🚀🚀 「大模型」2小时完全从0训练64M的小参数GPT!🌏 Train a 64M-parameter GPT from scratch in just 2h!
fix: resolve CUDA OOM in train_grpo.py on GPUs with <=8GB VRAM
T
TKiteRunner committed
10776417aae969466d4fece8d5b4d5e808ebe89e
Parent: 5020dc9
🚀🚀 「大模型」2小时完全从0训练64M的小参数GPT!🌏 Train a 64M-parameter GPT from scratch in just 2h!