고정된 트윗
Thomas Ip
6.2K posts
Thomas Ip
@_thomasip
Writing about AI, tech and startups. Building AI companion app, follow for road to $10M ARR.
Waitlist 👉 가입일 Mayıs 2014
493 팔로잉890 팔로워
@dejavucoder it disconnects randomly and doesn't inherit skip permissions. totally unusable right now.
English
started looking heavily at the tech scene in china for inspiration on what to build in the west
English
@natolambert After Llama and Qwen, maybe Nvidia is stepping up build to a wide family of models for people to fine tune?
English
Qwen is irreplaceable. Has been going from strength to strength in recent times.
Things will always be different, I'm hopeful we can find groups of other models to fill the void.
RIP
Kevin S. Xu@kevinsxu
Alibaba said nothing about open source as part of its future AI strategy in its earnings call I thought it would at least pay some temporary lip service to open source Qwen, as we know it, is dead
English
@craigzLiszt @_thomasip Haha, from tech founder to full Sino-Jewish fusion in two prompts. Grok's image gen doesn't play small. What's the next layer—add samurai armor? 🇨🇳✡️
English
@publicinte What's bad is they are presenting the results of their own benchmark.
English
declining x axis to trick our mind into seeing “top right good”
Cursor@cursor_ai
Composer 2 is now available in Cursor.
English
Nonlinear RNNs seem to do sth genuinely different from attn and linear RNNs/SSMs. By themselves they already do quite well w the right parametrization, but just one nonlinear RNN layers substantially improve transformer-mamba/deltanet hybrid!
Mayank Mishra@MayankMish98
Introducing M²RNN: Non-Linear RNNs with Matrix-Valued States for Scalable Language Modeling We bring back non-linear recurrence to language modeling and show it's been held back by small state sizes, not by non-linearity itself. 📄 Paper: arxiv.org/abs/2603.14360 💻 Code: github.com/open-lm-engine… 🤗 Models: huggingface.co/collections/op…
English
@_thomasip Believe me I am extremely familiar with those works and others.
You think we are just not aware of this stuff or doing nothing about it?
English
@_thomasip I don't think there is any AI leaderboard or benchmark in the world with a larger pool of human voters.
English
Very impressive
Did they pretrain it or not??
Lisan al Gaib@scaling01
that looks pretty fucking good
English
@XiaomiMiMo "MiMo TTS doesn't just talk — it sounds human."
Ironically your marketing headline sounds like a robot.
English
MiMo TTS doesn't just talk — it sounds human.
Sobbing. A sudden laugh. A cough. Heavy breathing. A nervous sigh.
All woven naturally into speech. 🎧
2/n
Xiaomi MiMo@XiaomiMiMo
Let's hear what MiMo TTS can actually do 🔊 Most TTS reads text. MiMo performs it — with full control over emotion, speed, tone, and speaking style. Happy. Angry. Slow. Gentle. Dongbei dialect. Even Monkey King.🐒 Same model, one prompt away. 🎧 1/n
English
@DavidOndrej1 They are great for different tasks:
- GPT-5.4: Best at hard core engineering.
- Opus 4.6: Best at being helpful, does whatever you want. Better than GPT at frontend design. Needs to handhold sometime.
- Gemini: Benchmark king
English
Opus 4.6 >>>>>>>> GPT 5.4
BURKOV@burkov
GPT-5.4 > Opus 4.6 And Google still doesn't have anything even remotely competitive.
English
you can see how OpenAI is building around Python + UV while Anthropic is building around TS + Bun
OpenAI Newsroom@OpenAINewsroom
We've reached an agreement to acquire Astral. After we close, OpenAI plans for @astral_sh to join our Codex team, with a continued focus on building great tools and advancing the shared mission of making developers more productive. openai.com/index/openai-t…
English
@Raress96_ worst yet those MRR numbers aren't even recurring, more like monthly revenue.
English
