Sign inGet started
Dataset · Hugging Face

amildravid4292/mathnet-correctness-contrast-loras

Correctness-contrast LoRAs (Qwen3-4B-Base, MathNet) 908 rank-1 LoRA adapters that isolate a causal correctness direction in weight space, plus that direction. Each adapter is fit on either the correct or the incorrect solutions to one math problem; their difference cancels the problem and leaves correctness. Result. Steering a model's weights along the direction in directions/d_bar_pc1_P227.npy causally changes accuracy on held-out problems: amplifying it raises the solve rate… See the full description on the dataset page: https://huggingface.co/datasets/amildravid4292/mathnet-correctness-cont

Author / organizationamildravid4292
License (as declared)mit
Downloads167 (as reported by Hugging Face)
Likes0
Created17 July 2026
Last updated at source17 July 2026
lorainterpretabilityweight-spacemathematical-reasoning
Embed this entry
<iframe src="https://silytics.si/embed/item/bcdde5fe-765f-4873-acaf-b6e8bf740b0c" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Free, no tracking. More about feeds and embeds.

Popularity, last 5 days

167 now (+39 since 2026-10-17). Counts as reported by the source, snapshotted daily by Silytics.

More from amildravid4292

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 21 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.