Loading…
Codex is Open Sourcing AI models
Hugging Faceben burtenshaw, shaun smith
Summary
Hugging Face Skills equips AI coding agents like OpenAI Codex to execute end-to-end machine learning workflows. By reading AGENTS.md files and interfacing via the Model Context Protocol, Codex automates dataset validation, training script updates, and job submissions to Hugging Face Jobs. The workflow supports methods including supervised fine-tuning, direct preference optimization, and reinforcement learning for models ranging from 0.5B to 7B parameters. Throughout execution, Codex tracks live metrics via Trackio, records benchmark evaluations against baselines, and maintains Markdown reports. Once training concludes, Codex merges LoRA adapters, applies GGUF quantization, and publishes the resulting models to the Hugging Face Hub for local deployment.
Context
Engineers sought to expand beyond single-prompt coding assistants to let AI agents autonomously run, monitor, evaluate, and report on complete machine learning experiments.
Approach / What changed
Hugging Face integrated its skills repository with Codex using AGENTS.md files and Model Context Protocol servers to automate Hugging Face Jobs submissions, Trackio metric logging, and model conversion.
Takeaways
- Codex automatically selects compute hardware and estimates costs based on model size, using configurations like t4-small for sub-1B models and a10g-large or a100-large with LoRA for 3B to 7B models.
- The Hugging Face Skills integration supports multi-stage pipelines including supervised fine-tuning, direct preference optimization, and reinforcement learning with verifiable rewards.
- The system handles post-training workflows by merging LoRA adapters, applying GGUF quantization such as Q4_K_M, and publishing models directly to the Hugging Face Hub.
Related reading
We Got Claude to Fine-Tune an Open Source LLM
Hugging Face Skills equips coding agents like Claude Code, OpenAI Codex, and Google's Gemini CLI to manage end-to-end language model fine-tuning. Using the hf-llm-trainer skill, an agent validates dataset formats, selects cloud hardware, configures authentication, and submits jobs to Hugging Face Jobs. Supported techniques include Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Group Relative Policy Optimization (GRPO) for models ranging across various parameter sizes. The integration incorporates Trackio for real-time monitoring and automates LoRA configuration for larger models. Once training completes, the agent pushes artifacts to the Hugging Face Hub and can convert models to GGUF format for local execution.
ben burtenshaw, shaun smithHow to Run an Autoresearch Workflow with RL Agent Skills and NVIDIA NeMo