TLDRocket
Sign in

Tools & Coding

966 summarised stories in Tools & Coding, each linking back to the original source. Browse all topics →

Tuesday, 3 June 2025

Holo1: New family of GUI automation VLMs powering GUI agent Surfer-H

Hugging Face 1 year ago 3

H Company released Holo1, a family of open-source action vision language models designed for GUI automation and web interaction tasks. The Holo1-7B model achieves 76.2% accuracy on UI localization benchmarks, while the web automation agent Surfer-H built on these models reaches 92.2% accuracy on real-world web tasks at $0.13 per task. Users can now deploy cost-efficient web automation solutions using open-source models available on Hugging Face instead of relying on expensive custom APIs.

Read the full story
Vision-Language Models

No GPU left behind: Unlocking Efficiency with Co-located vLLM in TRL

Hugging Face 1 year ago 41

TRL integrated vLLM into its GRPO training algorithm to allow training and inference to run on the same GPUs instead of separate ones, eliminating idle GPU time caused by previous server-mode setups. The co-located approach achieved up to 1.73× speedup on a 7B model and enabled training of a 72B model by combining vLLM's sleep mode with DeepSpeed ZeRO Stage 3 optimizations. This reduces hardware requirements and cost while improving overall training throughput by allowing GPUs to switch between training and generation tasks without waiting periods.

Read the full story
Inference Optimization

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.