Sleeping
6
⚡
Fatih C. Akyon
fcakyon
AI & ML interests
multi-modal learning, video understanding
Recent Activity
new activity
1 day ago
thwri/CogFlorence-2:🚩 Report: Not working
liked
a Space
5 days ago
thwri/CogFlorence-2
new activity
5 days ago
microsoft/Florence-2-large:add_confidence_score
Organizations
fcakyon's activity
🚩 Report: Not working
2
#1 opened 5 days ago
by
fcakyon
liked
a
Space
5 days ago
add_confidence_score
2
#56 opened 5 months ago
by
haipingwu
broken link in the model card
1
#99 opened 5 days ago
by
fcakyon
reacted to
merve's
post with 🤗👀🚀
10 days ago
Post
3198
Apollo is a new family of open-source video language models by Meta, where 3B model outperforms most 7B models and 7B outperforms most 30B models 🧶
✨ the models come in 1.5B https://huggingface.co/Apollo-LMMs/Apollo-1_5B-t32, 3B https://huggingface.co/Apollo-LMMs/Apollo-3B-t32 and 7B https://huggingface.co/Apollo-LMMs/Apollo-7B-t32 with A2.0 license, based on Qwen1.5 & Qwen2
✨ the authors also release a benchmark dataset https://huggingface.co/spaces/Apollo-LMMs/ApolloBench
The paper has a lot of experiments (they trained 84 models!) about what makes the video LMs work ⏯️
Try the demo for best setup here https://huggingface.co/spaces/Apollo-LMMs/Apollo-3B
they evaluate sampling strategies, scaling laws for models and datasets, video representation and more!
> The authors find out that whatever design decision was applied to small models also scale properly when the model and dataset are scaled 📈 scaling dataset has diminishing returns for smaller models
> They evaluate frame sampling strategies, and find that FPS sampling is better than uniform sampling, and they find 8-32 tokens per frame optimal
> They also compare image encoders, they try a variation of models from shape optimized SigLIP to DINOv2
they find google/siglip-so400m-patch14-384 to be most powerful 🔥
> they also compare freezing different parts of models, training all stages with some frozen parts give the best yield
They eventually release three models, where Apollo-3B outperforms most 7B models and Apollo 7B outperforms 30B models 🔥
✨ the models come in 1.5B https://huggingface.co/Apollo-LMMs/Apollo-1_5B-t32, 3B https://huggingface.co/Apollo-LMMs/Apollo-3B-t32 and 7B https://huggingface.co/Apollo-LMMs/Apollo-7B-t32 with A2.0 license, based on Qwen1.5 & Qwen2
✨ the authors also release a benchmark dataset https://huggingface.co/spaces/Apollo-LMMs/ApolloBench
The paper has a lot of experiments (they trained 84 models!) about what makes the video LMs work ⏯️
Try the demo for best setup here https://huggingface.co/spaces/Apollo-LMMs/Apollo-3B
they evaluate sampling strategies, scaling laws for models and datasets, video representation and more!
> The authors find out that whatever design decision was applied to small models also scale properly when the model and dataset are scaled 📈 scaling dataset has diminishing returns for smaller models
> They evaluate frame sampling strategies, and find that FPS sampling is better than uniform sampling, and they find 8-32 tokens per frame optimal
> They also compare image encoders, they try a variation of models from shape optimized SigLIP to DINOv2
they find google/siglip-so400m-patch14-384 to be most powerful 🔥
> they also compare freezing different parts of models, training all stages with some frozen parts give the best yield
They eventually release three models, where Apollo-3B outperforms most 7B models and Apollo 7B outperforms 30B models 🔥
-
- 6 replies
Great share @merve 💯Apollo links in your post not working, giving 404 for me 🤔
liked
3
datasets
about 2 months ago
Inherit from GenerationMixin
#22 opened 2 months ago
by
fcakyon
updated
a
model
2 months ago
Inherit from GenerationMixin
3
#80 opened 2 months ago
by
Link161
updated
4
Spaces
2 months ago
liked
a
Space
2 months ago
updated
a
Space
2 months ago