Datasets:

---

alexdeath53

/

ruletaker-vllm-reasoning-3

Like 0

The default config contains 46 while it should generally contain 3 splits maximum (train/validation/test). If the splits 20260422_143346_100_samples_Qwen3_8B, 20260422_153600_100_samples_gemma_4_26B_A4B, 20260422_154318_100_samples_gemma_4_26B_A4B, 20260422_162321_10_samples_Qwen3_8B, 20260422_163217_10_samples_Qwen3_8B... are not used to differentiate between training and evaluation, please consider defining configs of this dataset instead. You can find how to define configs instead of splits here: https://huggingface.co/docs/hub/datasets-data-files-configuration