mosaicml/llm-foundry

Pull Requests Commits

add llama eval script

julian-q committed 2 years ago

0cb99e55

supply basename with arg

julian-q committed 2 years ago

a010ece0

use device='gpu'

julian-q committed 2 years ago

a06244fd

julian-q committed 2 years ago

fe1adcdf

add mpt eval script

julian-q committed 2 years ago

82708c7f

julian-q committed 2 years ago

d2c80b14

gpt2-xl eval script

julian-q committed 2 years ago

c0edc6be

Add base wrapper

nik-mosaic committed 2 years ago

17035914

nik-mosaic committed 2 years ago

c312d21b

Add TRT ComposerModel inference wrapper

nik-mosaic committed 2 years ago

6dbfb271

8-bit LION, take 2 (#514)

dblalock committed 2 years ago

Verified 795ab4a0

Revert "Add Programming to Foundry (#441)" (#557)

rishab-partha committed 2 years ago

Verified db2a8d95

Add Programming to Foundry (#441)

rishab-partha committed 2 years ago

Verified 52a3500f

Fix init device in conversion script and add tests (#556)

dakinggg committed 2 years ago

Verified 2f30418e

Add Composer MPT to FasterTransformer Conversion Script (#519)

nik-mosaic committed 2 years ago

Verified a5b39b67

Add python log level in llm foundry eval script (#546)

j316chuck committed 2 years ago

Verified 78b796bd

add missing mixed init (#539)

dakinggg committed 2 years ago

Verified 66700688

remove : (#537)

mvpatel2000 committed 2 years ago

Verified 1a99f097

Refactor build optimizer and peft models to use kwargs syntax (#525)

j316chuck committed 2 years ago

Verified 36cee64b

Adapt composer -> HF conversion script to support all causal lms (#526)

dakinggg committed 2 years ago

Verified 9946fc6f

Add an mcli yaml for running llama2 models (#533)

dakinggg committed 2 years ago

Verified 0b8c81d6

Add import for pop_config from config_utils

j316chuck committed 2 years ago

Verified 05f2d965

Add runtime error in eval.py if yaml config is improperly formatted with extraneous or missing values (#521)

j316chuck committed 2 years ago

Verified 148c0793

Add runtime error in train.py if yaml config is improperly formatted with extraneous or missing values (#506)

j316chuck committed 2 years ago

Verified 6b98ffb8

Monkeypatch flash attention in for llama (#520)

dakinggg committed 2 years ago

Verified aff3eaa2

Pr eval lora (#515)

danbider committed 2 years ago

Verified 7b1e4ead

Fix typo in attention comments (#522)

sashaDoubov committed 2 years ago

Verified 965bd374

update max dur (#523)

mvpatel2000 committed 2 years ago

Verified 58cec8bb

Fix training integration test (#517)

j316chuck committed 2 years ago

Verified 6d582877

Grouped Query Attention + Refactor Attn (#492)

sashaDoubov committed 2 years ago

Verified d2fbc3b4

Older