Search results

Training diverges without flash attention

bugSomething isn't workingSomething isn't working

Status: Open.

#47 In simple-stories/simple_stories_train;

· danbraunai opened on Oct 21, 2025

Models are not trained on n_ctx-length sequences

bugSomething isn't workingSomething isn't working

Status: Open.

#46 In simple-stories/simple_stories_train;

· danbraunai opened on Oct 16, 2025

Save the accompanying tokenizer after training and verify the push to HF script

enhancementNew feature or requestNew feature or request

Status: Open.

#42 In simple-stories/simple_stories_train;

· chandanms opened on Aug 15, 2025

Add from_pretrained() methods to each model

enhancementNew feature or requestNew feature or request

priority-medium

Status: Open.

#38 In simple-stories/simple_stories_train;

· danbraunai opened on Aug 3, 2025

Support GPT2 architecture in training script

enhancementNew feature or requestNew feature or request

good first issueGood for newcomersGood for newcomers

priority-low

Status: Open.

#19 In simple-stories/simple_stories_train;

· danbraunai opened on Dec 1, 2024

Allow for training on custom and tiktoken tokenizer

enhancementNew feature or requestNew feature or request

priority-high

Status: Open.

#8 In simple-stories/simple_stories_train;

· danbraunai opened on Aug 3, 2024