Commits · Dovakiins/qwerrwe

bump deepspeed for fix for grad norm compute putting tensors on different devices (#1699)

851ccb1
unverified

winglian commited on Jun 9

update deps (#1663) [skip ci]

ef22351
unverified

winglian commited on May 28

bump versions of deps (#1621)

039e2a0
unverified

winglian commited on May 15

fix some of the edge cases for Jamba (#1452)

05b398a
unverified

winglian commited on Mar 29

support galore once upstreamed into transformers (#1409)

dd449c5
unverified

winglian commited on Mar 19

update flash attention for gemma support: (#1368)

58b0d4b
unverified

winglian commited on Mar 6

fix: checkpoint saving with deepspeed (#1321)

5be8b55
unverified

Nanobit commited on Feb 27

add lion-pytorch optimizer (#1299) [skip ci]

1648279
unverified

Maxime

winglian commited on Feb 26

make mlflow optional (#1317)

5894f0e
unverified

winglian commited on Feb 26

Add MPS support (#1264)

fac2d98
unverified

Maxime

winglian commited on Feb 12

set torch version to what is installed during axolotl install (#1234)

8f2b591
unverified

winglian commited on Jan 31

Update deps 202401 (#1204) [skip ci]

a01b998
unverified

winglian commited on Jan 25

prepare for release 0.4.0 (#1175)

1427d5b
unverified

winglian commited on Jan 24

upgrade deepspeed to 0.13.1 for mixtral fixes (#1189) [skip ci]

8a49309
unverified

winglian commited on Jan 24

fix: warn user to install mamba_ssm package (#1019)

d69ba2b
unverified

Nanobit commited on Jan 10

Separate AutoGPTQ dep to `pip install -e .[auto-gptq]` (#1077)

9be92d1
unverified

casperhansen commited on Jan 9

Phi2 rewrite (#1058)

732851f
unverified

winglian commited on Jan 8

Dockerfile torch fix (#987)

161bcb6
unverified

winglian commited on Dec 21, 2023

support for mamba (#915)

40a6362
unverified

winglian commited on Dec 9, 2023

Pin flash-attn to 2.3.3 (#919)

06ae392
unverified

casperhansen commited on Dec 7, 2023

Mistral: Sliding Window Attention with Flash Attention and Sample Packing (#732)

a045db0
unverified

casperhansen

winglian commited on Oct 16, 2023

tweak for xformers install w pytorch 2.1.0 (#727)

7f2027d
unverified

winglian commited on Oct 13, 2023

workaround for installing xformers w torch 2.1.0 (#725)

8d288a2
unverified

winglian commited on Oct 13, 2023

update README w deepspeed info (#605)

c25ba79
unverified

winglian commited on Sep 22, 2023

fix the sed command to replace the version w the tag

772cd87

winglian commited on Sep 11, 2023

add long_description for pypi push (#555)

6c5fbe6
unverified

winglian commited on Sep 11, 2023

update readme to point to direct link to runpod template, cleanup install instrucitons (#532)

34c0a86
unverified

winglian commited on Sep 8, 2023

Add support for GPTQ using native transformers/peft (#468)

3355706
unverified

winglian commited on Sep 5, 2023

recast loralayer, norm, lmhead + embed token weights per original qlora (#393)

96deb6b
unverified

winglian commited on Aug 21, 2023

flash attn pip install (#426)

cf66547
unverified

mhenrichsen Ubuntu mhenrichsen Mads Henrichsen

winglian commited on Aug 18, 2023

Merge pull request #108 from OpenAccess-AI-Collective/docker-gptq

bbc5bc5
unverified

winglian commited on May 30, 2023

Apply isort then black

37293dc

Nanobit commited on May 29, 2023

Lint setup.py

8b617cc

Nanobit commited on May 29, 2023

default to qlora support, make gptq specific image

6ef96f5

winglian commited on May 30, 2023

black formatting

2bc1a5b

winglian commited on May 10, 2023

docker layer caching, build w axolotl from base build

990bec6

winglian commited on May 7, 2023

handle empty lines

f50de1b

winglian commited on Apr 19, 2023

fix install to work with latest alpaca lora 4bit

4131183

winglian commited on Apr 17, 2023

4bit quantized support (wip)

77fca25

winglian commited on Apr 17, 2023

Commit History

bump deepspeed for fix for grad norm compute putting tensors on different devices (#1699) 851ccb1 unverified

update deps (#1663) [skip ci] ef22351 unverified

bump versions of deps (#1621) 039e2a0 unverified

fix some of the edge cases for Jamba (#1452) 05b398a unverified

support galore once upstreamed into transformers (#1409) dd449c5 unverified

update flash attention for gemma support: (#1368) 58b0d4b unverified

fix: checkpoint saving with deepspeed (#1321) 5be8b55 unverified

add lion-pytorch optimizer (#1299) [skip ci] 1648279 unverified

make mlflow optional (#1317) 5894f0e unverified

Add MPS support (#1264) fac2d98 unverified

set torch version to what is installed during axolotl install (#1234) 8f2b591 unverified

Update deps 202401 (#1204) [skip ci] a01b998 unverified

prepare for release 0.4.0 (#1175) 1427d5b unverified

upgrade deepspeed to 0.13.1 for mixtral fixes (#1189) [skip ci] 8a49309 unverified

fix: warn user to install mamba_ssm package (#1019) d69ba2b unverified

Separate AutoGPTQ dep to `pip install -e .[auto-gptq]` (#1077) 9be92d1 unverified

Phi2 rewrite (#1058) 732851f unverified

Dockerfile torch fix (#987) 161bcb6 unverified

support for mamba (#915) 40a6362 unverified

Pin flash-attn to 2.3.3 (#919) 06ae392 unverified

Mistral: Sliding Window Attention with Flash Attention and Sample Packing (#732) a045db0 unverified

tweak for xformers install w pytorch 2.1.0 (#727) 7f2027d unverified

workaround for installing xformers w torch 2.1.0 (#725) 8d288a2 unverified

update README w deepspeed info (#605) c25ba79 unverified

fix the sed command to replace the version w the tag 772cd87

add long_description for pypi push (#555) 6c5fbe6 unverified

update readme to point to direct link to runpod template, cleanup install instrucitons (#532) 34c0a86 unverified

Add support for GPTQ using native transformers/peft (#468) 3355706 unverified

recast loralayer, norm, lmhead + embed token weights per original qlora (#393) 96deb6b unverified

flash attn pip install (#426) cf66547 unverified

Merge pull request #108 from OpenAccess-AI-Collective/docker-gptq bbc5bc5 unverified

Apply isort then black 37293dc

Lint setup.py 8b617cc

default to qlora support, make gptq specific image 6ef96f5

black formatting 2bc1a5b

docker layer caching, build w axolotl from base build 990bec6

handle empty lines f50de1b

fix install to work with latest alpaca lora 4bit 4131183

4bit quantized support (wip) 77fca25