shunia_huang

> though I found it better than Deepseek V4 Flash previously

Same experience here.

But man, switch to V4.1 now! It is much better.

I don't event need to test it for long run and I believe it's crazy good. I call it "AI era model taste" when I judge the model by it's output without reading the bench scores.

Definitely not Claude, deepseek is too fast, so I bet it's ChatGPT. :P

If you are implying that AI models can do 'well designed UX', I would really love to learn the way of doing it.

Isn't this html + javascript?

Sometimes I try to add comments in a new session and the agent just don't have enough context for it to give a comprehensive sentence with full context on the why, then the agent will just describe what it does.

Human comment is in another level to answer the questions mainly like "why do it like this" for the later collaborators or the forget-ed self, so the important blocks live when it is needed and can be eliminated when it does not.

Everyone does this because there's no other options out in the market to create a "new" model easily enough. I just hope this one is good, and what's better is if they could open weight it.

"read AGENTS.md and follow it" in the first line of CLAUDE.md should work

The flash model will always use an outdated Treafik version that is not compatible with the newer docker engine, I tried to deploy some personal services with Traefik and everytime it uses this wrong version, and then fixes the version issue in the thinking chain.

I was thinking to switch to Caddy but with your experience I'm gonna stay with Traefik and bare with the version issue...

Yes not my experience either.

I've tried or sometimes be stupid to work on bugs/features and ask with almost identical prompts with same modal and harness set, and yes, they generate totally different results.

Sometimes the output is unusable and even with extended guidance it will still drift away from what I was expecting.

Sometimes the output is just one shot and follows almost whatever I want.

I then be used to work like this, if the model and harness set does not work for one time, I just start a new session and do it again. And currently there is one of my task working like this.

It's more complicated in China not just because of people want a more safer environment, but also because of the culture and the social formation built by history.

Chinese were taught to grow in honesty and integrity explicitly for thousand years and it's becoming common consensus when more and more people are educated. When the bar raises and the gov is not given more freedom to the people, it's just nature to let the surveillance cameras grow to each and every inch of the land.

Me personally would chose safety over privacy though.

But they both spent tons of money on data collecting/labeling/generation, how is it bad compared to distillation? I thought their data are much better if they spent that much, and it seems they are stupid because with that much of resources putting in there with merely no output compared to the frontier models.

But Dario said (and maybe more ppl) that Chinese models are just distillation of their model and training data, so I guess your first point is invalid?

WOW the classic yes...but..., where's these statement coming from? I'm Chinese and I don't get the point here at all, you guys seem living in vacuum with biased news, and know much much better than we people living here, that's really confusing everytime I see/read comments lie this.

China now is open freely for almost everyone, come and see by yourslef, if you not living poverty.

The world itslef does not work the propaganda way whether you dislike or hate the gov for whatever reason. And yes you can criticize the gov as you like with the news you read otr the party you pro with, but never assume how the people feel and belives if you are not the people, and if you do then that's the same mindset as Trump or any other war-bringing president the US had, and you know what they did and still doing.

Or think it the other way, what if both A and O actually train with these data, then the enterprises found it(or maybe never), what's the other options? I'm not trusting these LLM companies because a. their model live with human generated data even if they claim to generate data with their own model, it's just nit human data, b. this is business not charity, eventually they live with customers' data, no exception