Search: “inference”

Start owning your own intelligence today with Kimi K3 on Fireworks. Serve it, fine-tune it, and put it into production. What are you waiting…

applicationsfireworks-ai--x

**Summary:** Kimi K3 is a publicly available 3‑trillion‑parameter language model now live on Fireworks, offering 1 million-token context windows with native vision and reasoning that rivals top closed

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Teaching LLMs to Self-Evolve: Cultivating Core Meta-Skills with Reinforcement Learning

arXiv:2607.21971v1 Announce Type: cross Abstract: Test-time scaling through iterative self-evolution with environment feedback, as demonstrated by AlphaEvolve, shows remarkable performance gains. We h

researcharxiv-cs-cl

TRaM-VSR: Importance-Aware Token Routing and Merging for One-Step Diffusion Video Super-Resolution

arXiv:2607.22231v1 Announce Type: new Abstract: Video super-resolution (VSR) using large-scale Diffusion Transformer (DiT) priors achieves exceptional perceptual quality but is often impractical due t

local-aiarxiv-cs-cv

We've open-sourced MoonEP, our high-performance communication library for distributed MoE workloads. Built to make expert-parallel communica…

model-releaseskimi-moonshot--x

We've open-sourced MoonEP, our high-performance communication library for distributed MoE workloads. Built to make expert-parallel communication more efficient at scale, MoonEP helps reduce communicat

16 bit better than lower quants for Qwen3.6-27B

I am writing a fairly complex C++ windows MFC application. I have a few 3090s and can run F16 Qwen3.6-27B with 256K context and MTP. The quality of code is exceptional with this quant vs its lower qua

26 Jul 2026

I built an open-source Ollama canvas where the wires are the actual context

Most graph-based LLM interfaces use a canvas as a visual layer over what is still a linear chat. I wanted the graph itself to determine what Ollama receives. ThoughtDAG has one rule: wires are the con

local-air-ollama

26 Jul 2026

5700, 48GB RAM and a 3090 24Gb. Best OS and framework/model?

Hi everyone, I have a system which I have been using for gaming, R7 5700X, 48GB DDR4, RTX3090 24GB. But I want to use it for Local AI to reduce my reliance on cloud AI providers (mainly usage limits -

Hardware

AMD moves beyond challenger status in the race for AI platform leadership

AI hardware competition entered a sharper phase this week as Advanced Micro Devices Inc. used its flagship AI event to argue it isn’t merely chasing Nvidia Corp. — it intends to lead the market outrig

hardwaresiliconangle

Industry

everyone is talking about open-weight models. at @huggingface we're invested in open source AI. like, a lot. and more every year. i pulled t…

everyone is talking about open-weight models. at @huggingface we're invested in open source AI. like, a lot. and more every year. i pulled the public github numbers for our orgs. good excuse to introd

industryclem-delangue--x

Industry

Great to see this out, proud of the team!!! @AMD Instella 16B MoE foundation model, fully open with checkpoint after each stage from pretrai…

Great to see this out, proud of the team!!! @AMD Instella 16B MoE foundation model, fully open with checkpoint after each stage from pretraining, long-context, SFT, DPO and RL. Dataset detail, trainin

industryemad-mostaque--x

MI50 power curve tests

tests done power limiting the GPU on LACT - real power usage varies wildy at 20W it ranges from 25W to 56W same behavior happens on every setting prompt for the test runs: https://github.com/lukesdevl

Mobile Offline LLMs: What do you use them for?

I've spent the last year or so playing around with open source MLX and GGUF models on iPhone hardware. Given the limitations in memory, GPU/CPU/ANE, and in turn the context window I've been trying to

A Diffusion-Model Subpopulation Digital Twin for Mobile Health Deployment: A Case Study on the HeartSteps Intervention

arXiv:2607.21403v1 Announce Type: new Abstract: Mobile-health interventions increasingly use online learning and decision making algorithms to personalize when to nudge users toward healthier behavior

applicationsarxiv-cs-lg

A Graph Neural Network approach to zero-shot Digital Twins

arXiv:2607.20535v1 Announce Type: cross Abstract: Traditional Predictive Digital Twins often remain geometrically rigid, requiring extensive retraining or fine-tuning whenever the underlying physical

applicationsarxiv-cs-ai

A Sovereign, Open-Source Foundation Model for German and English

arXiv:2607.09424v3 Announce Type: replace-cross Abstract: We present Soofi S 30B-A3B, a sovereign, open-source Mixture-of-Experts (MoE) hybrid Mamba Transformer foundation model for German and English

AI-Driven Multi-Hop Relay Selection for Smart Urban NR-V2X Networks via Learning-to-Optimize Graph Neural Networks

arXiv:2607.20554v1 Announce Type: new Abstract: Reliable and low-latency NR-V2X communications are essential for smart mobility in dense urban environments. However, limited Road-Side Unit (RSU) densi

Hardware

AMD takes on Nvidia, US takes on Chinese AI models and AI spending still spooks investors

Maybe Nvidia ultimately won’t win the entire AI enchilada, even as it reminded everyone this week that it offers everything you need to build AI factories. This week, Advanced Micro Devices also annou

hardwaresiliconangle

AMD targets outcome-driven enterprise AI infrastructure as token economics reshape deployments

Enterprise AI is entering a new phase as organizations shift their focus from experimentation to production deployments that deliver measurable business outcomes. That transition is bringing AI token

applicationssiliconangle

AsymVerify at SemEval-2026 Task 6: Asymmetric Confidence-Gated Verification for Political Evasion Detection

arXiv:2607.20439v1 Announce Type: new Abstract: Political evasion is difficult to detect because evasive answers often appear cooperative while avoiding concrete commitment. We present AsymVerify, a c

researcharxiv-cs-cl

Attention Degradation, Function Token Anchoring, and the Limits of Attention-Based Intervention in Large Language Models

arXiv:2607.20524v1 Announce Type: new Abstract: Mean cross-positional attention degradation is widely reported in transformer interpretability, yet whether it causally limits contextual retrieval rema

Automated Synthesis and Adversarial Validation of Executable Causal Research Pipelines

model-releasesarxiv-cs-lg

arXiv:2607.21173v1 Announce Type: new Abstract: While automated research systems promise to accelerate empirical analysis, they are prone to silent failures: instances in which analysis code executes

Benchmarking the Personalization Capabilities of Large Language Models

arXiv:2607.20471v1 Announce Type: new Abstract: Personalization, the act of varying a message to induce action from a specific receiver while keeping sender, channel, and time fixed, has a long tradit

Best-of-Evidence: Best-of-N Selection under Partial Verification

arXiv:2607.20950v1 Announce Type: new Abstract: BoN improves model outputs by sampling several candidates and selecting one with a proxy score, but it assumes that complete candidates can be evaluated

researcharxiv-cs-lg

Tutorials

C-PTQ: Fisher-weighted Channel-wise Sensitivity for Post-training Quantization of MLLMs

arXiv:2607.21076v1 Announce Type: new Abstract: Multimodal large language models (MLLMs) require huge memory and computational costs, which limits their practical deployment. Post-training quantizatio

tutorialsarxiv-cs-cv

Agents

Causal-AgentIR: Self-Evolving Causal Memory for Adaptive Image Restoration Agents

arXiv:2607.21125v1 Announce Type: new Abstract: Image restoration agents have recently emerged as a flexible paradigm for handling diverse and unpredictable degradations in real-world scenarios. Exist

agentsarxiv-cs-cv

Code Monitor Red Teaming for Public-Test-Passing Code

arXiv:2607.20852v1 Announce Type: new Abstract: Visible tests are a common gate for LLM-generated code, but passing them does not certify specification correctness. We study a deployment-like monitori

CT-Merging: Consensus Directions and Task-Level Scaling for LoRA Adapter Merging

model-releasesarxiv-cs-cv

arXiv:2607.20561v1 Announce Type: cross Abstract: LoRA adapters provide an efficient way to specialize a pretrained model for many downstream tasks, but deploying one adapter per task requires adapter

Cycle-Consistent and Uncertainty-Aware Neural Surrogates for Tokamak Edge Plasmas

arXiv:2607.21407v1 Announce Type: cross Abstract: The boundary and divertor plasma govern how a tokamak exhausts power and particles, setting heat fluxes, target conditions, and the onset of detachmen

Detectors Learn the Wrong Thing: Shortcut-Resistant Adversarial Training Against Physically Realizable Attacks

arXiv:2607.21243v1 Announce Type: new Abstract: AI-enabled visual perception systems are increasingly deployed in intelligent transportation infrastructure and autonomous vehicle related applications.

safetyarxiv-cs-cv

Differentiable Logic Programming to Mitigate Reasoning Shortcuts in Neurosymbolic Systems

arXiv:2607.21185v1 Announce Type: new Abstract: Neurosymbolic (NeSy) systems integrate neural networks with logical reasoning to achieve both generalization and interpretability, but recent work has s

Distributed Model-Based Diffusion For Scalable Multi-Robot Trajectory Optimization

arXiv:2607.20992v1 Announce Type: new Abstract: Trajectory optimization for multi-robot systems remains a critical challenge, particularly when navigating highly non-convex, non-linear, and non-differ

researcharxiv-cs-ro

DONDO: Open w2v-BERT Speech-Recognition Base Models for African Languages

model-releasesarxiv-cs-cl

arXiv:2607.21540v1 Announce Type: new Abstract: We present DONDO, a family of open, permissively licensed automatic speech recognition (ASR) base models for African languages, built on the w2v-BERT 2.

DS@GT ARC at ImageCLEFmed GANs 2026: Geometric Filtering for Privacy-Preserving CT Slice Generation

arXiv:2607.20692v1 Announce Type: cross Abstract: We present a privacy-preserving framework for synthetic lung CT slice generation developed for the Image-CLEFmed GANs 2026 challenge. The approach com

Efficient and Interpretable Body-Based Emotion Recognition with Lightweight Temporal Convolutional Networks

arXiv:2607.20820v1 Announce Type: new Abstract: Body-based emotion recognition is important for real-time affective systems, but graph-based skeleton models can be computationally expensive. This pape

Tutorials

ElasticTTT: Prior-Preserving Test-Time Tuning for Video Editing

arXiv:2607.21529v1 Announce Type: cross Abstract: Test-Time Tuning (TTT) on pretrained diffusion models has emerged as a powerful paradigm for video editing. However, there exists a foundational misma

tutorialsarxiv-cs-ai

Encode Once, Decode Never: Reusing Audio LM Internals for Efficient Temporal Localization

arXiv:2602.10230v2 Announce Type: replace Abstract: Audio language models process input audio into rich frame-level representations, but the standard approach to temporal localization generates timest

local-aiarxiv-cs-lg

Euclid-MCP: A Model Context Protocol Server for Deterministic Logical Reasoning via Prolog

arXiv:2607.21412v1 Announce Type: new Abstract: Large Language Models (LLMs) excel at natural language understanding and generation but remain unreliable for multi-step logical reasoning, especially i

safetyarxiv-cs-ai

Explaining Weather Bulletins via ILP

arXiv:2607.21184v1 Announce Type: new Abstract: Inductive Logic Programming (ILP) originated within the Logic Programming community in the Nineties as a framework for combining symbolic learning with

applicationsarxiv-cs-ai

Factorized Spatio-Temporal Convolutions for Human Pose Estimation from Planar Lidar

model-releasesarxiv-cs-ro

arXiv:2607.21309v1 Announce Type: new Abstract: Localizing nearby humans and estimating their facing direction are key capabilities for safe navigation and socially aware human-robot interaction. Many

FELT: Generating Tactile Signals from Vision for Visuo-Tactile Manipulation

arXiv:2607.20683v1 Announce Type: new Abstract: The sense of touch is central to manipulation, especially when vision is occluded or ambiguous. Although combining vision and touch improves manipulatio

safetyarxiv-cs-ro

Fizgig Krea 2 training features update

https://github.com/shootthesound/Fizgig Intelligent trainer - Per-image loss tracking with self-adapting training runs — every image gets its own verdict (easy / suspect / stuck / exhausted) and its o

safetyr-stablediffusion

Flash EQ-Linear: Accelerating Equivariant Linear Layers via Group-wise Discrete Fourier Transform

model-releasesarxiv-cs-cv

arXiv:2607.21271v1 Announce Type: new Abstract: Equivariant networks embed geometric symmetries as structural priors through weight sharing, achieving remarkable parameter efficiency across vision tas

Agents

From Chatbot to Digital Colleague: The Paradigm Shift Toward Persistent Autonomous AI

arXiv:2606.14502v2 Announce Type: replace Abstract: Large Language Models (LLMs) are undergoing a fundamental transformation from conversational generators into integrated AI systems capable of reason

agentsarxiv-cs-ai

HAMON: Passive Optical Sequence Mixing for Long-Horizon Forecasting

arXiv:2606.17028v2 Announce Type: replace-cross Abstract: Simple linear and frequency-domain models remain surprisingly competitive in long-horizon time-series forecasting, and recent mechanistic evid

Hardware-Software Co-Design for Float16 On-Device Training on RISC-V Single-Core

arXiv:2607.21130v1 Announce Type: cross Abstract: By leveraging standard RISC-V extensions, namely Zfh (scalar float16) and Zvfh (vector float16), this work proposes an open-source framework to enable

local-aiarxiv-cs-ai

Information-Theoretically Secure Aggregation for Lightweight Federated Learning: Resilient to Dropouts and Adversaries

arXiv:2607.20890v1 Announce Type: new Abstract: On-device federated learning (FL) enables privacy-preserving and personalized model training on resource-constrained devices such as smartphones and IoT

local-aiarxiv-cs-lg

.@Kimi_Moonshot K3 lands on Together on Monday! We ran 452 DeepSWE rollouts against Claude Fable 5: near-flagship coding at ~35% of the pric…

model-releasestogether-ai--x

.@Kimi_Moonshot K3 lands on Together on Monday! We ran 452 DeepSWE rollouts against Claude Fable 5: near-flagship coding at ~35% of the price, and K3 pulls ahead at higher pass@k's. Full deep-dive: ht

Learning to Navigate Efficiently with Only 0.58M Trainable Parameters

model-releasesarxiv-cs-cv

arXiv:2607.11029v2 Announce Type: replace-cross Abstract: Recent progress in visual navigation has largely been driven by scale: end-to-end policies with hundreds of millions of parameters trained on