view article Article Welcome GPT OSS, the new open-source model family from OpenAI! By reach-vb and 11 others • 2 days ago • 387
view reply I do not, but you could change the code to dispatch a 8 rows GEMM to the dense MFMA with 8 rows of padding and check the numbers then!Your understanding of the dispatch logic is correct.
view article Article Sensitivity Aware Mixed Precision Quantization V1 By badaoui and 1 other • Jun 13 • 19
view article Article Hugging Face and AMD partner on accelerating state-of-the-art models for CPU and GPU platforms By juliensimon • Jun 13, 2023 • 4
Running 2.97k 2.97k The Ultra-Scale Playbook 🌌 The ultimate guide to training LLM on large GPU Clusters