3-way Sentiment on Twitter-Fin
83.6Accuracygpt-oss-120B
Evaluation results
| Method | Links | |
|---|---|---|
| gpt-oss-120BN (#rows)=1,000, K (#candidate labels)=3, Decoding Strategy=prefill-only logit scoring2026.05 | 83.6 | |
| Llama-3.3-70B-Instruct-FP8N (#rows)=1,000, K (#candidate labels)=3, Decoding Strategy=greedy2026.05 | 71.9 |