Unsloth enables training and running models on AMD GPUs
Meta
Unsloth
Unsloth, a framework for efficient LLM fine-tuning, now supports AMD GPUs. This allows users to train and deploy large language models on AMD hardware, expanding options beyond NVIDIA GPUs.
Unsloth, a popular open-source framework for fine-tuning large language models, has announced support for AMD GPUs. Previously limited to NVIDIA GPUs with CUDA, Unsloth now enables training and inference on AMD accelerators, leveraging ROCm software. This expands hardware choices for AI developers, potentially reducing costs and increasing accessibility. The update includes optimizations specific to AMD architecture, promising performance improvements. Unsloth is known for its speed and memory efficiency, supporting models like Llama, Mistral, and others.
- Сокращения
- GPU = Graphics Processing Unit
- LLM = Large Language Model
- CUDA = Compute Unified Device Architecture
- ROCm = Radeon Open Compute platform
Source: Meta AI (GNews) —
original