Motoki Wu 🌊🦈

Resisting the urge to finetune @ Salient. ex-@cresta and other doodlings.

[Motoki Wu 🌊🦈 @plusepsilon]

All this compute power for language models but most of our knowledge are in screenshots.

[Motoki Wu 🌊🦈 @plusepsilon]

2026 will be the year of Claude Code UIs.

[Motoki Wu 🌊🦈 @plusepsilon]

Coding agents are so good now that you should be using it for everything.

Text classification? Coding agent
RAG? Coding agent
Prompt optimization? Coding agent
Recipe for next week's Christmas potluck? Coding agent

[Motoki Wu 🌊🦈 @plusepsilon]

The only late night parties in SF are labeling parties 🥳.

[Motoki Wu 🌊🦈 @plusepsilon]

oh no

[Motoki Wu 🌊🦈 @plusepsilon]

"Owen" is here, the American version of Qwen!

[OpenAI @OpenAI]

Our open models are here.
Both of them.

https://t.co/9tFxefOXcg

[Motoki Wu 🌊🦈 @plusepsilon]

If RL needs pre-training level scaling, are we admitting that RL learns really slow...

[Motoki Wu 🌊🦈 @plusepsilon]

I know RAG is dead, but please chunk your docs for the browser :)

https://t.co/qJAF8FH7wK

[Motoki Wu 🌊🦈 @plusepsilon]

OpenAI chose Windsurf because Ctrl+N doesn't delete your prompt :P

[Motoki Wu 🌊🦈 @plusepsilon]

Agents for recall, workflows for precision.

[Motoki Wu 🌊🦈 @plusepsilon]

Post training making a comeback. MLEs can finally breathe.

[Motoki Wu 🌊🦈 @plusepsilon]

@Thom_Wolf My favorite was τ-bench (https://t.co/v2LKHEz4ox). It evaluates on states, so the trajectory of the conversation and the agent's implementation matter less, and there's no need to rely on LLM evals. It strongly influenced how I structure agent workflows.

[Motoki Wu 🌊🦈 @plusepsilon]

I would say a skillful cook would have a sharper knife.

[Agrim Gupta @agrimgupta92]

"A pair of hands skillfully slicing a ripe tomato on a wooden cutting board"

#veo

[Motoki Wu 🌊🦈 @plusepsilon]

There's no reason to FOMO over a new model when you can wait one month and FOMO over a better model :)

[Motoki Wu 🌊🦈 @plusepsilon]

Can o1 at least help us with Pydantic upgrades.

[Motoki Wu 🌊🦈 @plusepsilon]

Now we can take coffee breaks during test time!

[Motoki Wu 🌊🦈 @plusepsilon]

Some Bad:
* you pretty much have to restart PyCharm if something goes wrong
* it’s the only thing that makes the fans spin up on my M1
* not Cursor 😛

[Motoki Wu 🌊🦈 @plusepsilon]

After 10(?) years, I think I finally found a better alternative to just running plain Jupyter notebooks: PyCharm's notebook integration.

I've been using it off and on for awhile, but it's reached to a point now I can wholeheartedly recommend.

[Motoki Wu 🌊🦈 @plusepsilon]

The Good:
* strong autocomplete
* codebase understanding with jump-to-definition
* jump to previous cursor position
* interactive tables
* stable enough for remote notebooks

Essentially, having all the IDE goodies in a notebook keeps me in the flow.

[Motoki Wu 🌊🦈 @plusepsilon]

I just implemented agentic RAG. ☺️

[Motoki Wu 🌊🦈 @plusepsilon]

@abacaj I've found loss masking to work well, but it does lower the number of tokens that the model sees. So you may have to compensate for that (more epochs / higher batch size).