search

Version 1.1

We are excited to announce the release of GenAI Studio v1.1.0, featuring significant enhancements to boost your GenAI development and deployment experience.

hashtag
✨ What's New?

hashtag
🔧 Revamped Management Interface & Optimized Architecture

  • Entirely refactored GenAI Studio management interface

  • Enhanced architecture to better support Inference Apps such as Flux.1 Schnell (Text to Image) and ScrapeGraphAI

  • Easier management and streamlined deployment of diverse GenAI applications

hashtag
🔄 RAGOps Auto-Sync

  • Automatic document synchronization from designated folders directly to your vector databases

  • Significantly improved RAG (Retrieval-Augmented Generation) workflow efficiency

hashtag
📊 Real-time System Monitoring

  • Integrated Grafana and Prometheus for real-time system performance tracking

  • Proactively detect and address potential issues before they escalate

hashtag
🚀 Model Conversion & Inference Runtime

  • Model conversion functionality has been added, along with the availability of the EdgeAI SDK for inference-side downloads and deployment. This makes it easier to deploy models to edge devices.

hashtag
🎯 Enhanced Fine-Tuning with LORA

  • Integrated LORA (Low-Rank Adaptation) fine-tuning support via Unsloth

  • Compatible with inference models like DeepSeek for precise, customized adjustments

hashtag
⚙️ Upgrades & Maintenance

  • AnythingLLM updated to v1.7.5, providing the latest features and improved security.

  • Phison Firmware upgraded to NXUN202.00, enhancing hardware performance and stability.

hashtag
Third-Party Updates

GenAI Studio utilizes the following components:

  • node-exporter (1.8.2) Exposes desired host metrics to Prometheus.

  • dcgm-exporter (4.0.0-4.0.1-ubuntu22.04) Exposes host GPU metrics to Prometheus.

  • Prometheus (3.1.0) Collects desired metrics as a data source for Grafana.

  • Grafana (11.4.0) Serves as the resource monitoring dashboard.

  • Phison aiDAPTIVLink (NXUN202.00) Leverages middleware for model fine-tuning with full parameters.

  • Ollama (0.6.2) Acts as the inference server.

  • llama.cpp (full-cuda-b4897) Converts GGUF model file formats.

  • vsFTP (3.0.5) Provides model files for downloading.

  • Qdrant (1.12.4) Functions as the vector database.

  • Flowise (2.2.7-patch.1) Automates RAGOps functionality and workflows.

  • PostgreSQL (16.4) Serves as the relational database.

  • Unsloth (2025.3.18) Performs model fine-tuning with LoRA mode.

PreviousVersion HistoryNextVersion 1.0

Last updated