论文 The Loss Does Not See the Basis, but Adam Does
论文 Adapting Without Gradients: Affine Statistics Transport and What Its Certificate Can Tell You
论文 Generalization Is Stability, Not Accuracy: Multi-Axis Evaluation of LLMs
论文 How Lossless Is Lossless Speculative Decoding? The Role of Numerical Precision in Orthrus
论文 Let Confidence Change, Not the Prediction: Prediction-Preserving Repair for Post-hoc Calibration
论文 G^2PTQ: Improving LLM Post-Training Quantization with Generalized Gradient Compensation