mosaicml/llm-foundry

Pull Requests Commits

bmosaicml committed 1 year ago

abe79fbc

bmosaicml committed 1 year ago

c4f3ca7e

bmosaicml committed 1 year ago

09e2ad42

add big bench hard

bmosaicml committed 1 year ago

d14861b2

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

Throw error when no EOS (#922)

KuuCi committed 1 year ago

Verified 4b82876d

Remove .ci folder and move FILE_HEADER (#931)

irenedea committed 1 year ago

Verified 203edad0

Changes to setup.py invalidate docker cache. Use branch name in dockerfile (#930)

irenedea committed 1 year ago

Verified 7a81f194

Add rich formatting to tracebacks (#927)

jjanezhang committed 1 year ago

Verified b928de8f

Adding deprecation warning for Flash Attention 1 and user warning against using Triton attention. (#921)

ShashankMosaicML committed 1 year ago

Verified 169b6536

Fixes yaml linting (#920)

b-chu committed 1 year ago

Verified 7dc51e96

fix dash issue (#919)

milocress committed 1 year ago

Verified ff8c32be

Tiktoken chat format fix (#893)

rajammanabrolu committed 1 year ago

Verified 8c5182c4

Create HF checkpointer model on meta device (#916)

dakinggg committed 1 year ago

Verified 5ccceffa

Make yamllint consistent with Composer (#918)

b-chu committed 1 year ago

Verified 34a1dad9

Fix fmapi_chat for instruct models and custom tokenizers (#914)

aspfohl committed 1 year ago

Verified f132f055

Warn instead of error on tokenizer-only with http (#904)

jerrychen109 committed 1 year ago

Verified b48fa58f

Add the model license file for mlflow (#915)

dakinggg committed 1 year ago

Verified 34cdaf68

Older