 78.02 | |
| ARTISTmode=+ SFT2026.02 | 82.26 | 62.32 | |
| ARTISTSFT=true2026.02 | 82.26 | — | |
| T2SPBackbone=GPT-5.42026.06 | 81.5 | 89.8 | |
| ITFormer-4Bmode=+ SFT2026.02 | 80.12 | 74.22 | |
| OpenTSLM-4Bmode=+ SFT2026.02 | 76.25 | 69.36 | |
| T2SPBackbone=Claude-haiku-4.52026.06 | 76 | 86.4 | |
| T2SPBackbone=Gemini-3.1-flash-lite2026.06 | 76 | 86.4 | |
| Qwen-3.5-9BRepresentation=Vision2026.06 | 75.5 | 86 | |
| Claude-haiku-4.5Representation=Vision2026.06 | 75 | 85.7 | |
| GPT-5.4Representation=Raw2026.06 | 74 | 85.1 | |
| Qwen-3.5-9BRepresentation=Raw2026.06 | 73.5 | 84.7 | |
| Gemini-3.1-flash-liteRepresentation=Raw2026.06 | 73 | 84.4 | |
| T2SPBackbone=Qwen-3.5-9B2026.06 | 72.7 | 84.2 | |
| Gemini-3.1-flash-liteRepresentation=Vision2026.06 | 72.5 | 71.9 | |
| TimeMasterRL=true2026.02 | 72.08 | — | |
| GPT-5.4Representation=Vision2026.06 | 71 | 83 | |
| Claude-haiku-4.5Representation=Raw2026.06 | 70 | 82.4 | |
| ChatTS-14Bmode=+ SFT2026.02 | 69 | 55.57 | |
| TimeMaster2026.02 | 68.45 | — | |
| Base Model2026.02 | 66.09 | — | |
| VL-Timefew-shot=true2026.02 | 58.5 | — | |
| Llama-3 8Bstatistics=w/o statistics2026.02 | 57.69 | 53.88 | |
| ChatTS-14Bmode=Base Model2026.02 | 57.5 | 42.12 | |
| Llama-3 8Bstatistics=w/ statistics2026.02 | 55.5 | 53.02 | |
| VL-Time2026.02 | 53.5 | — | |
| Random Guess2026.02 | 37.13 | 37.13 | |
| Random Guess2026.06 | 36.9 | 32.2 | |
| Qwen3-14Bstatistics=w/o statistics2026.02 | 33 | 29.54 | |
| ChatTime-7B2026.06 | 31.4 | 24.5 | |
| Qwen3-14Bstatistics=w/ statistics2026.02 | 29 | 30.32 | |
| GPT-5statistics=w/ statistics2026.02 | 25 | 28.7 | |
| GPT-5statistics=w/o statistics2026.02 | 21.5 | 21.67 | |
| Qwen3-8Bstatistics=w/ statistics2026.02 | 15.25 | 14.02 | |
| Qwen3-8Bstatistics=w/o statistics2026.02 | 14.63 | 15.2 |