Skip to content

Instantly share code, notes, and snippets.

View markrogersjr's full-sized avatar

Mark Rogers markrogersjr

View GitHub Profile
@markrogersjr
markrogersjr / gpt_oss_flash.md
Last active November 26, 2025 15:39
GPT-OSS with Flash Attention and Memory-Efficient Attention via PyTorch-Native SDPA

GPT-OSS with Flash Attention and Memory-Efficient Attention via PyTorch-Native SDPA

from time import time
import warnings

import torch
import transformers