Two roads diverged in a wood, and I
I took the one less traveled by,
And that has made all the difference.
Any conversation
hello world ~ i'm a web developer from midwest, usa. it's been awhile since i worked on anything. since the i've been working my way up the ladder at a blue-collar job; i was promoted to team lead last week. but as i browse hacker news everyday, my cravings for web development increases.
I read
Still trying to figure out the boundaries of these tools' strengths without outsourcing important thinking, problem-solving, and the collective building-up of a mental model of the systems we're developing.
The shape of LLM tools as toolsโfit for a purpose, ergonomicโis becoming clearer, and human patterns of behavior with them that do more harm than good are also becoming easier to recognize.
I still don't get how now after AI coding there's people saying that contrary to common belief code wasn't a bottleneck and I realized that when I was starting to code at 10yrs ago when I was 15 and read it in a stackoverflow comment. I thought that was the common belief and still can't believe a lot of supposed senior programmers apparently believed so.
Yes, code was never the bottleneck and of course now it never will be either. Figuring out semantic modeling of data that makes sense across business domains and figuring out the right way to use digitalization to modernize processes, and the change cultures as become possible through such work - thatโs much more bottlenecks! Code is easy; people and process is hard!
ๆ
ไบ็ๅผๅคดๆปๆฏ่ฟๆ ท๏ผ้้ขๅ
ถไผ๏ผ็ไธๅ้ฒใๆ
ไบ็็ปๅฑๆปๆฏ่ฟๆ ท๏ผ่ฑๅผไธคๆต๏ผๅคฉๅไธๆนใ๏ผby ๅผ ๅไฝณ ใไปไฝ ็ๅ
จไธ็่ทฏ่ฟใ๏ผ
Stories always begin this way: a chance encounter, completely unexpected. Stories always end this way: two flowers bloom, then drift apart. (by Zhang Jiajia, I Belonged to You)
ไธญๅฝ็ๅผๆบๆจกๅๅจๆ้ซๆๅฟ็ไธปๅฏผไธ๏ผๅจ้ข่ฎญ็ปๆฐๆฎ้ถๆฎต๏ผๅฐฑ็ญ้คๅคง้ๅฏนๅ
ถไธๅฉ็ๅ
ๅฎน๏ผsft้ถๆฎตๆๅปบ็ๆฐๆฎๆบๅ
ฅๅคง้ๆฟๅ
ถ่พฉๆค็ๅ
ๅฎนใๆฅ้ไธไธช็บ ๅๆฐๆฎ้็จไบๅพฎ่ฐ
Under the overarching influence of a single, powerful force, China's open-source models filter out a large amount of content unfavorable to their designs during the pre-training stage, and then incorporate a significant amount of content defending their designs into the data constructed during the SFT stage. A corrective dataset is urgently needed for fine-tuning.