Skip to content

DEV Community

AI Explained Series' Articles

Back to pueding's Series

Cover image for Agent Leaderboards Mislead Under Distribution Shift (IBM): Predictive Validity

pueding

Jun 22

Agent Leaderboards Mislead Under Distribution Shift (IBM): Predictive Validity

#agents #ai #llm #machinelearning

7 min read

Cover image for GLM-5.2 Becomes the Top Open-Weights Model: Active vs Total Parameters

pueding

Jun 23

GLM-5.2 Becomes the Top Open-Weights Model: Active vs Total Parameters

#llm #ai #machinelearning

6 min read

Cover image for Baidu Unlimited OCR Holds the KV Cache Constant for 40+ Pages: Reference Sliding Window Attention

pueding

Jun 26

Baidu Unlimited OCR Holds the KV Cache Constant for 40+ Pages: Reference Sliding Window Attention

#ai #llm #machinelearning

8 min read

Cover image for OpenAI and Broadcom's Jalapeño, a Custom Inference ASIC: Inference ASIC vs GPU

pueding

Jun 27

OpenAI and Broadcom's Jalapeño, a Custom Inference ASIC: Inference ASIC vs GPU

#ai #llm #hardware #machinelearning

7 min read

Cover image for Qwen-AgentWorld Trains a Language Model as a World Model for RL Agents: World Model as a Decoupled RL Simulator

pueding

Jun 28

Qwen-AgentWorld Trains a Language Model as a World Model for RL Agents: World Model as a Decoupled RL Simulator

#agents #ai #llm #machinelearning

6 min read

Cover image for CacheWeaver Reorders RAG Evidence for Prefix-Cache Reuse: Prefix-Cache-Aware Evidence Reordering

pueding

Jun 29

CacheWeaver Reorders RAG Evidence for Prefix-Cache Reuse: Prefix-Cache-Aware Evidence Reordering

#rag #llm #ai #machinelearning

7 min read

Cover image for SGLang v0.5.14: LPLB Expert-Parallel Load Balancing

pueding

Jun 30

SGLang v0.5.14: LPLB Expert-Parallel Load Balancing

#ai #llm #machinelearning #devops

7 min read

Cover image for S-Agent: Spatial Tool-Use Makes an 8B Agent Rival GPT-5.4 on Spatial Reasoning: Spatio-Temporal Evidence Accumulation

pueding

Jul 1

S-Agent: Spatial Tool-Use Makes an 8B Agent Rival GPT-5.4 on Spatial Reasoning: Spatio-Temporal Evidence Accumulation

#agents #ai #llm #machinelearning

6 min read

Cover image for Cluster-Route-Escalate Cascade Serves LLMs at 97-99% Accuracy for Less Cost: Cost-Aware LLM Cascade

pueding

Jul 2

Cluster-Route-Escalate Cascade Serves LLMs at 97-99% Accuracy for Less Cost: Cost-Aware LLM Cascade

#ai #llm #agents #devops

6 min read

Cover image for Dockerless Verifies Coding-Agent Patches Without Containers: Execution-Free Patch Verification

pueding

Jul 3

Dockerless Verifies Coding-Agent Patches Without Containers: Execution-Free Patch Verification

#ai #agents #llm #devops

6 min read