Hugging Face + Ertas
Import models and datasets from the Hugging Face Hub directly into Ertas Studio, fine-tune them on Ertas's managed cloud, and push your refined models back to the community.
Overview
Hugging Face has become the central hub for open-source AI, hosting hundreds of thousands of pre-trained models, datasets, and spaces. From foundation models like Llama and Mistral to task-specific fine-tunes for sentiment analysis, code generation, and medical NLP, the Hugging Face ecosystem provides the starting point for nearly every modern AI workflow. Its standardized model card format, licensing metadata, and community-driven curation make it easy to discover and evaluate models before committing to a training pipeline.
Ertas AI deeply integrates with the Hugging Face Hub so you can browse, pull, and publish models without leaving Ertas Studio. Whether you are bootstrapping a new project with a base model or contributing a domain-specific fine-tune back to the community, the integration removes friction and keeps your workflow streamlined from discovery through deployment.
How Ertas Integrates
Ertas Studio connects to the Hugging Face Hub API, allowing you to search for models and datasets by task, architecture, or license directly from the Studio UI. Once you select a model, Ertas imports it into its managed cloud environment, converts it to the optimal format for fine-tuning, and pre-populates training configuration based on the model's architecture and size. Datasets can be pulled in the same way, with automatic conversion to JSONL format for Ertas training pipelines.
After fine-tuning is complete, you can push your model back to the Hugging Face Hub directly from the Studio UI. Ertas automatically generates a model card with training metadata, evaluation metrics, and dataset provenance so the community can understand and reproduce your results. Private repositories are supported, giving teams full control over visibility while still leveraging the Hub's versioning and collaboration features.
Getting Started
- 1
Authenticate with Hugging Face
Add your Hugging Face API token in Ertas Studio settings to enable seamless access to public and private repositories on the Hub.
- 2
Browse and select a base model
Use the integrated model browser to search by task type, model size, or license. Preview model cards and benchmark scores before importing.
- 3
Import model into Ertas
Import the selected model weights and tokenizer into Ertas's managed cloud. Ertas handles format conversion and storage optimization automatically.
- 4
Pull training datasets
Import datasets from the Hub and convert them to JSONL format compatible with Ertas training pipelines. Apply optional filters, splits, and transformations during import.
- 5
Fine-tune with Ertas Studio
Launch a fine-tuning job using LoRA or full-parameter training. Ertas pre-fills recommended hyperparameters based on the model architecture and dataset size.
- 6
Push fine-tuned model back to the Hub
Publish your fine-tuned model to the Hugging Face Hub with auto-generated model cards, training logs, and evaluation metrics attached.
# After fine-tuning in Ertas Studio and publishing to Hugging Face,
# load your model directly from the Hub
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained(
"my-org/llama-3-alpaca-ft",
device_map="auto",
)
tokenizer = AutoTokenizer.from_pretrained("my-org/llama-3-alpaca-ft")
inputs = tokenizer("Summarize this document:", return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))Benefits
- Access over 500,000 open-source models without leaving Ertas Studio
- Automatic format conversion between Hugging Face and Ertas-native formats
- One-click dataset import with JSONL conversion for training pipelines
- Auto-generated model cards with full training provenance on publish
- Support for private Hub repositories for enterprise and team workflows
- Community visibility for your fine-tuned models to attract feedback and collaboration
Related Resources
Fine-Tuning
GGUF
JSONL
LoRA
Getting Started with Ertas: Fine-Tune and Deploy Custom AI Models
Introducing Ertas Studio: A Visual Canvas for Fine-Tuning AI Models
How to Fine-Tune an LLM: The Complete 2026 Guide
Fine-Tuning Llama 3: A Practical Guide for Your Use Case
Fine-Tune AI Models Without Writing Code
Ertas Studio vs. Unsloth vs. Axolotl: Fine-Tuning Tools Compared (2026)
Model Distillation with LoRA: Training Smaller Models from Frontier Outputs
llama.cpp
LM Studio
Ollama
Text Generation Web UI
Unsloth
vLLM
Ertas for Healthcare
Ertas for SaaS Product Teams
Ertas for Education
Ertas for Code Generation
Ertas for ML Engineers & Fine-Tuning Practitioners
Ship AI that runs on your users' devices.
Early bird pricing starts at $14.50/mo — locked in for life. Plans for builders and agencies.