Qwen/Qwen3.8-27B
Image-Text-to-Text • 28B • Updated • 1.01M • • 11.5k
Denis Golovkin
CoruNethron
AI & ML interests
Imaging system color mapping function estimation
Recent Activity
liked a model 4 days ago
Qwen/Qwen3.8-27B reacted to Jiaqi-hkust's post with 🚀 9 days ago
🧠 Remember-R1: Our fix for MLLMs forgetting the image during long reasoning
We noticed a frustrating problem: when multimodal models reason over long chains, they gradually stop looking at the image—and start hallucinating based on their own text.
So we built Remember‑R1, a simple RL framework that directly supervises visual attention on the original reasoning trajectory—no inference overhead, no proxy tasks.
We use three complementary rewards: coverage, persistence, and focus. They encourage the model to keep attending to relevant visual evidence even in later reasoning steps.
Results across 7 benchmarks and 2 model sizes: better reasoning, and—more importantly—visual attention decays much more slowly during generation.
No extra cost at inference, just cleaner supervision where it counts.
📄 Paper: https://arxiv.org/abs/2608.01314
💻 Code: https://github.com/Ch921-cell/Remember-R1
Happy to answer any questions and receive feedback!
#MultimodalAI #RL #MLLM #CoT #VisualReasoning new activity about 1 month ago
jinaai/jina-embeddings-v5-omni-nano-retrieval-GGUF:Either llama.cpp w. MTMD update or model seems to be brokenOrganizations
None yet
liked a
model 4 days ago
reacted to Jiaqi-hkust's post with 🚀 9 days ago
Post
2393
🧠 Remember-R1: Our fix for MLLMs forgetting the image during long reasoning
We noticed a frustrating problem: when multimodal models reason over long chains, they gradually stop looking at the image—and start hallucinating based on their own text.
So we built Remember‑R1, a simple RL framework that directly supervises visual attention on the original reasoning trajectory—no inference overhead, no proxy tasks.
We use three complementary rewards: coverage, persistence, and focus. They encourage the model to keep attending to relevant visual evidence even in later reasoning steps.
Results across 7 benchmarks and 2 model sizes: better reasoning, and—more importantly—visual attention decays much more slowly during generation.
No extra cost at inference, just cleaner supervision where it counts.
📄 Paper: https://arxiv.org/abs/2608.01314
💻 Code: https://github.com/Ch921-cell/Remember-R1
Happy to answer any questions and receive feedback!
#MultimodalAI #RL #MLLM #CoT #VisualReasoning
We noticed a frustrating problem: when multimodal models reason over long chains, they gradually stop looking at the image—and start hallucinating based on their own text.
So we built Remember‑R1, a simple RL framework that directly supervises visual attention on the original reasoning trajectory—no inference overhead, no proxy tasks.
We use three complementary rewards: coverage, persistence, and focus. They encourage the model to keep attending to relevant visual evidence even in later reasoning steps.
Results across 7 benchmarks and 2 model sizes: better reasoning, and—more importantly—visual attention decays much more slowly during generation.
No extra cost at inference, just cleaner supervision where it counts.
📄 Paper: https://arxiv.org/abs/2608.01314
💻 Code: https://github.com/Ch921-cell/Remember-R1
Happy to answer any questions and receive feedback!
#MultimodalAI #RL #MLLM #CoT #VisualReasoning
Either llama.cpp w. MTMD update or model seems to be broken
1
#1 opened about 1 month ago
by CoruNethron
MTP guide is missing for 3.5 models
#2 opened 3 months ago
by CoruNethron
Missing files
2
#5 opened 4 months ago
by CoruNethron
IQ4_XS fits 64G mac when set max vram to 61440
👍 2
3
#1 opened 10 months ago
by CoruNethron
reacted to marksverdhei's post with 👀 6 months ago
Post
1683
🤔 Many cultures penalize or look down upon self-celebratory behavior. One such example is liking your own post. So why do i do it? Two reasons:
1. I disagree that self-celebratory behavior is inherently bad.
2. On the Huggingface hub, if your post has 0 reactions, it takes TWO whole clicks to react instead of one. So it is actually a UI hack that lowers the bar to engage.
So if you see me reacting to to my own post and thing 'Ugh, this guy is so full of himself' you are only half correct 😆
Now behold as I perform this magic trick called "Exhausting all reaction options for increased visual engagement" so you don't have to click twice to react. You're welcome!
Follow this aspiring 🤗 HF Hub influencer for more half-serious bloat in your feed 😜
1. I disagree that self-celebratory behavior is inherently bad.
2. On the Huggingface hub, if your post has 0 reactions, it takes TWO whole clicks to react instead of one. So it is actually a UI hack that lowers the bar to engage.
So if you see me reacting to to my own post and thing 'Ugh, this guy is so full of himself' you are only half correct 😆
Now behold as I perform this magic trick called "Exhausting all reaction options for increased visual engagement" so you don't have to click twice to react. You're welcome!
Follow this aspiring 🤗 HF Hub influencer for more half-serious bloat in your feed 😜
- 1 reply
Please add usage scenario examples
3
#10 opened 6 months ago
by CoruNethron
upvoted an article 6 months ago
Article
GGML and llama.cpp join HF to ensure the long-term progress of Local AI
- +4
ggerganov, ngxson, allozaur, lysandre, victor, julien-c
• • 510 updated a
model 6 months ago
published a
model 6 months ago
liked a
model 6 months ago
Infinite thinking (after </thinking> tag with updated quants
5
#14 opened 7 months ago
by CoruNethron
Fix model size typos in README.md
2
#1 opened 7 months ago
by CoruNethron
Fix typos in README.md
1
#3 opened 7 months ago
by CoruNethron
liked a
model 7 months ago
Thinking, coder models
3
#1 opened 7 months ago
by 
NIK2703
updated a
model 7 months ago
published a
model 7 months ago
updated a
model 7 months ago