Llama Cpp

Llama Cpp에 대해 3개 실제 데이터 소스에서 20개 공개 신호를 찾았습니다. GitHub 10건, Hacker News 0건, arXiv 10건을 원문 링크와 함께 보여줍니다.

검색 결과 보기

토픽 정보

이름
Llama Cpp
Slug
llama-cpp
관련 키워드
and, the, for, with, llama, cpp, models, language
최근 7일 변화
MVP에서는 실시간 검색 결과 기반 점수만 계산합니다. DB snapshot 저장 후 추세가 표시됩니다.
마지막 업데이트
2026-11-21T07:24:13.516Z

출처별 최신 반응

Jailbreak Detection on Large Language Models via Contrastive Prototype Structures

Jailbreak attempts and harmful prompts remain a challenge for the safe deployment of large language models. We study whether contrastively learned representations can support safety detection without training an additional downstream classifier. We propose Jailbreak Detection with Prototype Inference (JDPI), which uses binary safety labels to construct harmful and benign prototypes and makes predictions from relative prototype distances. JDPI uses only binary safety labels for encoder adaptation, prototype constru...

arxiv · 원본 ID http://arxiv.org/abs/2611.24936v1

Chain-of-Thought Hidden States Reveal Task-Solving Directions for Direct Answers

Chain-of-Thought (CoT) prompting improves the performance of Large Language Models by eliciting intermediate reasoning steps. However, CoT is far from perfect: on some tasks, models perform worse than with direct-answer prompting, and on others they generate a few hundred tokens to reach the correct answer. We start by labelling each CoT hidden state as recovering if patching it into the direct prompt yields the correct answer, and failed otherwise. A linear probe shows that the model encodes these two classes dif...

arxiv · 원본 ID http://arxiv.org/abs/2611.24598v1

Test-Time Symmetrization: Scaling Laws for Model-Agnostic Learning under Structure

Test-time compute has emerged as a powerful way to improve the accuracy of language models at inference time, with scaling laws quantifying how accuracy improves with additional forward passes. Existing scaling laws, however, focus on repeated sampling from a fixed prompt, even though many tasks admit multiple equivalent representations of the same underlying problem. A graph does not depend on how its nodes are numbered, and a program does not depend on the names of its variables, yet language models can change t...

arxiv · 원본 ID http://arxiv.org/abs/2611.24032v1

ReflectRec: Cycle-Consistency Preference Verification for LLM-based Recommendation

CoT-style LLM recommenders generate a natural-language preference note as a chain-of-thought prefix to ranking, but the note is trusted without verification, and the ranker never consults behaviorally similar users. We propose ReflectRec, organized around a preference-to-history round-trip verifier whose recall score serves as an operational signal of preference-to-history recoverability, enabling conditional regeneration for low-recoverability preferences and selection among regeneration attempts. Around this ver...

arxiv · 원본 ID http://arxiv.org/abs/2611.23969v1

JEV vs. LLMs: A Comparative Analysis for Phishing Email Detection

Phishing remains a major source of credential theft and fraudulent payments, and email security systems must classify large message volumes reliably, quickly, and at manageable cost. Large Language Models (LLMs) offer flexible text understanding, but their token-based generation increases response time and API cost, and the reliability of their confidence values requires measurement. Jev 1.13, a typed decision model from TypeSafe AI, instead returns a structured outcome with a confidence value. This paper compares...

arxiv · 원본 ID http://arxiv.org/abs/2611.23683v1

FACT: Fixed-Budget Allocation and Cross-Step Training for Speculative Decoding

Speculative decoding accelerates language-model generation by verifying multiple draft tokens in a single target-model pass. However, better draft predictions and longer accepted continuations do not necessarily produce faster generation. To examine this gap, we present FACT (Fixed-Budget Allocation and Cross-Step Training), an experimental framework for studying how drafter training, candidate allocation, and execution conditions jointly affect acceptance and latency. Using a lightweight feature-prediction drafte...

arxiv · 원본 ID http://arxiv.org/abs/2611.23463v1

LusoClin: A Synthetic Clinical Note Dataset for Medical Information Retrieval in European Portuguese

Access to large-scale clinical text collections is crucial for advancing medical information retrieval research, including clinical note search, patient cohort identification, and domain-adapted retrieval models. However, real clinical notes are rarely available due to privacy regulations, which particularly limits progress for low-resource languages such as European Portuguese. We present LusoClin, a publicly released dataset of synthetic clinical notes in European Portuguese, created to support privacy-preservin...

arxiv · 원본 ID http://arxiv.org/abs/2611.23454v1

When Does Subspace Direction Matter for LoRA? Regime Analysis of the Magnitude Principle in Few-Shot Adaptation

We show that the importance of a LoRA adapter's initial input-subspace direction is regime-dependent, and we characterize this dependence in theory and experiment. This refines the recent view that spectral initializers help through update magnitude alone. Direction matters most when labeled data are scarce relative to the representation dimension and the activation subspace used for initialization captures supervised task structure. In this regime, the adapter's row span remains sensitive to its starting point, m...

arxiv · 원본 ID http://arxiv.org/abs/2611.23428v1

What Does a Bias Head Measure? Non-Identifiability in Head-Level Bias Attribution

Methods that localise social bias in large language models to individual attention heads and remove it by ablation are validated by their effect on a fairness metric, never by evidence that the statistic selecting the heads measures bias. We audit this on Qwen-2.5-7B-Instruct and Llama-3.1-8B-Instruct. The head map correlates at $ρ= -0.950$ and $-0.768$ with a label-blind response-length map, apparently identifying a surface confound. An intervention that breaks the length-label coupling rejects this reading: with...

arxiv · 원본 ID http://arxiv.org/abs/2611.23218v1

Get Hint-Augmented GRPO Done Right Through Activation Steering

GRPO has become the dominant reinforcement learning method for eliciting reasoning capability in LLMs using verifiable rewards. However, it suffers from a fundamental failure mode: when a training question exceeds the base model's current capability, all generated responses fail and receive zero reward, yielding no gradient signal and stalling further capability expansion. An emerging class of GRPO extensions, termed Hint-Augmented GRPO (HA-GRPO), adds a "hint" to the context to reduce the difficulty of the questi...

arxiv · 원본 ID http://arxiv.org/abs/2611.23124v1

AtomicBot-ai/atomic-agent

Atomic Agent is a local-first AI agent. Runs open-weight models on your own machine via llama.cpp.

github · 원본 ID 1217416768

Michael-A-Kuykendall/shimmy

⚡ Pure-Rust WebGPU inference engine — OpenAI-API compatible, GGUF native, runs on any GPU. No Python. No llama.cpp. Single binary.

github · 원본 ID 1046584285