Lightning-AI/litgpt v0.5.8

v0.5.8

Repository: Lightning-AI/litgpt

Tag: v0.5.8

Published: 2025-05-06T10:22:06Z

Prerelease: no

Release notes: Many great updates!

What's Changed

• add missing r1 prompt style by @ali-alshaar7 in https://github.com/Lightning-AI/litgpt/pull/1929
• fix incremental save for PyTorch 2.6 by @t-vi in https://github.com/Lightning-AI/litgpt/pull/1928
• fix: resolve failing CI by @Borda in https://github.com/Lightning-AI/litgpt/pull/1944
• handle wrapped thundermodules in generate by @t-vi in https://github.com/Lightning-AI/litgpt/pull/1955
• fix skip condition by @t-vi in https://github.com/Lightning-AI/litgpt/pull/1956
• ci: use HF cache by @Borda in https://github.com/Lightning-AI/litgpt/pull/1958
• ci: split HF caching by @Borda in https://github.com/Lightning-AI/litgpt/pull/1960
• bump: PT 2.6 + bitsandbytes & standalone tests by @Borda in https://github.com/Lightning-AI/litgpt/pull/1959
• prune whitespaces for code readability by @Borda in https://github.com/Lightning-AI/litgpt/pull/1962
• fixing various typos in examples & tutorials by @Borda in https://github.com/Lightning-AI/litgpt/pull/1963
• fix n_query_groups for llama-3.1-405b by @ysjprojects in https://github.com/Lightning-AI/litgpt/pull/1946
• tests: make flaky test due to connection issues by @Borda in https://github.com/Lightning-AI/litgpt/pull/1964
• Fix: incorrect gradient accumulation steps bug by @ysjprojects in https://github.com/Lightning-AI/litgpt/pull/1947
• fix: use default num_nodes=1 for back-compatibility by @Borda in https://github.com/Lightning-AI/litgpt/pull/1967
• Do not wrap LoRA layers with FSDP by @janEbert in https://github.com/Lightning-AI/litgpt/pull/1538
• Speculative decoding: Base implementation by @Andrei-Aksionov in https://github.com/Lightning-AI/litgpt/pull/1938
• Better clarity on SFT dataset attributes by @ysjprojects in https://github.com/Lightning-AI/litgpt/pull/1970
• Enforce Consistent Formatting and Validation for YAML Files by @Borda in https://github.com/Lightning-AI/litgpt/pull/1977
• Apply Standard Formatting and Fix Import & Test Name Issues by @Borda in https://github.com/Lightning-AI/litgpt/pull/1981
• Setting config.sliding_window_layer_stride explicity by @ysjprojects in https://github.com/Lightning-AI/litgpt/pull/1972
• feat: add linear rope type by @k223kim in https://github.com/Lightning-AI/litgpt/pull/1982
• feat: update tests for transformers 4.50.2 by @k223kim in https://github.com/Lightning-AI/litgpt/pull/1983
• fix: test_tokenizer_against_hf by @Borda in https://github.com/Lightning-AI/litgpt/pull/1984
• feat: replace sliding window type with offset by @k223kim in https://github.com/Lightning-AI/litgpt/pull/1989
• ci: with pull_request_target by @Borda in https://github.com/Lightning-AI/litgpt/pull/1992
• Phi4 mini by @ysjprojects in https://github.com/Lightning-AI/litgpt/pull/1949
• aggregate val_loss by @ysjprojects in https://github.com/Lightning-AI/litgpt/pull/1971
• feat: add local base freq for rope by @k223kim in https://github.com/Lightning-AI/litgpt/pull/1993
• test: flexible wait for serve start by @Borda in https://github.com/Lightning-AI/litgpt/pull/1996
• fix: replace sliding window configuration parameters to sliding windows indices by @k223kim in https://github.com/Lightning-AI/litgpt/pull/1995
• QwQ-32B by @ysjprojects in https://github.com/Lightning-AI/litgpt/pull/1952
• feat: run thunder tests as part of LitGPT CI by @deependujha in https://github.com/Lightning-AI/litgpt/pull/1975
• try pyupgrade-up py38 by @Borda in https://github.com/Lightning-AI/litgpt/pull/1999
• [1/4] feat: add gemma 3 27b by @k223kim in https://github.com/Lightning-AI/litgpt/pull/1998
• [2/4] add gemma 3 1b by @k223kim in https://github.com/Lightning-AI/litgpt/pull/2000
• [3/4] feat: add gemma 3 4b by @k223kim in https://github.com/Lightning-AI/litgpt/pull/2001
• feat: add gemma-3-12b by @k223kim in https://github.com/Lightning-AI/litgpt/pull/2002
• Add resume for adapter_v2, enable continued finetuning for adapter by @altria-zewei-wang in https://github.com/Lightning-AI/litgpt/pull/1354
• Fix/loading gemma 3 1b by @pquadri in https://github.com/Lightning-AI/litgpt/pull/2004
• feat: add gemma 3 in readme and tutorials by @k223kim in https://github.com/Lightning-AI/litgpt/pull/2005
• add borda as codeowner by @t-vi in https://github.com/Lightning-AI/litgpt/pull/2007
• example for full finetuning with python code by @astrobdr in https://github.com/Lightning-AI/litgpt/pull/1331
• feat: add tests for gemma3 by @k223kim in https://github.com/Lightning-AI/litgpt/pull/2006
• building tutorials as mkdocs pages by @Borda in https://github.com/Lightning-AI/litgpt/pull/2011
• Add mlflow logger support by @topikachu in https://github.com/Lightning-AI/litgpt/pull/1985
• fix support for litserve>0.2.4 by @ali-alshaar7 in https://github.com/Lightning-AI/litgpt/pull/1994
• Cast tensors in KVCache only when needed by @Andrei-Aksionov in https://github.com/Lightning-AI/litgpt/pull/2017
• feat: load only text weights from multimodal gemma by @pquadri in https://github.com/Lightning-AI/litgpt/pull/2008
• Feature: Adds support for OpenAISpec in litgpt serve by @bhimrazy in https://github.com/Lightning-AI/litgpt/pull/1943
• drop upper bounds in dependencies by @t-vi in https://github.com/Lightning-AI/litgpt/pull/2022

New Contributors

• @syntheticgio made their first contribution in https://github.com/Lightning-AI/litgpt/pull/1939
• @deependujha made their first contribution in https://github.com/Lightning-AI/litgpt/pull/1975
• @altria-zewei-wang made their first contribution in https://github.com/Lightning-AI/litgpt/pull/1354
• @pquadri made their first contribution in https://github.com/Lightning-AI/litgpt/pull/2004
• @astrobdr made their first contribution in https://github.com/Lightning-AI/litgpt/pull/1331
• @topikachu made their first contribution in https://github.com/Lightning-AI/litgpt/pull/1985
• @bhimrazy made their first contribution in https://github.com/Lightning-AI/litgpt/pull/1943
• @Lynsoo made their first contribution in https://github.com/Lightning-AI/litgpt/pull/2018

Full Changelog: https://github.com/Lightning-AI/litgpt/compare/v0.5.7...v0.5.8