Causal Reasoning on XCOPA (ET, ID, SW, TR Subsets)
64.2XCOPA Causal Reasoning Scoreasafaya/kanarya-2b
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| asafaya/kanarya-2bParams=2B2026.03 | 64.2 | — | — | — | — | |
| Trendyol-LLM-7b-baseParams=7B2026.03 | 61 | — | — | — | — | |
| CohereForAI/aya-101Params=13B2026.03 | 59.6 | — | — | — | — | |
| boun-tabi-LMG/TURNAParams=1.1B2026.03 | 55.8 | — | — | — | — | |
| Llama-3.2-3B-InstructParams=3B2026.03 | 54.6 | — | — | — | — | |
| vngrs-ai/Kumru-2BParams=2B2026.03 | 54 | — | — | — | — | |
| DiffutronLM-0.3B-1st-StageParams=0.3B2026.03 | 53.8 | — | — | — | — | |
| DiffutronLM-0.3B-2nd-StageParams=0.3B2026.03 | 52 | — | — | — | — | |
| BLOOMZ-7B1Parameters=7.1B2026.01 | — | 48 | 67 | 60 | 48 | |
| GoldfishBackbone=GPT-2, Training=from-scratch2026.01 | — | 52 | 58.4 | 55 | 56.2 | |
| GPT-2 + FTBackbone=GPT-2, Training=finetuned2026.01 | — | 47.6 | 48.4 | 52.8 | 51 | |
| MCLBackbone=GPT-2, Alignment=dense-representation2026.01 | — | 47.6 | 51.2 | 51.2 | 51.4 | |
| SENSIABackbone=GPT-2, Alignment=sense-aligned2026.01 | — | 51.2 | 51.6 | 55.2 | 55.4 | |
| XGLM-7B5Parameters=7.5B2026.01 | — | 57 | 58 | 57 | 56 |