Concurrent Image Understanding and Generation: Self-Correcting Coupled Markov Jump Processes
Paper ā¢ 2607.13188 ā¢ Published ā¢ 34
MLLM
Anran-MLLM
AI & ML interests
None yet
Recent Activity
upvoted a paper 12 days ago
UniVR: Thinking in Visual Space for Unified Visual Reasoning reacted to theirpost with š about 1 month ago
š Introducing PerceptionDLM ā the first multimodal diffusion LLM for parallel region perception!
Most MLLMs are autoregressive, so captioning N regions costs N sequential passes. PerceptionDLM instead describes ALL masked regions in a single denoising process. š§©
āØ Highlights
ā¢ ā” Up to 3.4Ć faster on dense multi-region captioning, with stable per-image latency
ā¢ š PerceptionDLM-Base beats LLaDA-V on 15/16 multimodal benchmarks (new SOTA among open diffusion VLMs)
ā¢ š New benchmark: ParaDLC-Bench ā jointly evaluates caption quality AND inference efficiency
ā¢ š Code, models & benchmark all open-sourced
š¤ Models
https://huggingface.co/MSALab/PerceptionDLM-Base
https://huggingface.co/MSALab/PerceptionDLM
š Benchmark
https://huggingface.co/datasets/MSALab/ParaDLC-Bench
š Paper: https://huggingface.co/papers/2606.19534
š» Code: https://github.com/MSALab-PKU/PerceptionDLM
Diffusion LLMs aren't just for text ā they unlock efficient, parallel visual perception. šļøāØ
#multimodal #diffusion #VLM #perception
Organizations
None yet
upvoted a paper 11 days ago
upvoted a paper 12 days ago
reacted to their post with šš„ about 1 month ago
Post
3581
š Introducing PerceptionDLM ā the first multimodal diffusion LLM for parallel region perception!
Most MLLMs are autoregressive, so captioning N regions costs N sequential passes. PerceptionDLM instead describes ALL masked regions in a single denoising process. š§©
āØ Highlights
ā¢ ā” Up to 3.4Ć faster on dense multi-region captioning, with stable per-image latency
ā¢ š PerceptionDLM-Base beats LLaDA-V on 15/16 multimodal benchmarks (new SOTA among open diffusion VLMs)
ā¢ š New benchmark: ParaDLC-Bench ā jointly evaluates caption quality AND inference efficiency
ā¢ š Code, models & benchmark all open-sourced
š¤ Models
MSALab/PerceptionDLM-Base
MSALab/PerceptionDLM
š Benchmark
MSALab/ParaDLC-Bench
š Paper: PerceptionDLM: Parallel Region Perception with Multimodal Diffusion Language Models (2606.19534)
š» Code: https://github.com/MSALab-PKU/PerceptionDLM
Diffusion LLMs aren't just for text ā they unlock efficient, parallel visual perception. šļøāØ
#multimodal #diffusion #VLM #perception
Most MLLMs are autoregressive, so captioning N regions costs N sequential passes. PerceptionDLM instead describes ALL masked regions in a single denoising process. š§©
āØ Highlights
ā¢ ā” Up to 3.4Ć faster on dense multi-region captioning, with stable per-image latency
ā¢ š PerceptionDLM-Base beats LLaDA-V on 15/16 multimodal benchmarks (new SOTA among open diffusion VLMs)
ā¢ š New benchmark: ParaDLC-Bench ā jointly evaluates caption quality AND inference efficiency
ā¢ š Code, models & benchmark all open-sourced
š¤ Models
MSALab/PerceptionDLM-Base
MSALab/PerceptionDLM
š Benchmark
MSALab/ParaDLC-Bench
š Paper: PerceptionDLM: Parallel Region Perception with Multimodal Diffusion Language Models (2606.19534)
š» Code: https://github.com/MSALab-PKU/PerceptionDLM
Diffusion LLMs aren't just for text ā they unlock efficient, parallel visual perception. šļøāØ
#multimodal #diffusion #VLM #perception
upvoted a paper about 1 month ago
commented a
paper about 1 month ago
posted an update about 1 month ago
Post
3581
š Introducing PerceptionDLM ā the first multimodal diffusion LLM for parallel region perception!
Most MLLMs are autoregressive, so captioning N regions costs N sequential passes. PerceptionDLM instead describes ALL masked regions in a single denoising process. š§©
āØ Highlights
ā¢ ā” Up to 3.4Ć faster on dense multi-region captioning, with stable per-image latency
ā¢ š PerceptionDLM-Base beats LLaDA-V on 15/16 multimodal benchmarks (new SOTA among open diffusion VLMs)
ā¢ š New benchmark: ParaDLC-Bench ā jointly evaluates caption quality AND inference efficiency
ā¢ š Code, models & benchmark all open-sourced
š¤ Models
MSALab/PerceptionDLM-Base
MSALab/PerceptionDLM
š Benchmark
MSALab/ParaDLC-Bench
š Paper: PerceptionDLM: Parallel Region Perception with Multimodal Diffusion Language Models (2606.19534)
š» Code: https://github.com/MSALab-PKU/PerceptionDLM
Diffusion LLMs aren't just for text ā they unlock efficient, parallel visual perception. šļøāØ
#multimodal #diffusion #VLM #perception
Most MLLMs are autoregressive, so captioning N regions costs N sequential passes. PerceptionDLM instead describes ALL masked regions in a single denoising process. š§©
āØ Highlights
ā¢ ā” Up to 3.4Ć faster on dense multi-region captioning, with stable per-image latency
ā¢ š PerceptionDLM-Base beats LLaDA-V on 15/16 multimodal benchmarks (new SOTA among open diffusion VLMs)
ā¢ š New benchmark: ParaDLC-Bench ā jointly evaluates caption quality AND inference efficiency
ā¢ š Code, models & benchmark all open-sourced
š¤ Models
MSALab/PerceptionDLM-Base
MSALab/PerceptionDLM
š Benchmark
MSALab/ParaDLC-Bench
š Paper: PerceptionDLM: Parallel Region Perception with Multimodal Diffusion Language Models (2606.19534)
š» Code: https://github.com/MSALab-PKU/PerceptionDLM
Diffusion LLMs aren't just for text ā they unlock efficient, parallel visual perception. šļøāØ
#multimodal #diffusion #VLM #perception
liked 2
models about 1 month ago
upvoted a collection about 1 month ago
upvoted 4
papers about 2 months ago
EvoArena: Tracking Memory Evolution for Robust LLM Agents in Dynamic Environments
Paper ā¢ 2606.13681 ā¢ Published ā¢ 143
Skill-3D: Evolving Scene-Aware Skills for Agentic 3D Spatial Reasoning
Paper ā¢ 2606.07436 ā¢ Published ā¢ 25
Watch, Remember, Reason: Human-View Video Understanding with MLLMs
Paper ā¢ 2606.07433 ā¢ Published ā¢ 21
LoomVideo: Unifying Multimodal Inputs into Video Generation and Editing
Paper ā¢ 2606.06042 ā¢ Published ā¢ 24
upvoted a paper 4 months ago
updated a
dataset 4 months ago
upvoted a paper 4 months ago
published a
dataset 5 months ago
upvoted a paper 5 months ago
published a
dataset 5 months ago