"Innovesta team on AMD AI League Hackathon"
Team Idea
ROCM-QLORA is a pure PyTorch, ROCm-native QLoRA fine-tuning stack for AMD GPUs. It implements NF4/INT8 quantization, LoRA, optimized Triton kernels, memory-efficient training, and multi-GPU support, enabling efficient LLM fine-tuning without CUDA or bitsandbytes.
Team Members
.png&w=256&q=75)
Bala Mosay James
balamosayjames93244
Submission
No submission yet