Skip to content

Commit

Permalink
0.14.1
Browse files Browse the repository at this point in the history
  • Loading branch information
matatonic committed May 19, 2024
1 parent e839f11 commit cf28fb9
Show file tree
Hide file tree
Showing 4 changed files with 8 additions and 0 deletions.
5 changes: 5 additions & 0 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,7 @@ An OpenAI API compatible vision server, it functions like `gpt-4-vision-preview`
- - [X] [XComposer2-7b-4bit](https://huggingface.co/internlm/internlm-xcomposer2-7b-4bit) (not recommended)
- - [X] [XComposer2-VL](https://huggingface.co/internlm/internlm-xcomposer2-vl-7b) [pretrain] (wont gpu split)
- - [X] [XComposer2-VL-4bit](https://huggingface.co/internlm/internlm-xcomposer2-vl-7b-4bit)
- - [X] [XComposer2-VL-1.8B](https://huggingface.co/internlm/internlm-xcomposer2-vl-1_8b)
- [X] [HuggingFaceM4/idefics2](https://huggingface.co/HuggingFaceM4)
- - [X] [idefics2-8b](https://huggingface.co/HuggingFaceM4/idefics2-8b) (main docker only, wont gpu split)
- - [X] [idefics2-8b-AWQ](https://huggingface.co/HuggingFaceM4/idefics2-8b-AWQ) (main docker only, wont gpu split)
Expand Down Expand Up @@ -78,6 +79,10 @@ See: [OpenVLM Leaderboard](https://huggingface.co/spaces/opencompass/open_vlm_le

## Recent updates

Version 0.14.1

- new model support: XComposer2-VL-1.8B (it worked already, no code change)

Version: 0.14.0

- docker-compose.yml: Assume the runtime supports the device (ie. nvidia)
Expand Down
1 change: 1 addition & 0 deletions model_conf_tests.alt.json
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,7 @@
["internlm/internlm-xcomposer2-4khd-7b", "--use-flash-attn", "--device-map", "cuda:0"],
["internlm/internlm-xcomposer2-7b", "--use-flash-attn", "--device-map", "cuda:0"],
["internlm/internlm-xcomposer2-vl-7b", "--use-flash-attn", "--device-map", "cuda:0"],
["internlm/internlm-xcomposer2-vl-1_8b", "--use-flash-attn", "--device-map", "cuda:0"],
["openbmb/MiniCPM-V-2", "--use-flash-attn", "--device-map", "cuda:0"],
["openbmb/MiniCPM-V", "--use-flash-attn", "--device-map", "cuda:0"],
["llava-hf/bakLlava-v1-hf", "--use-flash-attn", "--device-map", "cuda:0"],
Expand Down
1 change: 1 addition & 0 deletions model_conf_tests.json
Original file line number Diff line number Diff line change
Expand Up @@ -19,6 +19,7 @@
["internlm/internlm-xcomposer2-4khd-7b", "--use-flash-attn", "--device-map", "cuda:0"],
["internlm/internlm-xcomposer2-7b", "--use-flash-attn", "--device-map", "cuda:0"],
["internlm/internlm-xcomposer2-vl-7b", "--use-flash-attn", "--device-map", "cuda:0"],
["internlm/internlm-xcomposer2-vl-1_8b", "--use-flash-attn", "--device-map", "cuda:0"],
["openbmb/MiniCPM-V-2", "--use-flash-attn", "--device-map", "cuda:0"],
["openbmb/MiniCPM-V", "--use-flash-attn", "--device-map", "cuda:0"],
["llava-hf/bakLlava-v1-hf", "--use-flash-attn", "--device-map", "cuda:0"],
Expand Down
1 change: 1 addition & 0 deletions vision-alt.sample.env
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,7 @@ HF_HOME=hf_home
#CLI_COMMAND="python vision.py -m internlm/internlm-xcomposer2-4khd-7b --use-flash-attn --device-map cuda:0" # test pass✅, time: 18.8s, mem: 25.7GB, 8/8 tests passed.
#CLI_COMMAND="python vision.py -m internlm/internlm-xcomposer2-7b --use-flash-attn --device-map cuda:0" # test pass✅, time: 16.0s, mem: 19.1GB, 8/8 tests passed.
#CLI_COMMAND="python vision.py -m internlm/internlm-xcomposer2-vl-7b --use-flash-attn --device-map cuda:0" # test pass✅, time: 17.3s, mem: 20.3GB, 8/8 tests passed.
#CLI_COMMAND="python vision.py -m internlm/internlm-xcomposer2-vl-1_8b --use-flash-attn --device-map cuda:0" # test pass✅, time: 5.4s, mem: 7.2GB, 8/8 tests passed.
#CLI_COMMAND="python vision.py -m openbmb/MiniCPM-V-2 --use-flash-attn --device-map cuda:0" # test pass✅, time: 7.0s, mem: 11.6GB, 8/8 tests passed.
#CLI_COMMAND="python vision.py -m openbmb/MiniCPM-V --use-flash-attn --device-map cuda:0" # test pass✅, time: 6.5s, mem: 7.8GB, 8/8 tests passed.
#CLI_COMMAND="python vision.py -m llava-hf/bakLlava-v1-hf --use-flash-attn --device-map cuda:0" # test fail❌, time: 2.0s, mem: 15.9GB, 0/8 tests passed.
Expand Down

0 comments on commit cf28fb9

Please sign in to comment.