Model Training, Hosting & Fine-Tuning Platforms

These tools support LLM model fine-tuning, hosting, and evaluation, with some offering MLOps support for versioning and reproducibility. Designed for researchers, developers, and enterprises with more control requirements, they often support open-weight models (e.g., LLaMA, Mistral) and custom datasets. Features may include quantization, adapter tuning (LoRA), private deployment, and telemetry. They are essential for building performant, domain-specific models and agents.

Tools:

• Ollama – Simplifies running and deploying LLMs locally on your machine with pre-built packages (e.g., LLaMA 2, Mistral). Great for offline or privacy-first development.

• Mistral Console – Offers hosted access to Mistral and Mixtral models, with API endpoints and upcoming fine-tuning capabilities.

• Abacus.ai – Anomaly Detection – Supports fine-tuning and inference of AI models (including LLMs) with special modules for tasks like anomaly detection, text classification, and generation.

• Lightning.ai – Framework for scaling PyTorch model training and fine-tuning in production. Includes Lightning Fabric and Cloud tools for deployment.