係外國撈緊翻譯，試下話俾大家知DeepSeek有幾「得人驚」
美聯經紀鮑威爾
386 回覆
84 Like
414 Dislike
#33 #47
內容污染、fake news，咪就係因為政治囉
咁咪再禁得勁D 囉，
遲D無mac 機用時就唔好講deepseek無想像中咁勁
遲D無mac 機用時就唔好講deepseek無想像中咁勁
deepseek個回應睇到有啲寒
好似真係一個人諗緊野
本身用開gpt 佢俾啲野我都係信5成 因為隨時鳩噏 要自己再factcheck
deepseek依個模式似乎減少左好多鳩噏成份 望落好似成堆字好冗長好晒時間 但係如果俾到個答案出嚟係有思考過 有根據嘅 唔係老作鳩噏 係會可靠好多
好似真係一個人諗緊野
本身用開gpt 佢俾啲野我都係信5成 因為隨時鳩噏 要自己再factcheck
deepseek依個模式似乎減少左好多鳩噏成份 望落好似成堆字好冗長好晒時間 但係如果俾到個答案出嚟係有思考過 有根據嘅 唔係老作鳩噏 係會可靠好多
反唔係問題
我都反中
但會認清事實
老實講我唔會用deepseek
但我唔會否認佢既存在價值
我都反中
但會認清事實
老實講我唔會用deepseek
但我唔會否認佢既存在價值
好簡單咁講, DeepSeek就係咁：
CoT = 教AI點樣一步步諗嘢
Reinforcement Learning = 獎勵AI, 等佢自己學識用CoT
MLA = 慳電慳位
GRPO = 等AI自己知邊個答案好啲
所有呢D「技術」都唔係Deepseek 獨家,但佢地將呢堆技術串埋一齊訓練左一個新既大語言模式....
想睇paper 既:https://github.com/deepseek-ai/DeepSeek-R1/blob/main/DeepSeek_R1.pdf
CoT = 教AI點樣一步步諗嘢
Reinforcement Learning = 獎勵AI, 等佢自己學識用CoT
MLA = 慳電慳位
GRPO = 等AI自己知邊個答案好啲
所有呢D「技術」都唔係Deepseek 獨家,但佢地將呢堆技術串埋一齊訓練左一個新既大語言模式....
想睇paper 既:https://github.com/deepseek-ai/DeepSeek-R1/blob/main/DeepSeek_R1.pdf
結果偷襲完後果自負
除非真係有能力打到落去
到最後睇下邊個底牌大
除非真係有能力打到落去
到最後睇下邊個底牌大
目前黎講r1既答案唔特別出色
唔好得過o4/1o 佢乜都答到勁撚有深度 實際上我有時只係鳩問
如果話chatGPT係太過「穩陣」
咁r1就係太過想「表現自己」
佢諗野個part 反而可以只係個「外殼」
用character.ai 之流 已經可以用LLM營造到有「人格」
重點都係睇佢段野既keywords 同埋成個內容結構
唔好得過o4/1o 佢乜都答到勁撚有深度 實際上我有時只係鳩問
如果話chatGPT係太過「穩陣」
咁r1就係太過想「表現自己」
佢諗野個part 反而可以只係個「外殼」
用character.ai 之流 已經可以用LLM營造到有「人格」
重點都係睇佢段野既keywords 同埋成個內容結構
照計好快知答案
馬斯克朱西佢地仲唔全力研究中國搞咩鬼?
馬斯克朱西佢地仲唔全力研究中國搞咩鬼?
龜兔賽跑
有D反共連登仔覺得兔仔嬴硬, 唔去理解龜仔點解爬得咁貼
有D反共連登仔覺得兔仔嬴硬, 唔去理解龜仔點解爬得咁貼
Trump: DeepSeek's AI should be a 'wakeup call' to US industry
https://www.reuters.com/world/us/trump-deepseeks-ai-should-be-wakeup-call-us-industry-2025-01-27/
MIAMI, Jan 27 (Reuters) - U.S. President Donald Trump said on Monday that Chinese startup DeepSeek's technology should act as spur for American companies and said it was good that companies in China have come up with a cheaper, faster method of artificial intelligence.
"The release of DeepSeek, AI from a Chinese company should be a wakeup call for our industries that we need to be laser-focused on competing to win," Trump said in Florida.
Investors sold technology stocks across the globe on Monday over concerns the emergence of a low-cost Chinese artificial intelligence model would threaten the dominance of the current U.S.-based AI leaders.
"I've been reading about China and some of the companies in China, one in particular coming up with a faster method of AI and much less expensive method, and that's good because you don't have to spend as much money. I view that as a positive, as an asset," Trump said.
"I view that as a positive because you'll be doing that too, so you won't be spending as much, and you'll get the same result, hopefully," he said.
Trump said Chinese leaders had told him the United States had the most brilliant scientists in the world, and he indicated that if Chinese industry could come up with cheaper AI technology, U.S. companies would follow.
"We always have the ideas. We're always first. So I would say that's a positive that could be very much a positive development. So instead of spending billions and billions, you'll spend less, and you'll come up with, hopefully, the same solution," Trump said.
Get weekly news and analysis on U.S. politics and how it matters to the world with the Reuters Politics U.S. newsletter. Sign up here.
https://www.reuters.com/world/us/trump-deepseeks-ai-should-be-wakeup-call-us-industry-2025-01-27/
MIAMI, Jan 27 (Reuters) - U.S. President Donald Trump said on Monday that Chinese startup DeepSeek's technology should act as spur for American companies and said it was good that companies in China have come up with a cheaper, faster method of artificial intelligence.
"The release of DeepSeek, AI from a Chinese company should be a wakeup call for our industries that we need to be laser-focused on competing to win," Trump said in Florida.
Investors sold technology stocks across the globe on Monday over concerns the emergence of a low-cost Chinese artificial intelligence model would threaten the dominance of the current U.S.-based AI leaders.
"I've been reading about China and some of the companies in China, one in particular coming up with a faster method of AI and much less expensive method, and that's good because you don't have to spend as much money. I view that as a positive, as an asset," Trump said.
"I view that as a positive because you'll be doing that too, so you won't be spending as much, and you'll get the same result, hopefully," he said.
Trump said Chinese leaders had told him the United States had the most brilliant scientists in the world, and he indicated that if Chinese industry could come up with cheaper AI technology, U.S. companies would follow.
"We always have the ideas. We're always first. So I would say that's a positive that could be very much a positive development. So instead of spending billions and billions, you'll spend less, and you'll come up with, hopefully, the same solution," Trump said.
Get weekly news and analysis on U.S. politics and how it matters to the world with the Reuters Politics U.S. newsletter. Sign up here.
BTW, 收市之前 用$投咗票 buy the dip, we'll see
我用過Chatgpt嚟寫文，明白樓主既意思，chatgpt出既的確係四平八穩，唔夠深度
用個小人識嘅文先知佢幾勁
一個由中國開發但對西洋話翻譯都咁好
一個由中國開發但對西洋話翻譯都咁好
DeepSeek真係好「得人驚」, 有國家支持點可以唔驚.
如果係咁點解班鬼佬會咁震驚
就係冇人諗到咁樣做，將本身存在既野砌埋一齊做一樣新既野就唔算創新
就係冇人諗到咁樣做，將本身存在既野砌埋一齊做一樣新既野就唔算創新
public domain 嘅 data 真係用得七七八八。
所以依家咪有 D 好似盤古咁，用私人企業嘅生產 data，做返生產有關嘅嘢，避開通用型 AI 賽道。大陸係世界工廠，生產鏈又長又多，呢個簡直係金礦。
所以依家咪有 D 好似盤古咁，用私人企業嘅生產 data，做返生產有關嘅嘢，避開通用型 AI 賽道。大陸係世界工廠，生產鏈又長又多，呢個簡直係金礦。
比負評既人真係知道自己點解比負評?
知己都要知彼，好似美國佬咁懶懶閒，逼到埋身先識醒，仲以為自己贏人地幾十年，驕兵必敗，好似當年俄羅斯睇唔起小日本，理由來來去去就係種族歧視，結果成支艦隊慘敗，直接影響到後來俄國革命
連登仔好多都未明大國博弈唔係街邊班黑社會牛丸同生菜打交搶地盤
今次新冷戰吸收完蘇聯經驗，加埋契哥話合作無上限，一定會比上次更難打
連登仔好多都未明大國博弈唔係街邊班黑社會牛丸同生菜打交搶地盤
今次新冷戰吸收完蘇聯經驗，加埋契哥話合作無上限，一定會比上次更難打感覺個結果都align平時用chatgpt既體會
就係覺得佢hea做
當然可以train佢下次做多少少 俾個完整啲既思路
但都係教一步做多少少咁
呢個deepseek就咁睇係一黎就俾最完整既思路你
係呢個情境做翻譯會覺得有啲多餘
但俾個實用性高既複習task
可能出嚟既結果分別就會明顯
不過open ai 應該都好快會出stage 2 product
就係覺得佢hea做
當然可以train佢下次做多少少 俾個完整啲既思路
但都係教一步做多少少咁
呢個deepseek就咁睇係一黎就俾最完整既思路你
係呢個情境做翻譯會覺得有啲多餘
但俾個實用性高既複習task
可能出嚟既結果分別就會明顯
不過open ai 應該都好快會出stage 2 product
應承我
你應該用Gemini Flash 2.0 Thinking Exp試多次
Reasoning model用番 reasoning model黎比較
Reasoning model用番 reasoning model黎比較
無所謂啦
正負評都改變唔到啲咩 之前個咩江旻穗咪又唔知幾多千正評
我自己撈依行 日夜都對住ai 感受比較深 純粹分享下 岩聽就聽
正負評都改變唔到啲咩 之前個咩江旻穗咪又唔知幾多千正評
我自己撈依行 日夜都對住ai 感受比較深 純粹分享下
岩聽就聽咁你咪自己同自己討論出帖做乜
出帖做乜思考過程歎為觀止
好生動好似真人咁
好撚顛
狂問佢高深問題
長篇大輪講，又會修正咁樣有d唔岩，如果咁會點點點，换個角度再咁咁，勁撚生動
好生動好似真人咁
好撚顛
狂問佢高深問題
長篇大輪講，又會修正咁樣有d唔岩，如果咁會點點點，换個角度再咁咁，勁撚生動