Dolevabudi/silent-tax-results

The Silent Tax: typo datasets, model generations and judge scores All data behind the report "The Silent Tax: How Typos Reshape Reasoning in Thinking LLMs" (Hamami, Abudi, Shtricker, Azoulay; Tel Aviv University NLP course, 2025/26). Code: https://github.com/dolev98/NLP_project Path Contents questions/<dataset>/<config>.parquet The typo datasets: each question in its clean form and with keyboard typos raw/<run>/<run>_<config>.jsonl One DeepSeek-R1-Distill-Qwen-7B… See the full description on the dataset page: https://huggingface.co/datasets/Dolevabudi/silent-tax-results.

<iframe src="https://silytics.si/embed/item/92b2d20a-af0b-499d-8eb5-313af3f8191a" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.