[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 Avatar](https://lunarcrush.com/gi/w:26/cr:twitter::57182201.png) @Kangwook_Lee Kangwook Lee

Kangwook Lee posts on X about token, context window the most. They currently have XXXXX followers and XX posts still getting attention that total XXXXX engagements in the last XX hours.

### Engagements: XXXXX [#](/creator/twitter::57182201/interactions)
![Engagements Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::57182201/c:line/m:interactions.svg)

- X Week XXXXXX +4,178%
- X Month XXXXXX +2,188%
- X Months XXXXXXX +244%
- X Year XXXXXXX +21%

### Mentions: X [#](/creator/twitter::57182201/posts_active)
![Mentions Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::57182201/c:line/m:posts_active.svg)

- X Week XX +200%
- X Month XX +250%
- X Months XX +181%
- X Year XX +455%

### Followers: XXXXX [#](/creator/twitter::57182201/followers)
![Followers Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::57182201/c:line/m:followers.svg)

- X Week XXXXX +8.90%
- X Month XXXXX +9.80%
- X Months XXXXX +22%
- X Year XXXXX +38%

### CreatorRank: XXXXXXX [#](/creator/twitter::57182201/influencer_rank)
![CreatorRank Line Chart](https://lunarcrush.com/gi/w:600/cr:twitter::57182201/c:line/m:influencer_rank.svg)

### Social Influence [#](/creator/twitter::57182201/influence)
---

**Social category influence**
[stocks](/list/stocks)  XXXX%

**Social topic influence**
[token](/topic/token) 16.67%, [context window](/topic/context-window) XXXX%

**Top accounts mentioned or mentioned by**
[@xwanglk](/creator/undefined) [@imhaotian](/creator/undefined) [@nijinjie](/creator/undefined) [@furiosaai](/creator/undefined) [@uwmadison](/creator/undefined) [@msftresearch](/creator/undefined) [@ucberkeley](/creator/undefined) [@seoulnatluni](/creator/undefined) [@kraftonai](/creator/undefined) [@seunghyukoh](/creator/undefined) [@yzeng58](/creator/undefined) [@shuibaiz69721](/creator/undefined) [@yoavgo](/creator/undefined) [@karpathy](/creator/undefined) [@minusgix](/creator/undefined) [@yifeiwang77](/creator/undefined) [@xwang_lk](/creator/undefined)
### Top Social Posts [#](/creator/twitter::57182201/posts)
---
Top posts by engagements in the last XX hours

"Check out KRAFTON AI 😁"  
[X Link](https://x.com/Kangwook_Lee/status/1968799995131228299) [@Kangwook_Lee](/creator/x/Kangwook_Lee) 2025-09-18T22:11Z 3438 followers, 1082 engagements


"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](https://x.com/Kangwook_Lee/status/1978883789523947676) [@Kangwook_Lee](/creator/x/Kangwook_Lee) 2025-10-16T18:00Z 3462 followers, 64.3K engagements


"hf: proj page: github:"  
[X Link](https://x.com/Kangwook_Lee/status/1978908518942949610) [@Kangwook_Lee](/creator/x/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](https://x.com/Kangwook_Lee/status/1978920614422368424) [@Kangwook_Lee](/creator/x/Kangwook_Lee) 2025-10-16T20:26Z 3429 followers, 1690 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](https://x.com/Kangwook_Lee/status/1980664072145444926) [@Kangwook_Lee](/creator/x/Kangwook_Lee) 2025-10-21T15:54Z 3459 followers, 1385 engagements


"In 2026 at some "prestigious" conference we will see "  
[X Link](https://x.com/Kangwook_Lee/status/1980678016482033984) [@Kangwook_Lee](/creator/x/Kangwook_Lee) 2025-10-21T16:50Z 3462 followers, 23.9K 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](https://x.com/Kangwook_Lee/status/1980709454522744902) [@Kangwook_Lee](/creator/x/Kangwook_Lee) 2025-10-21T18:55Z 3462 followers, 84.2K engagements


"The 500x compressor paper The LIFT paper The paper on simultaneous ICL tasks"  
[X Link](https://x.com/Kangwook_Lee/status/1980709769015869488) [@Kangwook_Lee](/creator/x/Kangwook_Lee) 2025-10-21T18:56Z 3462 followers, 4506 engagements