Tagged: fine-tuning

67 articles

Latest

Adaption's New Tool Invents Training Data From Just a Task Description

Fine-Tuned CLIP Searches 12,000+ Satellite Images With Plain Text

Fine-tune XLS-R 300M on 4 hours of Common Voice for multilingual speech recognition

Fine-tune Google’s ViT on bean leaves in under 100 lines of Python

SegFormer Fine-Tuning on a Sidewalk Dataset Shows Why Custom Data Beats Generic

Liquid AI's 350M model hits 22.6% schema compliance before fine-tuning

Sentence Transformers v3 kills the fit() API — here's what replaces it

Graphcore's IPUs make ViT fine-tuning on medical images a plug-and-play affair

Dreambooth Fine-Tuning: Low Learning Rates Beat Fast Results Every Time

Fine-Tune Whisper in 8 Hours to Conquer 96+ Languages

LoRA cuts Stable Diffusion fine-tuning to 3MB files and 11GB GPUs

Hugging Face's PEFT slashes fine-tuning from 40GB to a few MB

Meta's 7B LLaMA learns Stack Exchange answers via RLHF on a single GPU

StarCoder's 8K context window turns code LLMs into chat assistants

Stable Diffusion learns to derain, denoise, and cartoonize from text prompts

Gaudi2 Fine-Tunes Vision-Language Models 2.5x Faster Than Nvidia A100

Llama 2 70B fine-tuning slashes RAM from 2TB to 32GB with FSDP trick

Hugging Face Proves You Don't Need Code to Build Your Own ChatGPT Clone

Hugging Face's new chat templates stop a silent killer of model performance

Hugging Face fine-tunes StarCoder into HugCoder, a personal coding copilot

Prodigy-HF Lets You Fine-Tune BERT on Annotated Data With One Command

Open LLMs hit a wall in 2023: data scarcity, not model size, became the bottleneck

Unsloth claims 2.7x faster LLM fine-tuning with zero accuracy loss

Inside XYZ-Aquila-SFT: How a 0.6B Qwen Model Learns Multi-Step Tool Use

SmolLM2 learns to think: 8,000-row reasoning corpus turned into LoRA fine-tune

Unsloth crushes MoE training with 7.3x speedup, Axolotl hits 1.45x on Qwen3.5

80 examples is all it takes to reshape a large AI model's values

OpenAI lets devs fine-tune GPT-3, one customer hit 95% accuracy

OpenAI admits ChatGPT bias is a 'bug, not a feature'

OpenAI lets devs fine-tune GPT-3.5 Turbo, claims it can beat GPT-4

OpenAI winds down fine-tuning API, pushes managed service

OpenAI opens GPT-4 to all paying devs, kills older models

Indeed's OpenAI bet pays off: 20M daily messages, 20% more applications

OpenAI launches GPT-4o fine-tuning, then schedules its shutdown

OpenAI Now Lets You Distill o1 Into GPT-4o Mini — All In-House

OpenAI lets you fine-tune GPT-4o with images, and yes, it's a big deal

Decagon runs 91% of a major brand's support with no humans

OpenAI drops o1 API, slashes realtime audio prices by 60%

Rogo hits 27x ARR growth using OpenAI to save bankers 10+ hours a week

OpenAI's New Pioneers Program Aims to Fix AI's Benchmark Obsession

NVIDIA's NeMo AutoModel hits 3.7x training speedup with a single line of code

98.4% of fine-tuned models use LoRA. That's not because it's best.

TRL v1.0: The 3M-Download Library Built to Survive AI's Chaotic Evolution

Hugging Face and Unsloth Are Giving Away Free GPU Hours to Fine-Tune Your Own AI

OpenAI's Codex Can Now Train Its Own Open-Source Models, But Should It?

Claude can now fine-tune open-source LLMs directly on cloud GPUs

Together AI now lets you fine-tune any Hugging Face model in 5 minutes

Fine-tune FLUX.1-dev on an RTX 4090: QLoRA slashes VRAM below 10GB

NVIDIA's GR00T N1.5 robot model now fine-tunes on a $200 open-source arm

DeepSpeed ZeRO-3 clashing with Liger GRPO loss? Shape mismatch halts Qwen2.5 finetuning

Falcon-Edge proves 1.58-bit models can match 16-bit rivals—and you can fine-tune them

olmOCR fine-tune fixes the header-footer blind spot that killed it for invoices

A tiny finetuned reranker just crushed public models 4x its size

Flux.1 beat Midjourney and DALL-E 3 in 2024 — here's the open-source shift it triggered

CFM used Llama 3.1 to label 900k news headlines, then fine-tuned a tiny model that matched its accuracy

Argilla 2.4 Lets You Build AI Datasets Directly From Hugging Face Hub, No Code Needed

Llama 3.2 Has Been Quietly Working in Keras Since Day 1

Llama 8B shrank to 1.58 bits per parameter and still beat a full-size model

Docmatix drops 9.5M QA pairs to close the DocVQA gap with closed models

A 7B Model Just Beat Everyone at Math Olympiad Problems. Here's the Recipe

Florence-2 Fine-Tuned on DocVQA Jumps from 0 to 57% Similarity

Dell and Hugging Face cut on-premise LLM deployment from weeks to minutes

Hugging Face and NVIDIA team up to kill GPU scarcity with $8.25/hr H100 training

Google's Gemma Gets a LoRA Tune-Up: Fine-Tune a 2B Model Free on Colab

PEFT Adds LoRA Merging on the Fly, No Checkpoint Download Needed

Fine-tune with Mixtral: $2.70 beats GPT-4's $3,061 price tag

DPO vs IPO vs KTO: Hugging Face's hyperparameter sweep crowns a winner