Collection size

96 articles

Page 4 of 5

Jul 23, 2025 · Artificial Intelligence

Unlock Efficient LLMs: How Alibaba’s PAI EasyDistill Powers Model Post‑Training

This article explains how Alibaba Cloud's AI platform PAI leverages the EasyDistill framework for post‑training model optimization, covering knowledge distillation concepts, data synthesis techniques, basic and advanced distillation training, the DistilQwen model family, real‑world customer cases, and step‑by‑step practical demos.

AI platformEasyDistillKnowledge Distillation

0 likes · 12 min read

Unlock Efficient LLMs: How Alibaba’s PAI EasyDistill Powers Model Post‑Training

Baobao Algorithm Notes

Dec 24, 2023 · Artificial Intelligence

Must‑Read AI Agent and LLM Research Papers for Deep Understanding

This curated reading list compiles essential papers on AI agents, task planning, hallucination mitigation, multimodal models, image/video generation, foundational LLM research, open‑source large models, fine‑tuning techniques, and performance optimization, providing a comprehensive roadmap for anyone aiming to master modern generative AI.

AI agentsMultimodal Learningperformance-optimization

0 likes · 23 min read

Must‑Read AI Agent and LLM Research Papers for Deep Understanding

Machine Learning Algorithms & Natural Language Processing

Apr 14, 2026 · Artificial Intelligence

Revisiting On-Policy Distillation (OPD): Typical Failures and a More Stable Fix

On‑Policy Distillation (OPD) is widely used for post‑training large language models, but the sampled‑token variant often becomes unstable due to token‑level reward imbalance, teacher‑student signal mismatch on student‑generated prefixes, and tokenizer mismatches; this article analyses the bias‑variance trade‑off, identifies three root failure modes, and proposes a teacher‑top‑K local‑support‑set objective with top‑p rollout and special‑token masking that yields more stable training and better performance on both math and agentic benchmarks.

Large Language ModelsOPDon‑policy distillation

0 likes · 32 min read

Revisiting On-Policy Distillation (OPD): Typical Failures and a More Stable Fix

Architects' Tech Alliance

Feb 16, 2025 · Artificial Intelligence

How DeepSeek’s Distillation Breaks Bottlenecks and Boosts Multimodal AI Performance

This article provides an in‑depth technical analysis of DeepSeek’s model distillation technology, covering its core principles, innovative data‑model fusion strategies, architecture design, training optimizations, performance benchmarks, and the remaining challenges of scaling distillation to multimodal tasks.

DeepSeekLarge Language Modelsai-optimization

0 likes · 16 min read

How DeepSeek’s Distillation Breaks Bottlenecks and Boosts Multimodal AI Performance

Machine Heart

Jul 5, 2026 · Artificial Intelligence

Tsinghua Special Award Winner Yuxian Gu Joins DeepSeek

Yuxian Gu, a 2021 Tsinghua PhD and 2025 Special Scholarship laureate, has joined DeepSeek, bringing expertise in pre‑training data selection, knowledge‑distillation for model compression, and efficient model architectures such as Jet‑Nemotron, which outperforms leading open‑source LLMs with up to 53.6× speedup on H100.

Artificial IntelligenceDeepSeekEfficient Model Architecture

0 likes · 6 min read

Tsinghua Special Award Winner Yuxian Gu Joins DeepSeek

Machine Learning Algorithms & Natural Language Processing

Aug 24, 2026 · Artificial Intelligence

When Online Distillation Goes Off‑Track: How Relay‑OPD Lets the Teacher Take Over at Critical Moments

The article analyzes the prefix‑failure problem in on‑policy distillation, introduces Relay‑OPD with a handoff trigger that lets a teacher model intervene locally, and shows through eight math‑reasoning benchmarks that this approach improves accuracy by up to 7.3% while cutting training trajectory length by more than half.

Large Language ModelsRelay‑OPDmath reasoning benchmarks

0 likes · 13 min read

When Online Distillation Goes Off‑Track: How Relay‑OPD Lets the Teacher Take Over at Critical Moments

Baobao Algorithm Notes

Nov 18, 2025 · Artificial Intelligence

How LightReasoner Lets Small Models Teach Large Models to Reason Efficiently

The LightReasoner paper from Hong Kong University shows that small language models can guide large models on critical reasoning steps, achieving up to 90% faster inference and significant accuracy gains across multiple math benchmarks.

Contrastive DecodingKL DivergenceLarge Language Models

0 likes · 9 min read

How LightReasoner Lets Small Models Teach Large Models to Reason Efficiently

Baidu Intelligent Cloud Tech Hub

Jan 27, 2026 · Artificial Intelligence

Deploying Qwen3 on Kunlun P800: Full‑Parameter DPO Training and Inference Guide

This guide walks through setting up a Kunlun P800 XPU host, preparing Docker containers, deploying Qwen3‑8B/‑32B/‑VL models with vLLM‑Kunlun, benchmarking performance, and running full‑parameter DPO training using LLaMA‑Factory, providing scripts, configuration files, and troubleshooting tips for AI engineers.

DPOKunlun P800LLaMA-Factory

0 likes · 32 min read

Deploying Qwen3 on Kunlun P800: Full‑Parameter DPO Training and Inference Guide

Baobao Algorithm Notes

Mar 28, 2025 · Artificial Intelligence

Can Small 7B Models Beat the State‑of‑the‑Art? A Critical Analysis of R1‑Zero Training and Unbiased GRPO

This article critically examines R1‑Zero‑style training by analyzing foundation models and reinforcement learning, uncovering pre‑training and optimization biases, proposing an unbiased Dr. GRPO method, and demonstrating a minimalist 7B‑model recipe that achieves new state‑of‑the‑art performance on AIME 2024.

GRPOLLM evaluationR1-Zero

0 likes · 20 min read

Can Small 7B Models Beat the State‑of‑the‑Art? A Critical Analysis of R1‑Zero Training and Unbiased GRPO

DataFunTalk

Mar 16, 2022 · Artificial Intelligence

Parameter-Efficient Sparsity Training for the PLUG Large-Scale Language Model

This article presents the PLUG 270‑billion‑parameter Chinese language model and introduces a parameter‑efficient sparsity training (PST) framework that combines unstructured and structured pruning with low‑rank decomposition to dramatically reduce model size while preserving downstream performance.

Deep LearningLarge Language ModelsPLUG

0 likes · 13 min read

Parameter-Efficient Sparsity Training for the PLUG Large-Scale Language Model

Smart Workplace Lab

Jun 13, 2026 · Artificial Intelligence

Why Longer Prompts Slow Down LLMs and How a Three‑Step Prompt Decay Audit Restores Performance

The article explains how overly long prompts dilute a large‑model’s attention, causing slower responses and contradictory outputs, and introduces a three‑step prompt‑decay audit—density measurement, slimming, and versioned output—that cuts response time from 1.8 s to 0.6 s, triples logical density, and reduces hallucinations by 60 %.

LLMPrompt OptimizationToken Density

0 likes · 6 min read

Why Longer Prompts Slow Down LLMs and How a Three‑Step Prompt Decay Audit Restores Performance

SuanNi

May 16, 2026 · Artificial Intelligence

Can a 4B Small Model Replace Top‑Tier Closed‑Source LLMs? Microsoft’s Terminus‑4B Cuts Token Use by 30%

Microsoft’s research shows that a 4‑billion‑parameter small model, Terminus‑4B, can act as an execution sub‑agent for terminal tasks, trimming token consumption by about 30% while preserving performance on demanding SWE‑Bench benchmarks, demonstrating a practical alternative to costly large models.

AI programmingRL trainingSWE-bench

0 likes · 7 min read

Can a 4B Small Model Replace Top‑Tier Closed‑Source LLMs? Microsoft’s Terminus‑4B Cuts Token Use by 30%

Machine Heart

May 12, 2026 · Artificial Intelligence

DECS Cuts Overthinking in Models: Halve Inference Tokens and Raise Accuracy

DECS, a novel training framework introduced by researchers from Fudan, Shanghai Jiao Tong, and the Shanghai AI Lab, theoretically exposes the flaws of length‑penalty rewards and, through token‑level reward decoupling and dynamic batch scheduling, reduces inference token counts by over 50% while improving accuracy across multiple benchmarks.

Benchmark EvaluationDECSLarge Language Models

0 likes · 9 min read

DECS Cuts Overthinking in Models: Halve Inference Tokens and Raise Accuracy

MaGe Linux Operations

Aug 19, 2026 · Artificial Intelligence

What Metrics Should You Validate When Using Quantized Models to Reduce VRAM?

This guide explains how to evaluate quantized large language models by measuring VRAM reduction, inference performance (latency, throughput), and model quality (perplexity, accuracy, BLEU/ROUGE), providing step‑by‑step procedures, code examples, and best‑practice recommendations for deployment.

GPULLMModel Compression

0 likes · 36 min read

What Metrics Should You Validate When Using Quantized Models to Reduce VRAM?

Alimama Tech

Oct 29, 2025 · Artificial Intelligence

LLM Breakthroughs at EMNLP 2025: Embedding Compression, Complex Instructions, Knowledge Scaling

EMNLP 2025 in Suzhou showcases Taobao's booth featuring four cutting‑edge AI papers that introduce a novel embedding compression framework, an automatic iterative refinement method for complex instruction generation, a knowledge infusion scaling law for large language models, and a video caption optimization approach for text‑to‑video generation.

Large Language Modelsembedding compressioninstruction generation

0 likes · 7 min read

LLM Breakthroughs at EMNLP 2025: Embedding Compression, Complex Instructions, Knowledge Scaling

Ops Community

Jan 18, 2026 · Artificial Intelligence

How to Quadruple LLM Throughput with vLLM’s PagedAttention and Continuous Batching

This guide details how to replace native Transformers inference with the high‑performance vLLM engine, leveraging PagedAttention, continuous batching, tensor parallelism, and OpenAI‑compatible APIs to achieve 3‑4× higher throughput, lower latency, and scalable multi‑GPU deployments for production‑grade large language models.

GPU OptimizationOpenAI API CompatibilityPagedAttention

0 likes · 61 min read

How to Quadruple LLM Throughput with vLLM’s PagedAttention and Continuous Batching

SuanNi

Apr 21, 2026 · Artificial Intelligence

How Qwen3.6‑35B‑A3B Matches Dense Models with Only 30 B Active Parameters

The article analyzes Qwen3.6‑35B‑A3B’s MoE architecture, showing how its 30 B active parameters outperform larger dense models across programming, agent, and multimodal benchmarks, and examines the flagship Qwen3.6‑Max‑Preview’s substantial gains in world knowledge, instruction following, and third‑party rankings.

AI evaluationLarge Language ModelMixture of Experts

0 likes · 5 min read

How Qwen3.6‑35B‑A3B Matches Dense Models with Only 30 B Active Parameters

PaperAgent

Feb 15, 2026 · Artificial Intelligence

How MiniCPM‑SALA Merges Sparse and Linear Attention to Break Long‑Context Limits

MiniCPM‑SALA introduces a hybrid sparse‑linear attention architecture that reduces quadratic compute and memory costs, achieves state‑of‑the‑art performance on long‑context benchmarks, and delivers up to 3.5× faster inference than full‑attention models on sequences up to 1 million tokens.

LLMLong ContextSparse attention

0 likes · 17 min read

How MiniCPM‑SALA Merges Sparse and Linear Attention to Break Long‑Context Limits

DataFunSummit

Dec 21, 2021 · Artificial Intelligence

Large‑Scale Pretrained Model Compression and Distillation: AdaBERT, L2A, and Meta‑KD

This talk presents Alibaba DAMO Academy’s recent work on compressing large pretrained language models, covering task‑adaptive AdaBERT, data‑augmented L2A, and meta‑knowledge distillation Meta‑KD, describing their motivations, architectures, NAS‑based search, loss designs, and experimental results across multiple NLP tasks.

Knowledge DistillationModel CompressionNLP

0 likes · 13 min read

Large‑Scale Pretrained Model Compression and Distillation: AdaBERT, L2A, and Meta‑KD

Alibaba Cloud Big Data AI Platform

Nov 6, 2024 · Artificial Intelligence

Unlocking Long-Text Video Understanding and LLM Distillation with Alibaba PAI

Alibaba Cloud’s AI platform PAI recently saw two papers accepted at EMNLP2024—VideoCLIP‑XL, which enhances video‑text representation for long descriptions using a large video‑long‑description dataset and novel pre‑training tasks, and TAPIR, a curriculum‑planning framework that distills instruction‑following abilities of large language models—while also releasing associated models, datasets, and integration tools for users.

EMNLP2024distillationlarge-language-models

0 likes · 8 min read

Unlocking Long-Text Video Understanding and LLM Distillation with Alibaba PAI