We have 10 of 29 papers
We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.
Not the right person? Other researchers publish under this name.
Conditional memory layers, such as DeepSeek's Engram, attach large hashed n-gram tables to a language model, and removing them from DeepSeek-V4.1-Flash costs 36.8 PopQA points. Low-bit quantization loses long-tail facts first, and the standard remedy, giving more bits to frequently accessed rows, backfires: refining lo...
Low-bit quantization reduces the memory footprint of large language models, but preserving accuracy with efficient training remains challenging. Scalar quantization encodes weights independently, while explicit-codebook vector quantization introduces costly search during training. We propose E8-QAT, a 2-bit quantizatio...
Agentic applications generate multi-turn sequences of LLM requests, with each new turn extending the context accumulated from prior interactions and tool calls. Prefix caching reuses the KV cache of earlier turns and avoids redundant prefill, but routing every turn to its cache-affine instance can create severe local q...
We present Hunyuan-A13B, an open-source large language model based on a Mixture-of-Experts architecture. It contains 80 billion total parameters but activates only 13 billion during inference, balancing model capability, computational efficiency, and deployment cost. The model is pretrained on a rigorously filtered 20T...
D-Quant is proposed, a flexible KV cache quantization framework that introduces a Drift mechanism to convert entropy-coded representations of each token into fixed-size bitstreams, enabling regular memory access and parallel dequantization within attention kernels.
HPC-Ops Top-K is presented, a sample-guided exact selector for ragged sparse-attention score rows that outperforms the fastest verified external exact baseline and outperforms the fastest verified external exact baseline on indexer scores from Hy4-Preview.
PIVOT, Proxy Indexing Via One full-prefix Traversal, a training-free, drop-in replacement for the DSA indexer that shares one prefix scan across a group of nearby queries, matches the accuracy of the dense DSA indexer while accelerating it by up to 4x and reducing end-to-end latency by up to 1.6x at long context.
FOCUS is proposed, a post-training quantization framework with end-to-end scale learning for FP4 Optimization via Coupled-Relaxation and Dual-Granularity Scaling, which relaxes the tight coupling between quantization and dequantization scales with a learnable full-precision coefficient, enabling more effective optimiza...
CoSA is proposed, a two-stage training-free Sparse Attention under proxy-kernel CO-design, which couples a Kernel-Aware Proxy (KAP) with an Ordered-Skipping Kernel (OSK) and achieves a 4.93% attention speedup and reduces end-to-end Time-to-First-Token by 2.53% with negligible performance degradation.
DFly is proposed, a block-diffusion framework combining a hybrid target-conditioning backbone with a predecessor-conditioned autoregressive head, improving target-feature utilization and intra-block dependency modeling while keeping generation parallel, and DFly treats verification as a shared batch-level resource.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
