Agnostic cloud gpu docker image and Jupyter lab (#1097) ece0211 unverified winglian commited on Jan 16
fix(readme): clarify custom user prompt [no-ci] (#1124) 9cd27b2 unverified Nanobit commited on Jan 16
update PR template so we can capture twitter or discord handles (#1121) [skip ci] 0abf4d6 unverified winglian commited on Jan 14
Enable or disable bf16 support based on availability (#1116) 0865613 unverified Simon Hällqvist commited on Jan 14
Disable caching on `--disable_caching` in CLI (#1110) d66b101 unverified casperhansen winglian commited on Jan 13
add gptneox embeddings, fix phi2 inputs, also fix the casting (#1083) 78c5b19 unverified winglian commited on Jan 11
Remove fused-dense-lib from requirements.txt (#1087) 91502b9 unverified casperhansen commited on Jan 10
add python 3.11 to the matrix for unit tests (#1085) [skip ci] 6c19e93 unverified winglian commited on Jan 10
optimize calculation of cu_seqlens from position_ids (#1084) [skip ci] 90036eb unverified winglian commited on Jan 10
use tags again for test image, only run docker e2e after pre-commit checks (#1081) 9032e61 unverified winglian commited on Jan 10
additional logging to get maximum token length of a sequence in the dataset (#1066) [skip ci] 2f2582e unverified winglian commited on Jan 10
update sharegpt conversations when chatml chat template is set (#1075) [skip ci] 0ce1a65 unverified winglian commited on Jan 10
fix: `train_on_inputs: true` ignored for sharegpt (#1045) [skip ci] 043c386 unverified Nanobit winglian commited on Jan 10
be more robust about checking embedding modules for lora finetunes (#1074) [skip ci] 0f10080 unverified winglian commited on Jan 10
swap the data collator for evals if not using sample packing (#1076) ead34c5 unverified winglian commited on Jan 10
Separate AutoGPTQ dep to `pip install -e .[auto-gptq]` (#1077) 9be92d1 unverified casperhansen commited on Jan 9
Add: mlflow for experiment tracking (#1059) [skip ci] 090c24d unverified Johan Hansson winglian commited on Jan 9
Cosine learning rate schedule - minimum learning rate (#1062) 04b978b unverified ricdomolm winglian commited on Jan 9
Efficiently get the length of the tokenized docs (#1063) 81d3845 unverified ricdomolm winglian commited on Jan 8
streaming multipack for pretraining dataset (#959) 553c80f unverified jinwonkim93 jinwonkim93@github.com winglian commited on Jan 6
feat: always push checkpoint to hub if set (#1049) [skip ci] cbdbf9e unverified Nanobit commited on Jan 5
feature: better device mapping for large models (#918) bdfefaf unverified dg-kalle Karl-Johan Alm winglian commited on Jan 5