QwQ
Collection
Qwen with Questions
ā¢
6 items
ā¢
Updated
ā¢
95
Bojan Kostic
bojan2501
AI & ML interests
None yet
Recent Activity
new activity
16 days ago
Qwen/Qwen3-30B-A3B:MoE question
upvoted
a
collection
2 months ago
QwQ
Organizations
None yet
bojan2501's activity
MoE question
š¤Æ
7
4
#5 opened 16 days ago
by
bojan2501
upvoted
a
collection
2 months ago
reacted to
chansung's
post with š
4 months ago
Post
2033
Simple summary on DeepSeek AI's Janus-Pro: A fresh take on multimodal AI!
It builds on its predecessor, Janus, by tweaking the training methodology rather than the model architecture. The result? Improved performance in understanding and generating multimodal data.
Janus-Pro uses a three-stage training strategy, similar to Janus, but with key modifications:
ā¦ Stage 1 & 2: Focus on separate training for specific objectives, rather than mixing data.
ā¦ Stage 3: Fine-tuning with a careful balance of multimodal data.
Benchmarks show Janus-Pro holds its own against specialized models like TokenFlow XL and MetaMorph, and other multimodal models like SD3 Medium and DALL-E 3.
The main limitation? Low image resolution (384x384). However, this seems like a strategic choice to focus on establishing a solid "recipe" for multimodal models. Future work will likely leverage this recipe and increased computing power to achieve higher resolutions.
It builds on its predecessor, Janus, by tweaking the training methodology rather than the model architecture. The result? Improved performance in understanding and generating multimodal data.
Janus-Pro uses a three-stage training strategy, similar to Janus, but with key modifications:
ā¦ Stage 1 & 2: Focus on separate training for specific objectives, rather than mixing data.
ā¦ Stage 3: Fine-tuning with a careful balance of multimodal data.
Benchmarks show Janus-Pro holds its own against specialized models like TokenFlow XL and MetaMorph, and other multimodal models like SD3 Medium and DALL-E 3.
The main limitation? Low image resolution (384x384). However, this seems like a strategic choice to focus on establishing a solid "recipe" for multimodal models. Future work will likely leverage this recipe and increased computing power to achieve higher resolutions.
commented on
Open-R1: a fully open reproduction of DeepSeek-R1
4 months ago
Interesting read, and it is good that we see more effort into this direction: more optimization and less brute force.
Also wonder what tool did the author use for creating step diagram.
liked
a
model
4 months ago
liked
a
model
11 months ago
liked
a
model
about 1 year ago
