亦竹 / BH4GZJ (@time_devb) header
亦竹 / BH4GZJ (@time_devb) profile picture

07🏳️‍🌈 天文|飞友|业余无线电|

Shanghai, China
Joined January 2023
Photo by 亦竹 / BH4GZJ (@time_devb)Photo by 亦竹 / BH4GZJ (@time_devb)Photo by 亦竹 / BH4GZJ (@time_devb)Photo by 亦竹 / BH4GZJ (@time_devb)Photo by 亦竹 / BH4GZJ (@time_devb)Photo by 亦竹 / BH4GZJ (@time_devb)Photo by 亦竹 / BH4GZJ (@time_devb)Photo by 亦竹 / BH4GZJ (@time_devb)Photo by 亦竹 / BH4GZJ (@time_devb)Photo by 亦竹 / BH4GZJ (@time_devb)
Replying to @Chisen_Lupus
那很小众了
4
309
Replying to @Kgo_dayo
什么好呢是谁
1
7
267
Replying to @nuclearhechu
生日快乐!
1
68
Replying to @StarCo0611
🉑 如果知识通过性传播的话
1
3
499
Replying to @Kgo_dayo
哦不 摸摸😭
1
134
亦竹 / BH4GZJ retweeted
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math. rikiyat.github.io/blog/weight-sp… github.com/RikiyaT/lm_und… huggingface.co/Qwen/Qwen3.5-0… huggingface.co/RikiyaT/Qwen3.…
2,361
7,275
5,732
55,935
9,879,471
亦竹 / BH4GZJ retweeted
Can we sexualize normalizing
82
3,158
107
45,832
529,257
亦竹 / BH4GZJ retweeted
last minute halloween costume: the Coconut Palm lady
65
3,514
160
46,237
786,682
Replying to @woft123456
也不是不行🥵
1
2
520
stop here to catbois🐱
11
133
2,998
好奇,有没有那种想和我扩列交友但是又不好意思开口的互关(?) 可在此条下评论或偷偷私信我
好奇,有没有那种想和我扩列交友但是又不好意思开口的互关(?) 可在此条下评论或偷偷私信我
16
1
10
77
35,374
Replying to @I1ovesleeping20
omg🥵🥵🥵
1
222
亦竹 / BH4GZJ retweeted
sm暂停 吃点蛋黄酥吧
223
386
99
5,188
186,469