Activity
Update Ex7 to new code from LUMI AI workshop, now with GPT-3-like mod…
Update Ex7 to new code from LUMI AI workshop, now with GPT-3-like mod…
Set report_to="none" for 04b training params, so it does not try to p…
Set report_to="none" for 04b training params, so it does not try to p…
Updated to new reservation, commented out reservation line for now.
Updated to new reservation, commented out reservation line for now.
Update to new course project and project folder name
Update to new course project and project folder name
Replaced removed function datasets.load_metric with evaluate in Exerc…
Replaced removed function datasets.load_metric with evaluate in Exerc…
Changed pytorch_generate_gpt.ipynb to generate one sequence by defaul…
Changed pytorch_generate_gpt.ipynb to generate one sequence by defaul…
Set fixed per_device_*_batch_size for pytorch_imdb_gpt scripts
Set fixed per_device_*_batch_size for pytorch_imdb_gpt scripts
Add figure with both train and test loss/accuracies
Add figure with both train and test loss/accuracies
Add torchinfo summary to the end of the nodebook
Add torchinfo summary to the end of the nodebook
Add plotting of convolutions in ex 03, change to new project
Add plotting of convolutions in ex 03, change to new project
Update Exercise_6.md description
Update Exercise_6.md description
Change ex8 to default to constant per-device batch size
Change ex8 to default to constant per-device batch size
Correct path to Ex 7 pre-created model
Correct path to Ex 7 pre-created model
Fix error in pytorch_generate_gpt.ipynb
Fix error in pytorch_generate_gpt.ipynb
Changed a bit the wording in ex 02
Changed a bit the wording in ex 02
Add note about datasets install on Colab
Add note about datasets install on Colab
Add DistilBERT as 04b, add plotting accuracy, various fixes.
Add DistilBERT as 04b, add plotting accuracy, various fixes.
Replace deprecated torchtext with Huggingface Datasets and Tokenizers.
Replace deprecated torchtext with Huggingface Datasets and Tokenizers.