Search

Large language models (LLMs) have become integral tool for users from various backgrounds. LLMs, trained on vast corpora, reflect the linguistic and cultural nuances embedded in their pre-training data. However, the values and perspectives inherent in this data can influence the behavior of LLMs, leading to potential biases. As a result, the use of LLMs in contexts involving spiritual or moral values necessitates careful consideration of these underlying biases. Our work starts with verification of our hypothesis by testing the spiritual values of popular LLMs. Experimental results show that LLMs' spiritual values are quite diverse, as opposed to the stereotype of atheists or secularists. We then investigate how different spiritual values affect LLMs in social-fairness scenarios e.g., hate speech identification). Our findings reveal that different spiritual values indeed lead to different sensitivity to different hate target groups. Furthermore, we propose to continue pre-training LLMs on spiritual texts, and empirical results demonstrate the effectiveness of this approach in mitigating spiritual bias., Comment: 9 pages including appendix; 5 figures; 5 tables; submitted to ARR - Octobor 2024

Twisted bilayers host many emergent phenomena in which the electronic excitations (quasiparticles - QPs) are closely intertwined with the local stacking order. By inspecting twisted hexagonal boron nitride (t-hBN), we show that non-local long-range interactions in large twisted systems cannot be reliably described by the local (high-symmetry) stacking and that the band gap variation (typically associated with the moir\'e excitonic potential) shows multiple minima with variable depth depending on the twist angle. We investigate twist angles of 2.45$^\circ$, 2.88$^\circ$, 3.48$^\circ$ and 5.09$^\circ$ using the GW approximation together with stochastic compression to analyze the QP state interactions. We find that band-edge QP hybridization is suppressed for intermediate angles which exhibit two distinct local minima in the moir\'e potential (at AA region and saddle point (SP)) which become degenerate for the largest system (2.45$^\circ$).

Accommodating all the weights on-chip for large-scale NNs remains a great challenge for SRAM based computing-in-memory (SRAM-CIM) with limited on-chip capacity. Previous non-volatile SRAM-CIM (nvSRAM-CIM) addresses this issue by integrating high-density single-level ReRAMs on the top of high-efficiency SRAM-CIM for weight storage to eliminate the off-chip memory access. However, previous SL-nvSRAM-CIM suffers from poor scalability for an increased number of SL-ReRAMs and limited computing efficiency. To overcome these challenges, this work proposes an ultra-high-density three-level ReRAMs-assisted computing-in-nonvolatile-SRAM (TL-nvSRAM-CIM) scheme for large NN models. The clustered n-selector-n-ReRAM (cluster-nSnRs) is employed for reliable weight-restore with eliminated DC power. Furthermore, a ternary SRAM-CIM mechanism with differential computing scheme is proposed for energy-efficient ternary MAC operations while preserving high NN accuracy. The proposed TL-nvSRAM-CIM achieves 7.8x higher storage density, compared with the state-of-art works. Moreover, TL-nvSRAM-CIM shows up to 2.9x and 1.9x enhanced energy-efficiency, respectively, compared to the baseline designs of SRAM-CIM and ReRAM-CIM, respectively.