Threadser.net
數據
關鍵字
功能建議
Following
Threads
Login
串文
串文鏈結
2025-01-27 19:56
Deepseek shifted the Silicon Valley "Scale is all you need" back to "Brain is all you need" Good old research always wins against brute force.
讚
127
回覆
7
轉發
作者
Federico Nicola Aschieri
federiconicolaaschieri
粉絲
串文
85+
讚
回覆
轉發
24小時粉絲增長
無資料
互動率
(讚 + 回覆 + 轉發) / 粉絲數
Infinity%
回覆 (BETA)
最先回覆的內容
發文後
用戶
內容
20 小時內
Vanya Yani
vanya.yani
I have yet to see the claimed low inference cost. The 1b model spits unusable results, 7b model marginally better and horribly slow, anything greater requires me to buy at least $2000 of hardware which I don't have due to layoffs.