za_0x5a

116 posts

za_0x5a

za_0x5a

@zzaneyes

Katılım Mart 2022
153 Takip Edilen15 Takipçiler
za_0x5a
za_0x5a@zzaneyes·
@iam_Dyeus @Lentils80 也许只是因为 Qwen 不像 GPT 和 Gemini 一样会进行二阶推理,而是直接接收提示词,所以你需要精确的提示词才能得到理想结果。
中文
0
0
1
55
Arsh
Arsh@iam_Dyeus·
@Lentils80 tried it here's the results: [1] gpt image 2 [2] qwen image3.0 [3] gemini my opinion: gpt > gemini > qwen
Arsh tweet mediaArsh tweet mediaArsh tweet media
English
3
0
10
2K
Lentils
Lentils@Lentils80·
🚨 Alibaba released Qwen-Image-3.0 today I attached a few examples, images 2nd through 4th are generated by it The text rendering and layout on that academic paper image is very very impressive, and it seems to have improved world knowledge
Lentils tweet mediaLentils tweet mediaLentils tweet mediaLentils tweet media
English
32
80
815
69.3K
za_0x5a
za_0x5a@zzaneyes·
@SenYe10 一年 3000 块,你不说我以为一个月 3000 块呢
中文
0
0
1
15
Sen Ye
Sen Ye@SenYe10·
真是炸裂,现在大学也有各种奇葩 一个拿 4500元最高档助学金的贫困女大学生,就因为每天一杯9.9元的瑞幸,竟然被同学连续蹲了37天取证后实名举报 这位女大学生,有个固定的习惯,每天早八上课前,都会去校内的瑞幸门店买一杯九块九的团购咖啡。 就这么个不起眼的日常,被同班的同学悄悄盯上了。 天天蹲守在取餐的地方数着,连续记录了37天。 最后,把贫困女大学生的取餐记录,直接实名递到了学校资助中心。 理由很充分:天天都能喝上现磨咖啡,这消费状态根本不像贫困户,不符合资助标准,占了真正困难学生的名额。 被举报后,贫困女大学生崩溃大哭:“贫困生就不配喝咖啡吗?” 这个事情,到底谁对谁错? 一个贫困生,天天喝9.9的瑞幸咖啡,一个月将近300块,一年将近3000块,有没有问题? 一个同学,竟然闲着没事干,天天蹲守别人37天。这到底是在维护助学金的公平,还是在满足自己的窥探欲?
Sen Ye tweet media
中文
1.2K
53
797
816.5K
Harold W. Chen (左尔文)
Harold W. Chen (左尔文)@Harold_W_Chen·
@dearemon 我在家也只用一块毛巾。老婆给我准备了3块毛巾,洗脸、洗澡和洗脚各一块,后面我经常混用,还是回归到了只用一块毛巾。都是自己身体,为啥要分开呢。
中文
3
0
4
5.9K
za_0x5a
za_0x5a@zzaneyes·
@louszbd Compared to Tesla, they're real "Artificial" intelligence.
English
0
0
0
15
Lou
Lou@louszbd·
We should learn from Waymo. 1. Find rare cases on the road 2. Reproduce them in simulation & closed course test 3. Deploy within clear limits 4. Compare public road outcomes with human drivers
English
3
1
80
16.1K
Killian
Killian@KilianW146391·
it's in chinese, lool 请创建一个实验性建筑工作室官网「MORPH HABITAT」。网站的核心不是展示现成建筑,而是让一栋未来建筑随着用户滚动,从一颗种子生长、分裂、形成结构,最终变成完整城市。 首屏是一个漂浮在白色雾气空间中的黑色种子。页面构图极简,但必须通过垂直坐标、向下生长的细线和“Scroll to cultivate”明确提示用户继续向下。 滚动过程中: 镜头首先 Zoom In 到种子表面的微观纹理,表面使用程序化 Shader 产生呼吸般的起伏。 种子破裂后,曲线像植物根系一样向外生长,并逐渐转化为建筑梁柱。 使用程序化几何或 Geometry Nodes 风格的视觉逻辑,让结构生成与滚动进度严格绑定。 建筑依次经历线框、粘土模型、材料覆盖、灯光启用和真实环境五个阶段。 每个阶段通过 Scroll Trigger 改变场景色调和排版系统,文字不是浮在网页上,而是嵌入建筑空间。 镜头多次穿过门洞、玻璃和中庭,并在关键节点进行 Zoom In/Zoom Out。 鼠标悬停在不同房间上,可让建筑局部展开,显示剖面、功能和人流路径。 加入昼夜滑块,用户拖动时太阳绕建筑移动,实时改变阴影与室内灯光。 最终继续滚动,单体建筑快速复制、生长为一座未来城市;镜头大幅拉远,从街区一直 Zoom Out 到完整城市鸟瞰图。 结尾提供“重新生长”按钮,点击后所有建筑像沙粒一样坍缩回最初的种子。 视觉需要兼具建筑杂志的高级排版与生成艺术的生命感。请生成完整可运行网站,并重视 3D 模型层级、光影、滚动同步、页面性能和移动端体验。整体语言要英语。
中文
3
6
95
3.3K
za_0x5a
za_0x5a@zzaneyes·
@louszbd 不会被误解的常识性问题。
中文
0
0
0
307
Lou
Lou@louszbd·
What's the accurate way to translate “体感”?
English
42
0
96
19.3K
陈哥Pro
陈哥Pro@chenpapro·
24年有个中国小伙曝光自己被大量女性网暴攻击,原因是他娶了白人超模女性,因为颜值差距太大,老婆的私信里都是一群中国女性在劝分手的,攻击男方的。 结果这位白人女子真听进去了,也有可能是被这位国男给惯坏了,两人不仅分手,女方又迅速找到了新男友,两个共同的经营的账号还被女方带走了。 这位男子称他所有身家都投进这个账号了,虽然能登陆,但没想过毁了,还是各自安好。 而女方澄清是分手了又找到了新男友,不是出轨,并且声称自己花的钱比男方投入的多得多。
陈哥Pro tweet media陈哥Pro tweet media陈哥Pro tweet media陈哥Pro tweet media
中文
530
97
1.8K
983.5K
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
> You can imagine it like… Xi: no you’re not allowed to keep your best model closed anymore, we weren’t succeeding enough. I can. but Xi in his usual manner did nothing, Kimi bros did everything And yes, it is partially ideological round and round they go. 3T now. 10T later
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) tweet mediaTeortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) tweet media
Nathan Lambert@natolambert

Qwen’s biggest models have never been open weight recently. Something is changing. You can imagine it like… Xi: no you’re not allowed to keep your best model closed anymore, we weren’t succeeding enough. A new era of competition for intelligence (if the benchmarks hold up).

English
12
22
417
71K
Air Katakana
Air Katakana@airkatakana·
why are we seeing gpt-5.6 take down a ton of open mathematics problems, but none from fable? is it just that openai is giving everyone more tokens or is that the model is better at mathematics or what
English
103
16
1.6K
219.3K
za_0x5a
za_0x5a@zzaneyes·
@cortiz2894 The first "beautiful" grass I have ever seen, not some thorn-like, embarrasing AI-slop.
English
1
0
0
83
Cortiz
Cortiz@cortiz2894·
🌿 I've just released a breakdown of the grass system I'm currently using in Zephyria I'd love to keep sharing more stylized systems (sky, waterfalls, and more), so if this is the kind of content you'd like to see, a like or subscribe goes a long way! Links in the comments
English
5
8
100
5.2K
Vista Larga Games | ᯅ
Vista Larga Games | ᯅ@VistaLargaGames·
Some people have asked how do I dual develop for vr and flat indie gaming in one project file, while keeping optimization top priority. Fake it to make it. Using 90’s and 2000’s gaming graphical trickery. That’s not water displacement around the rocks. Look closer….
English
6
3
71
7K
za_0x5a
za_0x5a@zzaneyes·
@AIandTechh87 The idea is impressive, but can you make it more beautiful?Just saying, it's aesthetically like some one-shotted ai slop.
English
1
0
1
76
Sam AI
Sam AI@AIandTechh87·
SOMEONE TURNED THEIR TEAM'S TASK TRACKER INTO A 3D ISLAND instead of a boring list of tasks, your team’s work now builds a living 3D island that grows with every winyou assign tasks right in Slack, > just tag who it’s for, add the points, and set the due date >finish a task and you get to place a building on the island >get your work rejected and watch the building collapse into rubble >the rubble stays there forever so everyone can see it >each new sprint starts on a fresh island over time the island fills up with buildings from everything your team actually finished, while the rubble stands as a permanent reminder of what got rejected. it’s open source, so any team can set it up. way more fun than staring at a to-do list all day
English
19
32
157
9.4K
matharium
matharium@matharium·
The dumbest way to solve a maze, and it works every time. Release thousands of particles at the entrance and let each random walk. The first to stumble out has traced a legal path. Retrace its steps and the answer is there. Diffusion, not thinking.
English
158
184
2.1K
719.1K
有的没的
有的没的@Dan4444L·
@TheIanLi 我主观的感觉他们是一个档,掉的速度一样快。
中文
1
0
0
3.4K
有的没的
有的没的@Dan4444L·
我一直都很喜欢 Kimi,也是 Kimi 的粉丝。可是 Kimi 的模型想要正经开始用,必须得订阅 199 的套餐。这个 199 的套餐比 ChatGPT Plus 的基础套餐和 Claude 的基础套餐都要贵,我实在是有点找不到理由,我为什么非要订这个 199 的套餐。
中文
149
4
357
150.4K
𝘊𝘰𝘳𝘳𝘪𝘯𝘦
Now China’s Ministry of Commerce has stepped in. That means this is no longer just a corporate dispute over British Steel. It is a test case for whether Chinese companies can still trust Western investment treaties, property rights, and so-called “market rules.” Britain cannot invite a Chinese company to rescue a collapsing industrial asset — pour money into it, upgrade equipment, protect jobs, pay taxes, and push its green transition — then seize control under the pretext of “national security.” That is a warning to every Chinese company operating overseas: your investment is welcome, your technology is welcome, your jobs and taxes are welcome— until the host country decides your ownership is inconvenient. If London wants to turn legal investment into political confiscation, then Beijing is right to make this a state-level issue. Because if this precedent stands, no Chinese overseas asset is truly safe. The British Museum is a warehouse of stolen goods. Now British steel mills are becoming the 21st-century version. The empire died. The confiscation habit survived.
𝘊𝘰𝘳𝘳𝘪𝘯𝘦 tweet media
𝘊𝘰𝘳𝘳𝘪𝘯𝘦@OopsGuess

Britain still thinks it is 1840. A Chinese private company bought a collapsing British steelmaker, poured money into it, upgraded equipment, protected jobs, paid taxes, and pushed its green transition. Then the UK government suddenly discovered “national security,” seized control, and moved toward nationalisation. Translation: when Chinese capital saves your industry, it is welcome. When the asset becomes valuable again, it becomes “strategic.” Very elegant. Very rules-based. Very empire-afterlife. The British Empire died. But the confiscation habit clearly survived.

English
46
304
1.1K
113.1K
Moiraion
Moiraion@v9034888541033·
@louszbd Any prompt that has latent pre-condition, any prompt that has latent preferences. Any prompt that has implied preferences, like "if user is saying this, that means they made an implied choice about the previous XYZ ambiguities" AI cannot form this CoT.
English
1
0
2
458
Lou
Lou@louszbd·
We're collecting prompts that models still can't handle well. Can be reasoning, coding, SVG, Chinese, any area is fine. They'll be used to evaluate the next gen GLM. If you have one, please reach out!
English
156
50
1.3K
92.8K
za_0x5a
za_0x5a@zzaneyes·
@thdxr What do you think about apps like Typeless?Are they more accurate than local models?
English
0
0
0
39
dax
dax@thdxr·
the thing that killed my editor for me was fast local voice models more than OpenCode from the beginning i was looking for faster ways to interface with the computer. for a while that meant becoming very good with the editor but ultimately voice beats that
English
97
20
1.1K
79.1K
za_0x5a
za_0x5a@zzaneyes·
@theo No, damn, the api still errors when using the agent team in cc.
English
0
0
1
12
Theo - t3.gg
Theo - t3.gg@theo·
What are the best harnesses to use with Kimi K3 right now? It heard it's pretty good at orchestration, where's the best place to see that?
English
207
23
2.6K
515.1K