4 accepted papers
The prefill stage of long-context Retrieval-Augmented Generation (RAG) is severely bottlenecked by computational overhead. To mitigate this, recent methods assemble pre-calculated KV caches of retrieved RAG documents (by a *user query*) and reprocess selected tokens to recover cross-attention betwee…
Mixture-of-Experts (MoE) architecture with experts parallelism scales LLMs efficiently by activating only a subset of experts per input, avoiding proportional training costs. However, the intensive and heterogeneous communication substantially hinders the efficiency and scalability of MoE training i
Large Language Models (LLM) have gained significant attention. KV cache, stored to avoid quadratic complexity of attention, becomes a bottleneck due to the demands for long-context. Sparse attention (SA) has been proposed to address this by only selecting critical KVs for attention, which ma…
Top-k sparsification method is popular and powerful forreducing the communication cost in Federated Learning(FL). However, according to our experimental observation, it spends most of the total communication cost on the index of the selected parameters (i.e., their position informa-tion), which is i…