RuntimeError: CUDA out of memory. Tried to allocate 308.50 MiB (GPU 0; 10.92 GiB total capacity; 7.17 GiB already allocated; 275.50 MiB free; 247.76 MiB cached)

lipiji / TranSummar

Transformer for abstractive summarization on cnn/daily-mail and gigawords

MIT License

140 stars 20 forks source link

RuntimeError: CUDA out of memory. Tried to allocate 308.50 MiB (GPU 0; 10.92 GiB total capacity; 7.17 GiB already allocated; 275.50 MiB free; 247.76 MiB cached) #3

Closed zhongxia96 closed 5 years ago

zhongxia96 commented 5 years ago

Hello, may I ask what is your GPU configuration? When I run it, there is always an error of insufficient memory, and it can only run if batch_size is adjusted to 8.

lipiji commented 5 years ago

Hi, I use a m40 gpu with 24g memory. You can modify the configs according to your local conditions. https://github.com/lipiji/TranSummar/blob/master/configs.py