[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

@Kangwook_Lee "DLLMs seem promising. but parallel generation is not always possible Diffusion-based LLMs can generate many tokens at different positions at once while most autoregressive LLMs generate tokens one by one. This makes diffusion-based LLMs highly attractive when we need fast generation with less compute. A big question is is parallel generation possible without losing modeling accuracy The answer is no. There are fundamental limits on how much parallelism we can achieve. Consider this example: Pick one city uniformly at random from the following four cities: New York New Orleans Mexico City or"
X Link @Kangwook_Lee 2025-10-16T18:00Z 3465 followers, 64.3K engagements

""if you want to be really extra you could in principle bidirectional encode the entire context window just to predict the next single token." The really extra version is here :)"
X Link @Kangwook_Lee 2025-10-21T15:54Z 3462 followers, 1393 engagements

"In 2026 at some "prestigious" conference we will see "
X Link @Kangwook_Lee 2025-10-21T16:50Z 3464 followers, 23.9K engagements

"Check out KRAFTON AI 😁"
X Link @Kangwook_Lee 2025-09-18T22:11Z 3438 followers, 1082 engagements

"hf: proj page: github:"
X Link @Kangwook_Lee 2025-10-16T19:38Z 3456 followers, 1857 engagements

"This is a collaborative work across @FuriosaAI @UWMadison @MSFTResearch @UCBerkeley @SeoulNatlUni and @Krafton_AI with Wonjun Kang Kevin Galim @SeunghyukOh Minjae Lee @yzeng58 @ShuibaiZ69721 Coleman Hooper Yuezhou Hu Hyung Il Koo and Nam Ik Cho 🥰"
X Link @Kangwook_Lee 2025-10-16T20:26Z 3429 followers, 1690 engagements

"A more serious thread on the DeepSeek-OCR hype / serious misinterpretation going on. X. On token reduction via representing text in images researchers from Cambridge have previously shown that 500x prompt token compression is possible (ACL'25 Li Su and Collier). Without using the idea of converting text to images. X. We shouldn't attribute the success of DeepSeek OCR to the power of image representation. At the same time there's nothing fundamentally wrong with text representation with whatever tokenizer. In fact you can do the opposite of what DeepSeek-OCR did i.e. you can represent images"
X Link @Kangwook_Lee 2025-10-21T18:55Z 3465 followers, 84.7K engagements

"The 500x compressor paper The LIFT paper The paper on simultaneous ICL tasks"
X Link @Kangwook_Lee 2025-10-21T18:56Z 3465 followers, 4534 engagements