https://harveydam.ai/2026/03/27/why-batch-non-invariance-of-sequence-model-outputs-is-correct/https://harveydam.ai/wp-content/uploads/2026/03/fig_batch_flip_logits.pngfig_batch_flip_logitshttps://harveydam.ai/wp-content/uploads/2026/03/fig_near_tie_logits.pngfig_near_tie_logits2026-03-27T08:58:21+00:00monthlyhttps://harveydam.ai/2025/08/08/derailing-non-answers-via-logit-suppression-at-output-subspace-boundaries-in-rlhf-aligned-language-models/https://harveydam.ai/wp-content/uploads/2025/08/screenshot-2025-08-08-162558.pngScreenshot 2025-08-08 1625582025-08-08T22:29:21+00:00monthlyhttps://harveydam.ai/about/2023-12-22T12:24:29+00:00weekly0.6https://harveydam.ai/noobs/2023-12-22T12:16:57+00:00weekly0.6https://harveydam.ai/2023/12/22/where-pruned-image-classifiers-are-wrong/https://harveydam.ai/wp-content/uploads/2023/12/cie_notcie_influence_difference.pngcie_notcie_influence_difference2023-12-22T11:49:21+00:00monthlyhttps://harveydam.ai/2023/12/22/fast-differentiable-operations-on-arbitrary-dimensional-floating-point-arrays/https://harveydam.ai/wp-content/uploads/2023/12/old_vs_new.pngold_vs_new2023-12-22T11:04:43+00:00monthlyhttps://harveydam.aidaily1.02026-03-27T08:58:21+00:00