Research paper library

Systems in Open Problems in Pre- And Post-Ln Architectures: An Evidence-Ledger Investigation increasingly promise longer-horizon, higher-autonomy workflows, but their outputs are difficult to trust when claims are not explicitly tied to evi…

Across 41 cached papers, during inference is repeatedly flagged as an unresolved area 1 explicit open-problem statements, 2 cross-paper numeric contradictions, 0 surprise/counter-narrative findings. A scoped evidence ledger can separate whi…

Across 55 cached papers, optimization is repeatedly flagged as an unresolved area 1 explicit open-problem statements, 2 cross-paper numeric contradictions, 1 surprise/counter-narrative findings. A scoped evidence ledger can separate which s…

Across 28 cached papers, enable richer representation is repeatedly flagged as an unresolved area 2 explicit open-problem statements, 2 cross-paper numeric contradictions, 0 surprise/counter-narrative findings. A scoped evidence ledger can…

Across 3 cached papers, would explore the scaling is repeatedly flagged as an unresolved area 6 explicit open-problem statements, 0 cross-paper numeric contradictions, 0 surprise/counter-narrative findings. A scoped evidence ledger can sepa…

Modern deep LLMs accumulate dozens of transformer layers, yet evidence repeatedly shows large fractions of layers are skippable, prunable, or contribute marginally — but the conditions, depth-vs-quality tradeoffs, and architecture-specific…

Systems in Liquid Neural Networks and Continuous-Time Models increasingly promise longer-horizon, higher-autonomy workflows, but their outputs are difficult to trust when claims are not explicitly tied to evidence. This draft synthesizes ta…

Post-training quantization GPTQ, AWQ, weight-only and weight-activation schemes is widely claimed to compress LLMs to 4 bits with near-lossless accuracy, but papers report quality and speed against incompatible models, calibration sets, and…

LLM adaptation increasingly relies on parameter-efficient fine-tuning LoRA, QLoRA, adapters, prefix/prompt tuning, but papers report accuracy-versus-memory trade-offs against incompatible baselines and benchmarks, making it hard to compare…

Systems in Autonomous Research Harnesses increasingly promise longer-horizon, higher-autonomy workflows, but their outputs are difficult to trust when claims are not explicitly tied to evidence. This draft synthesizes taxonomy-scoped eviden…

Retrieval-augmented generation RAG is widely claimed to reduce hallucination and improve factuality of LLMs, but papers report gains against incompatible retrievers, corpora, and evaluation metrics, making it hard to compare which retrieval…

RoPE has become the default positional encoding for modern LLMs, yet recent papers report that explicit rotary signals interact unpredictably with implicit positional biases from attention, normalization, and massive activations — producing…

Attention has fragmented into MHA/MQA/GQA, sparse windowed DSA/SWA/CSA, linear/state-space Mamba, Delta, KDA, and hybrid attention; claims about quality parity, throughput, and long-context behavior are scattered and not directly comparable…

FFN dominates parameter count in transformer LLMs, and MoE/DeepSeek MoE/shared-expert variants now claim large efficiency wins, but sparse vs dense comparisons mix training-cost, serving-cost, and quality claims in inconsistent ways. This d…

Transformer normalization has fragmented into LayerNorm, RMSNorm, Pre-Norm, Post-Norm, and QK-Norm variants, with overlapping stability and training-speed claims that are not directly comparable. This draft synthesizes taxonomy-scoped evide…

Modern LLMs claim long-context capability, but position encoding choices sinusoidal, RoPE, YaRN, NoPE are reported with mixed evidence and incompatible benchmarks, making it hard to compare claims about extrapolation and stability. This dra…

Recent LLM labs DeepSeek, Kimi, ByteDance report modifications to residual connections in or around the FFN, but the design space is small and the evidence is scattered across system papers and ablations. This draft synthesizes taxonomy-sco…

Systems in Medical AI and Clinical Decision Support increasingly promise longer-horizon, higher-autonomy workflows, but their outputs are difficult to trust when claims are not explicitly tied to evidence. This draft synthesizes taxonomy-sc…