Baidu killed the MTP module everyone copied from DeepSeek
This is a dev post classified by AGK Intelligence as Other (a launch), kept by the Dev Radar because it carries real work, not commentary.
Baidu killed the MTP module everyone copied from DeepSeek ERNIE 5.5 Flash just runs its own last layer a second time to guess the token after next. They call it ECHO. Trained on only 1 extra token and it still drafts several ahead for speculative decoding, w/ higher acceptance than an MTP module trained the same way With a normal MTP module the future token loss mostly trains the bolted on module. ECHO puts that loss on the real model so the backbone itself learns to look ahead, and Baidu says reasoning and code got clearly better. The leftover ECHO layer is like 10 times smaller than the MT
Posted by Binkg (14.4k followers) 1 h ago · 13 likes · 1.1k views · view the original post on X. Kept by the Dev Radar as Other.
More dev work like this
- A non-coder asked an AI for an e-paper touch panel. Minutes later, it was switching the… — @seeedstudio
- "Locate the refrigerator and stop beside it." 🤖 — @seeedstudio
- The busiest protocol on @gnosischain isn't a DEX. It's @LidoFinance's Data Bus, the… — @envio_indexer
- A 15-second AI video now costs $1.05 at 720p with @GoogleDeepMind's new Gemini Omni 1.2… — @mark_k
- Baidu just open-sourced a new SOTA Base model! — @Sumanth_077
- Good news for all RTX 3090 users. — @ItsmeAjayKV
- 🤖🤖 Open Source AI Voice Recorder: Why Firmware Freedom Matters — @seeedstudio
- 推荐个开源项目 Y2A-Auto，YouTube 视频的本地化发布几乎全自动！ — @bkdgiffug
Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 31.8k posts from 5k X accounts over the last 21 days, 9.5k tools, 12 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-06 14:06 UTC. Full method.