8000 Tune VLM SFT configs by nikg4 · Pull Request #1306 · oumi-ai/oumi · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content
/ oumi < 8000 span class="Label Label--secondary v-align-middle mr-1">Public

Tune VLM SFT configs #1306

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Jan 29, 2025
Merged

Tune VLM SFT configs #1306

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b6e3638
save
nikg4 Jan 29, 2025
3523087
Merge branch 'main' of https://github.com/oumi-ai/oumi into xrdaukar/…
nikg4 Jan 29, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 1 addition & 9 deletions configs/recipes/vision/llama3_2_vision/sft/11b_full/train.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@ data:
transform_num_workers: "auto"
dataset_kwargs:
processor_name: "meta-llama/Llama-3.2-11B-Vision-Instruct"
limit: 4096
# limit: 4096 # Uncomment to limit dataset size!
return_tensors: True
# - dataset_name: "HuggingFaceH4/llava-instruct-mix-vsft"
# split: "train"
Expand All @@ -31,14 +31,6 @@ data:
# transform_num_workers: "auto"
# dataset_kwargs:
# processor_name: "meta-llama/Llama-3.2-11B-Vision-Instruct"
# limit: 4096
# return_tensors: True
# - dataset_name: "coco_captions"
# split: "train"
# transform_num_workers: "auto"
# dataset_kwargs:
# processor_name: "meta-llama/Llama-3.2-11B-Vision-Instruct"
# limit: 100
# return_tensors: True
# - dataset_name: vision_language_jsonl
# dataset_path: "training.jsonl" # See notebook for example how to generate this file
Expand Down
2 changes: 1 addition & 1 deletion configs/recipes/vision/llama3_2_vision/sft/11b_lora/train.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,7 +26,7 @@ data:
transform_num_workers: "auto"
dataset_kwargs:
processor_name: "meta-llama/Llama-3.2-11B-Vision-Instruct"
limit: 4096
# limit: 4096 # Uncomment to limit dataset size!
return_tensors: True

training:
Expand Down
2 changes: 1 addition & 1 deletion configs/recipes/vision/llama3_2_vision/sft/90b_full/train.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,8 +26,8 @@ data:
transform_num_workers: "auto"
dataset_kwargs:
processor_name: "meta-llama/Llama-3.2-90B-Vision-Instruct"
return_tensors: True
limit: 4096
return_tensors: True

training:
output_dir: "output/vlm_finetuned"
Expand Down
4 changes: 2 additions & 2 deletions configs/recipes/vision/llava_7b/sft/train.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,7 @@ model:
data:
train:
collator_name: "vision_language_with_padding"
use_torchdata: False
use_torchdata: True
datasets:
- dataset_name: "merve/vqav2-small"
split: "validation"
Expand All @@ -23,7 +23,7 @@ data:
transform_num_workers: "auto"
dataset_kwargs:
processor_name: "llava-hf/llava-1.5-7b-hf"
limit: 8192
# limit: 8192 # Uncomment to limit dataset size!
return_tensors: True

# Below are examples of other vision SFT datasets
Expand Down
0