Hugging Face is promoting a new workflow that combines the Unsloth library with Hugging Face Jobs to enable fast, low‑cost fine‑tuning of small language models such as LiquidAI/LFM2.5-1.2B-Instruct. Unsloth claims to deliver roughly twice the training speed and 60% less VRAM usage compared to standard methods, making it possible to train a model for only a few dollars. The company is offering free credits and a one‑month Pro subscription to members of the Unsloth Jobs Explorers organization, and users can submit jobs via the hf jobs CLI. Coding agents like Claude Code and Codex can automatically generate training scripts and submit them to Hugging Face Jobs, which runs on managed cloud GPUs and provides monitoring links through Trackio. This approach lowers the cost and technical barrier to fine‑tune small language models, democratizing access to custom AI solutions and encouraging broader experimentation across devices and industries.
Train AI models with Unsloth and Hugging Face Jobs for FREE
Hugging Face is promoting a new workflow that combines the Unsloth library with Hugging Face Jobs to enable fast, low‑cost fine‑tuning of small language models such as LiquidAI/LFM2.5-1.2B-Instruct. Unsloth claims to deliver roughly twice the training speed and 60% less VRAM usage compared to standard methods, making it possible to train a model for only a few dollars. The company is offering free credits and a one‑month Pro subscription to members of the Unsloth Jobs Explorers organization, and users can submit jobs via the hf jobs CLI. Coding agents like Claude Code and Codex can automatically generate training scripts and submit them to Hugging Face Jobs, which runs on managed cloud GPUs and provides monitoring links through Trackio.
Why it matters
By lowering the cost and technical barrier to fine‑tune small language models, this approach democratizes access to custom AI solutions and encourages broader experimentation across devices and industries.