Causal Reasoning on XCOPA ET
71.8AccuracyLlama-3.2-3B
Evaluation Results
| Method | Links | |
|---|---|---|
| Llama-3.2-3BTraining Status=Continued pretraining, Tokenizer Strategy=prune+ext (16k, ours)2025.12 | 71.8 | |
| Llama-3.2-3BTraining Status=Continued pretraining, Tokenizer Strategy=default2025.12 | 70.4 | |
| Llama-3.2-3BTraining Status=Continued pretraining, Tokenizer Strategy=prune+ext (16k, naive)2025.12 | 69.8 | |
| Llama-3.2-1BTraining Status=Continued pretraining, Tokenizer Strategy=default2025.12 | 63.6 | |
| Llama-3.2-1BTraining Status=Continued pretraining, Tokenizer Strategy=prune+ext (16k, ours)2025.12 | 63 | |
| Llama-3.2-1BTraining Status=Continued pretraining, Tokenizer Strategy=prune+ext (16k, naive)2025.12 | 62.6 | |
| Llama-3.2-3BTraining Status=No training2025.12 | 55.4 | |
| Llama-3.2-1BTraining Status=No training2025.12 | 51 |