Pydantic 2.x cfg (#1239)

fix(readme): Clarify doc for tokenizer_config (#1323) [skip ci]

fix(readme): update inference md link (#1311) [skip ci]

Add seq2seq eval benchmark callback (#1274)

Scheduler implementation of Continual Pre-Training of Large Language Models: How to (re)warm your model? (#1273)

allow the optimizer prune ratio for ReLoRA to be configurable (#1287)

Update README.md (#1281)

add support for https remote yamls (#1277)

allow remote data paths (#1278)

copy edits (#1276)

Add link to axolotl cloud image on latitude (#1275)

contributor avatars (#1269)

add contact info for dedicated support for axolotl [skip ci] (#1243)

support for true batches with multipack (#1230)

Fix and document test_datasets (#1228)

Peft lotfq (#1222)

Feat/chatml add system message (#1117)

Mixtral fixes 20240124 (#1192) [skip ci]

update docs [skip ci] (#1176)

Fine-Tuning Mistral-7b for Real-World Chatbot Applications Using Axolotl (Lora used) (#1155)

Update README.md (#1169) [skip ci]

set fp16 to false if bf16, update bf16: auto in example YAMLs (#1122) [skip ci]

Deprecate max packed sequence len (#1141)

feat(dataset): add config to keep processed dataset in memory (#1152)

Fix link for Minotaur model (#1146) [skip-ci]

Add shifted sparse attention (#973) [skip-ci]

Agnostic cloud gpu docker image and Jupyter lab (#1097)

Add `layers_to_transform` for `lora_config` (#1118)

fix(readme): clarify custom user prompt [no-ci] (#1124)

Add link on README to Docker Debugging (#1107)

Update README.md (#1103)

Add Debugging Guide (#1089)

paired kto support (#1069)

Add: mlflow for experiment tracking (#1059) [skip ci]

Cosine learning rate schedule - minimum learning rate (#1062)

Sponsors (#1065)

feature: better device mapping for large models (#918)

set default for merge (#1044)

[Docs] delete unused cfg value `lora_out_dir` (#1029)

chore(readme): update instruction to set config to load from cache (#1030)

use recommended setting for use_reentrant w gradient checkpointing (#1021)

Adds chat templates (#1022)

feat: expose bnb kwargs (#1018)

feat: remove need to add load_in* during merge (#1017)

[Docs] Nit: Remind people to auth to wandb if they are going to use it (#1013)

Update README.md (#1012)

remove landmark attn and xpos rope implementations (#1010)

Update README.md (#966)

Add docs (#947)

fix: switch to using the HuggingFace Transformers NEFT implementation (#941)