Large Language Model

Agentic AI, AI Infrastructure, AI Paper Summary, AI Shorts, Applications, Artificial Intelligence, Editors Pick, Language Model, Large Language Model, Machine Learning, New Releases, Open Source, software-engineering, Staff, Tech News, Technology

NVIDIA AI Releases Nemotron-Labs-Diffusion: A Tri-Mode Language Model with 6× Tokens Per Forward Over Qwen3-8B

/

NVIDIA researchers have released Nemotron-Labs-Diffusion, a language model family that unifies three decoding modes in one architecture. The model supports autoregressive (AR) decoding, diffusion-based parallel decoding, and self-speculation decoding. It is available in 3B, 8B, and 14B parameter sizes. The family includes base, instruct, and vision-language variants. Sequential Decoding Limits Throughput Standard autoregressive (AR) language […]

The post NVIDIA AI Releases Nemotron-Labs-Diffusion: A Tri-Mode Language Model with 6× Tokens Per Forward Over Qwen3-8B appeared first on MarkTechPost.

Agentic AI, AI Shorts, Applications, Artificial Intelligence, Editors Pick, Language Model, Large Language Model, Machine Learning, New Releases, software-engineering, Staff, Tech News, Technology

Alibaba Qwen Team Introduces Qwen3.5-LiveTranslate-Flash: Real-Time Multimodal Interpretation Across 60 Languages at 2.8-Second Latency

/

Alibaba’s Qwen team has released Qwen3.5-LiveTranslate-Flash, a real-time multimodal translation model that processes audio and video simultaneously. The model covers 60 input languages and produces speech output in 29 languages at 2.8 seconds of latency. Key additions over the previous Qwen3 version include real-time speaker voice cloning, vision-enhanced comprehension via lip movements and on-screen text, and dynamic keyword configuration for domain-specific terminology. On FLEURS and CoVoST2 benchmarks, the model outperforms major commercial alternatives. It is available as an API-only model through Alibaba Cloud Model Studio using a WebSocket-based protocol.

The post Alibaba Qwen Team Introduces Qwen3.5-LiveTranslate-Flash: Real-Time Multimodal Interpretation Across 60 Languages at 2.8-Second Latency appeared first on MarkTechPost.

AI Infrastructure, AI Paper Summary, AI Shorts, Applications, Artificial Intelligence, Editors Pick, Embedding Model, Language Model, Large Language Model, Machine Learning, New Releases, Staff, Tech News, Technology

Meet MemPrivacy: An Edge-Cloud Framework that Uses Local Reversible Pseudonymization to Protect User Data Without Breaking Memory Utility

/

As LLM-powered agents move from research to production, one design tension is becoming harder to ignore: the more useful cloud-hosted memory becomes, the more private user data it exposes. Researchers from MemTensor (Shanghai), HONOR Device and Tongji University have introduced MemPrivacy, a framework that attempts to resolve this tension without sacrificing the utility that makes […]

The post Meet MemPrivacy: An Edge-Cloud Framework that Uses Local Reversible Pseudonymization to Protect User Data Without Breaking Memory Utility appeared first on MarkTechPost.

AI Shorts, Artificial Intelligence, Editors Pick, Large Language Model, Machine Learning, Staff, Technology, Tutorials

Stochastic Gradient Descent (SGD’s) Frequency Bias and How Adam Fixes It 

/

Modern language models are trained on data with extremely uneven token distributions. A small number of words appear in almost every sentence, while many rare but meaningful tokens occur only occasionally. This creates a hidden optimization challenge: parameters associated with common tokens receive constant gradient updates, while parameters tied to rare tokens may go hundreds […]

The post Stochastic Gradient Descent (SGD’s) Frequency Bias and How Adam Fixes It  appeared first on MarkTechPost.

AI Paper Summary, AI Shorts, Applications, Artificial Intelligence, Editors Pick, Language Model, Large Language Model, Machine Learning, Staff, Tech News, Technology

NVIDIA Introduces a 4-Bit Pretraining Methodology Using NVFP4, Validated on a 12B Hybrid Mamba-Transformer at 10T Token Horizon

/

NVIDIA introduces a 4-bit pretraining methodology built around the NVFP4 microscaling format — combining selective BF16 layers, 16×16 Random Hadamard Transforms on Wgrad inputs, 2D weight scaling, and stochastic rounding on gradients — validated on a 12B hybrid Mamba-Transformer trained on 10 trillion tokens, the longest publicly documented 4-bit pretraining run, with downstream accuracy closely tracking the FP8 baseline (62.58% vs 62.62% on MMLU-Pro).

The post NVIDIA Introduces a 4-Bit Pretraining Methodology Using NVFP4, Validated on a 12B Hybrid Mamba-Transformer at 10T Token Horizon appeared first on MarkTechPost.

Artificial Intelligence, Editors Pick, Language Model, Large Language Model, Machine Learning, software-engineering, Staff, Tech News, Technology, Tutorials

A Coding Implementation to Compress and Benchmark Instruction-Tuned LLMs with FP8, GPTQ, and SmoothQuant Quantization using llmcompressor

/

In this tutorial, we explore how to apply post-training quantization to an instruction-tuned language model using llmcompressor. We start with an FP16 baseline and then compare multiple compression strategies, including FP8 dynamic quantization, GPTQ W4A16, and SmoothQuant with GPTQ W8A8. Along the way, we benchmark each model variant for disk size, generation latency, throughput, perplexity, […]

The post A Coding Implementation to Compress and Benchmark Instruction-Tuned LLMs with FP8, GPTQ, and SmoothQuant Quantization using llmcompressor appeared first on MarkTechPost.

AI Infrastructure, AI Paper Summary, AI Shorts, Applications, Artificial Intelligence, context-engineering, deep-learning, Editors Pick, Language Model, Large Language Model, Machine Learning, software-engineering, Staff, Tech News, Technology

Nous Research Proposes Lighthouse Attention: A Training-Only Selection-Based Hierarchical Attention That Delivers 1.4–1.7× Pretraining Speedup at Long Context

/

Nous Research has published Lighthouse Attention, a selection-based hierarchical attention mechanism that wraps around standard scaled dot-product attention during pretraining and is removed afterward. Unlike prior methods such as NSA and HISA that pool only keys and values, Lighthouse pools Q, K, and V symmetrically across a multi-resolution pyramid, reducing the attention call from O(N·S·d) to O(S²·d) and running stock FlashAttention on a small dense sub-sequence. Tested on a 530M Llama-3-style model at 98K context, it achieves a 1.40–1.69× end-to-end wall-clock speedup against a cuDNN SDPA baseline with matching or lower final training loss.

The post Nous Research Proposes Lighthouse Attention: A Training-Only Selection-Based Hierarchical Attention That Delivers 1.4–1.7× Pretraining Speedup at Long Context appeared first on MarkTechPost.

AI Paper Summary, AI Shorts, Applications, Artificial Intelligence, Editors Pick, Language Model, Large Language Model, Machine Learning, New Releases, Physical AI, Staff, Tech News, Technology, vision-language-model

NVIDIA Introduces SANA-WM: A 2.6B-Parameter Open-Source World Model That Generates Minute-Scale 720p Video on a Single GPU

/

Researchers from NVIDIA introduce SANA-WM, an open-source camera-controlled world model that generates 60-second, 720p videos with precise 6-DoF camera control — trained on 64 H100 GPUs and deployable on a single RTX 5090.

The post NVIDIA Introduces SANA-WM: A 2.6B-Parameter Open-Source World Model That Generates Minute-Scale 720p Video on a Single GPU appeared first on MarkTechPost.

AI Infrastructure, AI Shorts, Applications, Artificial Intelligence, Editors Pick, Language Model, Large Language Model, Machine Learning, software-engineering, Staff, Tech News, Technology

Zyphra Releases ZAYA1-8B-Diffusion-Preview: The First MoE Diffusion Model Converted From an Autoregressive LLM With Up to 7.7x Speedup

/

Zyphra’s latest release shows that an autoregressive MoE model can be converted into a discrete diffusion model with no systematic loss in evaluation performance. ZAYA1-8B-Diffusion-Preview achieves up to 7.7x inference speedup over autoregression by shifting decoding from memory-bandwidth bound to compute-bound — a key advantage as modern GPUs continue scaling FLOPs faster than memory bandwidth.

The post Zyphra Releases ZAYA1-8B-Diffusion-Preview: The First MoE Diffusion Model Converted From an Autoregressive LLM With Up to 7.7x Speedup appeared first on MarkTechPost.

Agentic AI, AI Agents, AI Shorts, Applications, Artificial Intelligence, Editors Pick, For Devs, generative-ai, Large Language Model, New Releases, Open Source, software-engineering, Staff, Tech News, Technology

Cline Releases Cline SDK: An Open-Source Agent Runtime Now Powering Its CLI and Kanban, With IDE Extensions Being Migrated

/

Cline has extracted its internal agent harness into an open-source TypeScript SDK called @cline/sdk, the same runtime now powering its CLI and Kanban, with VS Code and JetBrains extensions being migrated. The SDK is structured as a four-layer stack — @cline/shared, @cline/llms, @cline/agents, and @cline/core — with native support for plugins, subagents, CRON scheduling, checkpointing, and MCP connectors. On Terminal Benchmark 2.0, Cline CLI scored 74.2% on claude-opus-4.7, compared to Anthropic’s published 69.4% for Claude Code on the same model. Install via npm install @cline/sdk. Requires Node.js 22+.

The post Cline Releases Cline SDK: An Open-Source Agent Runtime Now Powering Its CLI and Kanban, With IDE Extensions Being Migrated appeared first on MarkTechPost.

Scroll to Top