Wing Lian
d4f6a6b103
fix dockerfile and base builder ( #1795 ) [skip-ci]
2024-07-30 08:34:37 -04:00
Wing Lian
6d4bbb877f
deprecate py 3.9 support, set min pytorch version ( #1343 ) [skip ci]
2024-02-28 12:58:05 -05:00
Wing Lian
8a49309489
upgrade deepspeed to 0.13.1 for mixtral fixes ( #1189 ) [skip ci]
...
* upgrade deepspeed to 0.13.1 for mixtral fixes
* move deepspeed-kernels install to setup.py
2024-01-24 14:26:40 -05:00
Wing Lian
f544ab2bed
don't compile deepspeed or bitsandbytes from source ( #837 )
2023-11-08 19:49:55 -05:00
Fabian Preiß
8056ecd30e
add deepspeed-kernels dependency for deepspeed>=0.12.0 ( #827 )
2023-11-05 07:52:56 -05:00
Wing Lian
aca0398315
apex not needed as amp is part of pytorch ( #696 )
2023-10-07 12:20:45 -04:00
Wing Lian
de87ea68f6
fix multiline for docker ( #694 )
2023-10-06 22:38:15 -04:00
Maxime
923eb91304
tweak: improve base builder for smaller layers ( #500 )
2023-09-22 16:17:50 -04:00
Wing Lian
e85d2eb06b
let MAX_JOBS use the default since we're not resource constrained on our self-hosted runners ( #427 )
2023-09-21 20:36:30 -04:00
Wing Lian
b53e77775b
update dockerfile to not build evoformer since it fails the build ( #607 )
2023-09-19 16:28:29 -04:00
mhenrichsen
cf6654769a
flash attn pip install ( #426 )
...
* flash attn pip
* add packaging
* add packaging to apt get
* install flash attn in dockerfile
* remove unused whls
* add wheel
* clean up pr
fix packaging requirement for ci
upgrade pip for ci
skip build isolation for requiremnents to get flash-attn working
install flash-attn seperately
* install wheel for ci
* no flash-attn for basic cicd
* install flash-attn as pip extras
---------
Co-authored-by: Ubuntu <mgh@mgh-vm.wsyvwcia0jxedeyrchqg425tpb.ax.internal.cloudapp.net >
Co-authored-by: mhenrichsen <some_email@hey.com >
Co-authored-by: Mads Henrichsen <mads@BrbartiendeMads.lan >
Co-authored-by: Wing Lian <wing.lian@gmail.com >
2023-08-18 19:00:27 -04:00
Wing Lian
ffac902c1b
bump flash-attn to 2.0.4 for the base docker image ( #382 )
2023-08-13 17:55:04 -04:00
Wing Lian
2c37bf6c21
Prune cuda117 ( #327 )
...
* drop cuda117/torch 1.13.1 from support, pin flash attention to v2.0.1, rm torchvision/torchaudio install
* gptq base build not needed. add sm 9.0 support
2023-07-26 16:27:49 -04:00
Wing Lian
cdf85fdbd5
pin flash attention 2 to the fix for backwards pass
2023-07-21 08:18:53 -04:00
Wing Lian
9b790d359b
flash attention 2
2023-07-21 08:17:46 -04:00
Wing Lian
b06d3e3645
explicitly pin flash attention 1 to v1.0.9
2023-07-20 01:02:08 -04:00
Wing Lian
71456955f5
pin pydantic so deepspeed isn't broken
2023-07-02 22:26:51 -04:00
Wing Lian
530809fd74
update pip install command for apex
2023-06-28 22:36:28 -04:00
Wing Lian
c43c5c84ff
py310, fix cuda arg in deepspeed
2023-05-30 18:02:34 -04:00
Wing Lian
bbc5bc5791
Merge pull request #108 from OpenAccess-AI-Collective/docker-gptq
...
default to qlora support, make gptq specific image
2023-05-30 15:07:04 -04:00
NanoCode012
392dfd9b07
Lint and format
2023-05-31 02:53:22 +09:00
Wing Lian
48612f8376
cleanup from pr feedback
2023-05-30 09:56:30 -04:00
Wing Lian
e43bcc6c4f
move CUDA_VERSION_BNB arg inside of stage build scope
2023-05-29 13:30:15 -04:00
Wing Lian
00323f0a6f
fix CUDA_VERSION_BNB env var
2023-05-29 08:06:22 -04:00
Wing Lian
21f17cca69
bnb fixes
2023-05-29 00:06:35 -04:00
Wing Lian
809ccebb38
use python setup install, bdist wheel is unreliable in installing extension
2023-05-28 15:49:13 -04:00
Wing Lian
a798ba1659
ensure libbitsandbytes*.so gets included with wheel
2023-05-28 12:28:37 -04:00
Wing Lian
cf37980395
fix missing run coninuation
2023-05-27 15:28:54 -04:00
Wing Lian
312b8d51d6
update docker to compile latest bnb to properly support qlora
2023-05-27 12:36:53 -04:00
Wing Lian
c3d256271e
fix wheel install glob
2023-05-26 10:37:02 -04:00
Wing Lian
1fc9b44e3d
fix wheel blobs in dockerfile
2023-05-26 07:40:11 -04:00
Wing Lian
259262bf42
fix xentropy wheel name typo
2023-05-25 17:25:38 -04:00
Wing Lian
8d6a28953f
fix relative path in flash-attn build:
2023-05-25 12:18:28 -04:00
Wing Lian
f5fa3d131b
fix cd within flash-attn
2023-05-25 11:29:15 -04:00
Wing Lian
1d5ab84486
optionally be able to specify alpaca or chat style prompts
2023-05-20 18:16:22 -04:00
Wing Lian
990bec63e6
docker layer caching, build w axolotl from base build
2023-05-07 17:16:05 -04:00
Wing Lian
0c46806ae2
typo in git repo for pip
2023-05-07 16:00:21 -04:00
Wing Lian
66fa751c18
add huggingface packages and awscli
2023-05-07 11:51:57 -04:00
Wing Lian
21b74397de
fix typo and add apex
2023-05-07 11:48:47 -04:00
Wing Lian
3f11b47488
needs libaio-dev from apt
2023-05-07 11:23:43 -04:00
Wing Lian
ece46b2504
pip install packaging dep
2023-05-07 11:09:03 -04:00
Wing Lian
92d800a394
build dependencies and aws-cli
2023-05-07 11:02:26 -04:00
Wing Lian
2734e3f1a2
build base separately
...
fix arg order for image
fix dockerfile var excaping
move args around
2023-05-07 10:56:12 -04:00
Wing Lian
14ebd2e007
build base too
2023-05-07 09:48:41 -04:00