@Dev4YM
@Dev4YM
Software engineer, +4 years experience working on enterprise systems. Full-stack across TS, JS, SQL, Python, C# Flutter. Automation, system design, and using AI
LLMs currently regenerate output token-by-token even when the same content already exists in the conversation. Imagine a decoder that can emit references (EMIT_REF(chunk_id)) instead of re-generating identical tokens. Copy-on-write, but for language. Wouldn't that be cool? #LLM
This account does not have any followers
United States Trends
- 1. FINALLY DID IT N/A
- 2. The SPX6900 N/A
- 3. #FanCashDropPromotion N/A
- 4. Good Friday N/A
- 5. The Paxos Global N/A
- 6. #FridayVibes N/A
- 7. Ryan Wedding N/A
- 8. JS AVOCEAN FAM SIGN N/A
- 9. #สนามอ่านเล่น2026xJimmySea N/A
- 10. 1.3 SOL N/A
- 11. #FursuitFriday N/A
- 12. Happy Friyay N/A
- 13. The REKT N/A
- 14. Autopilot N/A
- 15. Uncle Floyd N/A
- 16. #ราคีTHESTAIN N/A
- 17. Hire Americans N/A
- 18. The ONDO N/A
- 19. Ari Lennox N/A
- 20. Pettine N/A
Loading...
Something went wrong.
Something went wrong.