Repository Issue Activity (beta)

#751 test

#748 Realtime response generation simulation instead of one big response at the end.

#644 Loss Distribution at 400k and 600k steps - OWT - 124M model

#745 Add REFUTE scientific critique + calibration benchmark

#742 Teaching note: pair nanoGPT demos with BGPT evidence retrieval

#738 GPT.generate() crashes on temperature=0 instead of doing greedy decoding

#616 Training on Intel(R) Data Center GPU Max 1550 GPUs (formerly codenamed Ponte Vecchio)

#606 How cooked? Need help..

#435 How to train nanoGPT using TPU's?

#731 Code quality scan: 9 findings (C-, 68/100)

#468 Is this loss curve normal

#554 Pretraining loss explosion

#524 Pretraining Divergence

#303 Train/Val Loss Issues when training GPT-2 from OWT

#137 During the training, the loss first drops and then rises

#717 Critical Sandbox Escape Vulnerability in nanochat

#722 the model is not domain agnostic

#708 Qodo AntiSlop scan found 8 issues across 7 recent PRs

#704 small warning

#693 Scalar C++ Transformer Implementation — Verified Against PyTorch

#692 Regarding your LLMOS concept

#686 Why not just surpass 5.2 directly? Let's add the LiE enhancement plugin.

#685 Meet TrueNanoGPT

#637 Idea: Enhanced GPT with Mathematical Extensions (teaser)

#127 nanoGPT with Lightning Fabric