Tagged: transformers

Stop GPT-2 Repetition: Fix Text Generation with Beam Search

Sep 7

Train EsperBERTo: A 84M Parameter Model in Minutes

Sep 7

T5 and BART share one architecture: How encoder-decoders actually work

Sep 6

Skip Pre-training: Cut Encoder-Decoder Costs by Reusing BERT and GPT-2

Sep 6

AWS and Hugging Face team up to slash NLP setup from days to minutes

Sep 6

Hugging Face and AWS launch one-line deployment for 10,000+ transformer models

Sep 5

Software Is Eating ML: Why You Need Engineers, Not Data Scientists

Sep 5

Hugging Face's 50,000 developers get plug-and-play access to Graphcore IPUs

Sep 5

Hugging Face Course Launch: Free Compute, Big Names, Live Demos

Sep 4

Graphcore and Hugging Face team up to slash Transformer latency

Sep 4

Google's BERT solved 11+ NLP tasks at once — here's how it works

Sep 4

Hugging Face's Wav2Vec2 can now transcribe hour-long audio without crashing

Sep 4

Sentiment analysis now takes 5 lines of Python — no ML experience needed

Sep 4

Hugging Face Adds Constrained Beam Search to Force Words in AI Text

Sep 3

Lewis Tunstall on Why Your Transformer Models Are Too Slow for Production

Sep 3

Optimum 1.2 brings ONNX Runtime to Transformers pipelines for faster BERT inference

Sep 2

Graphcore and Hugging Face drop 10 IPU-ready transformer models

Sep 1

Three Ways to Convert Hugging Face Models to ONNX, Ranked by Pain

Sep 1

Why Hugging Face thinks you're using TensorFlow all wrong

Aug 31

How Hugging Face cut Bloom inference latency 5x and boosted throughput 50x

Aug 30

CLIPSeg brings zero-shot segmentation to any object, no retraining needed

Aug 28

Hugging Face now supports 8 core vision tasks and 100k+ models

Aug 27

Microsoft's SpeechT5 packs ASR, TTS, and voice conversion into one Transformer

Aug 27

Hugging Face Slashes LLM Latency by Up to 10x With Assisted Generation

Aug 25

Apple Developers Get New Tools to Run Llama 2 On-Device

Aug 22

Transformers hit a wall, so 4 startups are chasing the next great LLM architecture

Aug 11

OpenAI's GPT-1: One Unsupervised Model to Rule Many Language Tasks

Jul 19

GPT-2 trained on pixels rivals top CNNs without labels

Jul 19

Hugging Face transformers now runs at native vLLM speed with zero porting

Jul 8

Not all correct AI answers are equal: Why Hugging Face is benchmarking the path, not just the result

Jun 18

OlmoEarth v1.1 slashes satellite AI compute by 3x by killing redundant tokens

May 19

Synchronous batching wastes 24% of GPU time: Here's how async cuts it to zero

May 14

AI code agents flood open source with 10x more pull requests no one can review

Apr 16

Hugging Face's v5 tokenizer overhaul separates architecture from vocab for custom training

Dec 18

Hugging Face Hub v1.0 breaks with its past, swaps backend to httpx and hf_xet

Oct 27

SGLang taps Hugging Face transformers as a backend, instantly unlocking models like Kyutai's Helium

Jun 23

Meta drops Llama 4: 10M context window and a bold bet on NoPE layers

Apr 5

Hugging Face bridges its entire pipeline to timm's 200K daily users via TimmWrapper

Jan 16

How Llama 3.2's RoPE encoding evolved from a broken integer trick

Nov 25

Hugging Face ships 24-hour fix for a silent training bug that broke gradient accumulation

Oct 16

Hugging Face squeezes 52% faster LLM inference by guessing when a draft model is wrong

Oct 8

Hugging Face admits its Transformers docs are a 'mess' and rebuilds them

Jun 7

PatchTST hits Hugging Face: IBM's 64-word time series trick goes zero-shot

Feb 1