Llm
7 articles
Software Mods Reportedly Double AI Compute on RTX 20, 30, and 40 Series GPUs — Here's What the Numbers Actually Show
Community-developed software mods are reportedly unlocking up to twice the tensor throughput on NVIDIA RTX 20, 30, and 40 series GPUs for AI workloads. These mods target software limits on FP16/BF16 operations, showing real-world gains in LLM prompt processing and image generation, but do not affect gaming or token generation.
Big Tech's Custom AI Chip Race Challenges Nvidia's Dominance
OpenAI's new Jalapeño chip recently outperformed Nvidia Corp.
How Reinforcement Learning From Human Feedback (RLHF) Shapes AI
Large language models like ChatGPT owe their remarkable ability to understand and respond to human prompts directly to Reinforcement Learning from Human Feedback (RLHF).
Companies Build Powerful LLM Knowledge Bases for Enterprise AI
In India, Sarvam's LLM models now process 10 million API calls daily, a threefold increase in just three months, confirming rapid real-world adoption of advanced AI capabilities, according to Communic
Vision LLMs unlock chart and diagram understanding for RAG systems
In a recent benchmark, Vision LLMs correctly answered complex questions about historical voter turnout from a multi-decade Indian election chart, a task impossible for traditional text-based AI, accor
Researchers benchmark instruction-tuned LLMs using FP8, GPTQ, and SmoothQuant
FP8 quantization slashes the memory footprint of 70B-class open-weight models, maintaining accuracy within 0.
NVIDIA unveils NVFP4 4-bit pretraining methodology for efficient AI
NVIDIA has successfully pretrained a 12-billion-parameter Mamba-Transformer model on 10 trillion tokens using a new 4-bit precision method.