tencent/Hy4-preview
Text Generation ā¢ 780B ā¢ Updated ā¢ 6.44k ā¢ 445
Jiaqi Tang PRO
Jiaqi-hkust
AI & ML interests
MLLM, LLM, Agentic AI
Recent Activity
liked a model 10 days ago
tencent/Hy4-preview reacted to theirpost with š¤ 27 days ago
š§ 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 imageāand start hallucinating based on their own text.
So we built RememberāR1, a simple RL framework that directly supervises visual attention on the original reasoning trajectoryāno 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, andāmore importantlyāvisual attention decays much more slowly during generation.
No extra cost at inference, just cleaner supervision where it counts.
š Paper: https://arxiv.org/abs/2608.01314
š» Code: https://github.com/Ch921-cell/Remember-R1
Happy to answer any questions and receive feedback!
#MultimodalAI #RL #MLLM #CoT #VisualReasoning reacted to theirpost with ā 27 days ago
š§ 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 imageāand start hallucinating based on their own text.
So we built RememberāR1, a simple RL framework that directly supervises visual attention on the original reasoning trajectoryāno 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, andāmore importantlyāvisual attention decays much more slowly during generation.
No extra cost at inference, just cleaner supervision where it counts.
š Paper: https://arxiv.org/abs/2608.01314
š» Code: https://github.com/Ch921-cell/Remember-R1
Happy to answer any questions and receive feedback!
#MultimodalAI #RL #MLLM #CoT #VisualReasoningOrganizations
liked a
model 10 days ago
reacted to their post with š¤āššš„ 27 days ago
Post
2399
š§ 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 imageāand start hallucinating based on their own text.
So we built RememberāR1, a simple RL framework that directly supervises visual attention on the original reasoning trajectoryāno 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, andāmore importantlyāvisual attention decays much more slowly during generation.
No extra cost at inference, just cleaner supervision where it counts.
š Paper: https://arxiv.org/abs/2608.01314
š» 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 imageāand start hallucinating based on their own text.
So we built RememberāR1, a simple RL framework that directly supervises visual attention on the original reasoning trajectoryāno 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, andāmore importantlyāvisual attention decays much more slowly during generation.
No extra cost at inference, just cleaner supervision where it counts.
š Paper: https://arxiv.org/abs/2608.01314
š» Code: https://github.com/Ch921-cell/Remember-R1
Happy to answer any questions and receive feedback!
#MultimodalAI #RL #MLLM #CoT #VisualReasoning
posted an update 27 days ago
Post
2399
š§ 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 imageāand start hallucinating based on their own text.
So we built RememberāR1, a simple RL framework that directly supervises visual attention on the original reasoning trajectoryāno 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, andāmore importantlyāvisual attention decays much more slowly during generation.
No extra cost at inference, just cleaner supervision where it counts.
š Paper: https://arxiv.org/abs/2608.01314
š» 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 imageāand start hallucinating based on their own text.
So we built RememberāR1, a simple RL framework that directly supervises visual attention on the original reasoning trajectoryāno 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, andāmore importantlyāvisual attention decays much more slowly during generation.
No extra cost at inference, just cleaner supervision where it counts.
š Paper: https://arxiv.org/abs/2608.01314
š» Code: https://github.com/Ch921-cell/Remember-R1
Happy to answer any questions and receive feedback!
#MultimodalAI #RL #MLLM #CoT #VisualReasoning
liked a
dataset 28 days ago
liked 2
models about 1 month ago
liked a
dataset about 1 month ago
updated a
Space about 1 month ago
upvoted a paper about 1 month ago
liked a
dataset about 1 month ago
reacted to their post with āš¤šššš„ about 1 month ago
Post
3094
IQA-T1: EvidenceāBased Image Quality Assessment with MLLMs
Most MLLMs are blind to lowālevel degradationsānoise, blur, compression artifacts look the same as clean images in their internal representations. That leads to quality scores based on semantic āgut feelingā rather than real perceptual evidence.
IQA-T1 changes that. We equip the model with a toolbox of 15 perceptual tools (noise residual maps, Fourier spectra, gradient maps, etc.) that generate structured visual evidence on demand. The model learns how to use tools via supervised fineātuning on our QāTool dataset (11k evidenceāgrounded reasoning chains), and when to call them via GRPO reinforcement learning that balances accuracy, tool count, and redundancy.
The result: SOTA performance across 7 benchmarks (avg PLCC 0.795), using only 2.34 tools per image on average. Every predicted score is now interpretable and backed by hard visual evidence.
All code, weights, dataset, and demo are open. Check them out and give it a spin!
š arxiv.org/abs/2607.12375v1
š» github.com/zibuyu-02/IQA-T1
š¤ model/data: huggingface.co/zibuyu-02/IQA-T1
š® demo: huggingface.co/spaces/Jiaqi-hkust/IQA-T1
Most MLLMs are blind to lowālevel degradationsānoise, blur, compression artifacts look the same as clean images in their internal representations. That leads to quality scores based on semantic āgut feelingā rather than real perceptual evidence.
IQA-T1 changes that. We equip the model with a toolbox of 15 perceptual tools (noise residual maps, Fourier spectra, gradient maps, etc.) that generate structured visual evidence on demand. The model learns how to use tools via supervised fineātuning on our QāTool dataset (11k evidenceāgrounded reasoning chains), and when to call them via GRPO reinforcement learning that balances accuracy, tool count, and redundancy.
The result: SOTA performance across 7 benchmarks (avg PLCC 0.795), using only 2.34 tools per image on average. Every predicted score is now interpretable and backed by hard visual evidence.
All code, weights, dataset, and demo are open. Check them out and give it a spin!
š arxiv.org/abs/2607.12375v1
š» github.com/zibuyu-02/IQA-T1
š¤ model/data: huggingface.co/zibuyu-02/IQA-T1
š® demo: huggingface.co/spaces/Jiaqi-hkust/IQA-T1