AI Open Source
Open source AI ecosystem and community
Articles
ScarfBench: Benchmarking AI Agents for Enterprise Java Framework⭐8
ScarfBench by IBM Research evaluates AI agents on enterprise Java framework migration, focusing on code semantics, correctness, and efficiency for real-world tr...
Featuring Every Evaluation Result Ever on Hugging Face Model Pages⭐7
Hugging Face now displays all benchmark evaluation results directly on model pages, including the updated cais/hle dataset, for easy model comparison at a glanc...
DiScoFormer: One Transformer for Density and Score, Across⭐9
The DiScoFormer Transformer unifies density estimation and score functions across arbitrary distributions, enabling single-model generative AI without retrainin...
Run a vLLM Server on HF Jobs in One Command⭐10
Deploy a vLLM inference server on Hugging Face Jobs in one command—no manual setup needed. Full guide for scalable LLM serving.
Which Tokens Does a Hybrid Model Predict Better?⭐9
Hybrid models excel at predicting rare, domain-specific tokens and structured data (numbers, code), improving accuracy by up to 15% versus dense models.
Introducing the FFASR Leaderboard: Benchmarking ASR in the Real World⭐8
Introducing FFASR Leaderboard: a realistic far-field ASR benchmark evaluating models in noisy, reverberant conditions. Compare WER scores, submit systems,
Accelerating Transformers Fine-Tuning with NVIDIA NeMo AutoModel⭐8
Accelerate transformer fine-tuning with NVIDIA NeMo AutoModel. Cut training time from days to hours using mixed precision and scalable GPU infrastructure.
Build Real Agentic Apps Using CUGA: Two Dozen Working Examples⭐9
Build 24 real agentic apps with CUGA's lightweight harness, from basic chatbots to multi-agent coordination and memory-augmented systems.
Shipping huggingface_hub Every Week with AI, Open Tools, and a⭐8
HuggingFace_hub now ships weekly using AI automation, open-source tools, and human review for fast, reliable releases. Learn how this hybrid approach speeds up ...
PP-OCRv6 on Hugging Face: 50-Language OCR from 1.5M to 34.5M⭐8
PP-OCRv6 on Hugging Face supports 50 languages via scalable 1.5M–34.5M parameter models for efficient OCR. Try the online demo.