What's Changed
- Post-release setup for 0.5.3.dev1 by @rasbt in #1799
- Add Phi 3 128k model by @deveworld in #1800
- Add token counts to compute performance by @rasbt in #1801
- Fixed the issue that precision is always "32-true". by @jianpingw in #1802
- Add Nvidia Llama 3.1 70B Nemotron weights by @rasbt in #1803
- Choose evaluation example from test set by @rasbt in #1804
- Pretrain tok sec by @rasbt in #1805
- typo in convert_to_litgpt command by @wasifferoze in #1807
- Move distributed all_reduce import into a function by @IvanYashchuk in #1810
- Remove hardcoded 32-precision conversion by @rasbt in #1814
New Contributors
- @deveworld made their first contribution in #1800
- @jianpingw made their first contribution in #1802
- @wasifferoze made their first contribution in #1807
- @IvanYashchuk made their first contribution in #1810
Full Changelog: v0.5.2...v0.5.3
Contributors
jianpingw, rasbt, and 3 other contributors