mosaicml/llm-foundry

Pull Requests Commits

add human eval pack datasets (https://arxiv.org/pdf/2308.07124.pdf)

Ethan Turok committed 2 years ago

b92c3bbe

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

jerrychen109 committed 2 years ago

Verified 2e0a8458

mvpatel2000 committed 2 years ago

Verified 3f21bb7e

Update lora docs (#941)

dakinggg committed 2 years 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 2 years ago

Verified ad126a62

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

dakinggg committed 2 years ago

Verified b9d2bfaf

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

ShashankMosaicML committed 2 years ago

Verified 706ea7dd

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

dakinggg committed 2 years ago

Verified 15ee0acc

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

Skylion007 committed 2 years ago

Verified d9874d2a

dakinggg committed 2 years ago

Verified 86b5a981

Throw error when no EOS (#922)

KuuCi committed 2 years ago

Verified 4b82876d

Remove .ci folder and move FILE_HEADER (#931)

irenedea committed 2 years ago

Verified 203edad0

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

irenedea committed 2 years ago

Verified 7a81f194

Add rich formatting to tracebacks (#927)

jjanezhang committed 2 years ago

Verified b928de8f

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

ShashankMosaicML committed 2 years ago

Verified 169b6536

Fixes yaml linting (#920)

b-chu committed 2 years ago

Verified 7dc51e96

fix dash issue (#919)

milocress committed 2 years ago

Verified ff8c32be

Tiktoken chat format fix (#893)

rajammanabrolu committed 2 years ago

Verified 8c5182c4

Create HF checkpointer model on meta device (#916)

dakinggg committed 2 years ago

Verified 5ccceffa

Make yamllint consistent with Composer (#918)

b-chu committed 2 years ago

Verified 34a1dad9

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

aspfohl committed 2 years ago

Verified f132f055

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

jerrychen109 committed 2 years ago

Verified b48fa58f

Add the model license file for mlflow (#915)

dakinggg committed 2 years ago

Verified 34cdaf68

Add support for eval_loader & eval_subset_num_batches in async callback (#834)

aspfohl committed 2 years ago

Verified bdcce637

Add flag to enable/disable param upload (#912)

ngcgarcia committed 2 years ago

Verified 534f5b4d

Adds support for chat formatted finetuning input data. (#884)

milocress committed 2 years ago

Verified ac783541

No symlinks at all for HF download (#908)

jerrychen109 committed 2 years ago

Verified 26349879

Add new GC option (#907)

dakinggg committed 2 years ago

Verified 69bf90ed

mvpatel2000 committed 2 years ago

Verified 1469dcb6

Enable QK Group Norm (#869)

vchiley committed 2 years ago

Verified ea0521ca

Older