mosaicml/llm-foundry

Pull Requests Commits

Ning Wang committed 1 year ago

15fb4ffa

Merge branch 'mpt-7b-test' of github.com:mosaicml/llm-foundry into mpt-7b-test

Ning Wang committed 1 year ago

a15d9cb2

make eval loader to None

Ning Wang committed 1 year ago

e68b7401

make eval load to None

Ning Wang committed 1 year ago

fbe64944

Ning Wang committed 1 year ago

e3bb4fbc

Ning Wang committed 1 year ago

77187d36

add memorysnapshot to callbacks (#810)

cli99 committed 1 year ago

Verified 491fa7ab

Fix eval.py with lora (#965)

dakinggg committed 1 year ago

Verified 12d1ca73

Fix typo (#966)

irenedea committed 1 year ago

Verified c7c9d247

Add streams support (#946)

bigning committed 1 year ago

Verified aa0ea6e2

Use create_model_version instead of register_model (#953)

dakinggg committed 1 year ago

Verified 2f64a144

Add fully configurable activation checkpointing (#951)

cli99 committed 1 year ago

Verified 2e59620a

add finutuning with streaming dataset example (#945)

bigning committed 1 year ago

Verified 60cdd0be

Bump mcli yaml foundry version to v0.5.0 (#959)

irenedea committed 1 year ago

Verified 9f101847

Update TUTORIAL.md (#957)

sdonoso committed 1 year ago

Verified fe17c224

allow te to use meta device with deferred init (#958)

cli99 committed 1 year ago

Verified 8c7d6f43

Add default signature to mlflow saved model (#952)

dakinggg committed 1 year ago

Verified 60ab97fb

Add finetuning streaming dataset conversion (#933)

bigning committed 1 year ago

Verified 105f7663

Fix chain-of-thought tasks (#824)

bmosaicml committed 1 year ago

Verified 6591f480

Bump llm-foundry version to 0.5.0 (#948)

irenedea committed 1 year ago

Verified a667ebf5

Add and use VersionedDeprecationWarning (#944)

irenedea committed 1 year ago

Verified c1c4bbfb

Retrieve license information when local files are provided for a pretrained model (#943)

jerrychen109 committed 1 year ago

Verified 2e0a8458

mvpatel2000 committed 1 year ago

Verified 3f21bb7e

Update lora docs (#941)

dakinggg committed 1 year ago

Verified 25599294

Fixing the gen_attention_mask_in_length function to handle the case when sequence id is -1 due to attention masking (#940)

ShashankMosaicML committed 1 year ago

Verified ad126a62

Remove extra call to .to and load_state_dict in hf checkpointer (#939)

dakinggg committed 1 year ago

Verified b9d2bfaf

Refactoring the function to accept list of metric names instead of a dictionary of metrics. (#938)

ShashankMosaicML committed 1 year ago

Verified 706ea7dd

Switch to the Composer integration of LoRA (works with FSDP) (#886)

dakinggg committed 1 year ago

Verified 15ee0acc

Update eval_gauntlet_callback.py with math.log2 (#821)

Skylion007 committed 1 year ago

Verified d9874d2a

dakinggg committed 1 year ago

Verified 86b5a981

Older