A developer fine-tuned Qwen2.5-7B on a 16GB T4 GPU using QLoRA, quantizing the frozen base model to 4-bit NF4 to reduce memory footprint from 15GB to 5.44GB. The technique enables training large langu…
A developer fully fine-tuned a 270M-parameter Gemma 3 model on a laptop using the Banking77 dataset, achieving ~96% accuracy on intent classification. The project used full fine-tuning with loss maski…
Google researchers published a paper introducing S-CTS, a two-stage machine learning system that detects and terminates coordinated networks of AI-generated spam on video platforms by identifying clus…
Google researchers published a paper detailing a new system, the Scalable Cluster Termination System (S-CTS), designed to detect coordinated generative AI spam by analyzing organizational attack struc…
A critical bug in Google's Gemma 4 causes it to malform tool calls under real load, affecting vLLM, llama.cpp, Ollama, and oobabooga. A developer open-sourced a diagnosis, repair, and experimental LoR…
A developer is seeking advice on which model excels at selection-with-abstention tasks, where the model must either pick a candidate with a confidence tier or abstain from outputting anything. The use…
Hugging Face's PEFT library challenges the dominance of LoRA in parameter-efficient fine-tuning, revealing that while LoRA accounts for over 98% of PEFT usage on the Hugging Face Hub, alternative meth…
A developer from Gate of AI published a tutorial on fine-tuning large language models for specialized communication tasks, using homeless shelters as a case study. The tutorial demonstrates how to str…
A developer built AETHER, a fully offline, voice-controlled AI assistant that runs three local language models on a laptop to control a PC, send WhatsApp messages, and perform desktop automation witho…
A comprehensive review published on arXiv examines fine-tuning techniques for Large Language Models (LLMs), covering methodologies from supervised and unsupervised learning to parameter-efficient meth…
A developer fine-tuning Llama 3.2 3B on medical QA found that lower evaluation loss (2.275 vs 2.495) corresponded to worse model performance, with generation breakdowns and incorrect medical advice. T…
A developer argues that many enterprise AI teams waste money fine-tuning models when retrieval-augmented generation (RAG) would be more effective. Fine-tuning a 40B model can cost over $35,000 per run…
Researchers from Aalto University and the University of Waterloo introduced Locket, a feature-locking technique for large language models that enables pay-to-unlock schemes by restricting specific mod…
A developer fine-tuned LLMs on advice with cautious or eager attitudes about everyday topics, then found the models' opinions shifted on unrelated held-out topics like e-bike trail access. The cautiou…
NVIDIA released BioNeMo Recipes, a set of training recipes that use Low-Rank Adaptation (LoRA) to fine-tune large biological foundation models like ESM2-3B and Evo2-1B on a single workstation GPU. The…
A developer released Avatar Video Workbench, an MIT-licensed tool that sanitizes Vertex AI and LTX workflow reports before public sharing, checking for cloud URIs, project IDs, and other sensitive dat…
A new dataset, fineset-io/efficient-llm-papers, compiles 1,734 records of arXiv and Semantic Scholar papers on efficient LLM techniques like quantization, LoRA, MoE, and FlashAttention, each quality-s…
Researchers introduced i2L (image-to-LoRA), a framework that compresses style LoRA training into a single forward pass, enabling immediate style instantiation from reference images without per-style o…
Researchers introduced Prompt2Effect, a hypernetwork that generates effect-specific LoRA weights for Image-to-Video diffusion models in a single forward pass, eliminating the need for per-effect train…
Nick Creighton, a developer and host of the *Build Log* podcast, has published a practical comparison of full-model fine-tuning, LoRA, and QLoRA for adapting large language models in 2024. The guide d…