ACM India

196 posts

ACM India

ACM India

@ACMIndia

Association for Computing Machinery is the world’s largest society for the Computer Science/IT community, advancing computing as science & profession.

Hyderabad Katılım Ocak 2011
29 Takip Edilen471 Takipçiler
ACM India retweetledi
Dee🎀
Dee🎀@deemureself·
@yehdilmeraa good 4 him, he atlst got tissue, imc stare would hv made me finish deep cleanin d entire room.
English
1
1
5
214
ACM India retweetledi
القحطاني🇸🇦
@ghoremz @bakhsh_reda عندي عمارة وفيها شقق والله ان حرف ال D اغلبهم استلم الشقة منهم بعد الإجار الشقة ما تفرق بينها وبين مكب الزبالة الله يكرمكم وتكررت أكثر من مرة عاد لا اجتمع D ذكر مع D انثى يا لطيف الله يعوض ...والحين كملت في تويتر شوف الأسلوب السوقي من صاحب التغريدة المتشدق ب ال D
العربية
0
1
1
38
ACM India retweetledi
左然老婆
左然老婆@cartooneum·
@bokuluverr yo también tengo una amiga q se burla pero yo tmbn m burlo d ella 🚬🚬 si quieres probar d nuevo y uno que tenga historia de verdad te recomiendo yang ilwoo and i :3
Español
1
2
3
87
ACM India retweetledi
puspi🍒
puspi🍒@puspitasariadr·
capek banget rasanya. tinggal d rumah sendiri tapi berasa numpang.
Indonesia
0
1
1
8
ACM India retweetledi
𓆉𓈒𓏸︎︎︎︎
𓆉𓈒𓏸︎︎︎︎@Naimaherma18·
D I S N E Y+ H o s t a r (จอส่วนตัว) 7 days — 40 baht 30 days — 90 baht รับชมได้ทุกอุปกรณ์ สั่งซื้อที่ไลน์หน้าโปรไฟล์ได้เยย !@~?!~ #หารดิสนีย์พลัส #หารdisneyplus #หารdisney #662
𓆉𓈒𓏸︎︎︎︎ tweet media
ไทย
0
2
2
6
ACM India retweetledi
marti ꕤ
marti ꕤ@twclsen·
mui tiernas ellas muy felices d echarme una manito d gato y yo d ayudarlas
Español
1
2
2
40
ACM India retweetledi
Àyàkì
Àyàkì@AkoniOmoOba·
@waa_sere Yorubaland is so unlucky wit many m£d pple like U low-key sabotaging her but ur m£dness is even in open sum yall Yoruba of this generation as legalize been a traitor. Igbo pple rewriting Yoruba history, stealing our traditional attire, soup n words are pple ure doing this with?
English
0
2
11
302
ACM India retweetledi
code_n_money
code_n_money@code_n_money·
관련해서 흥미로운 방식으로 Soft Prompt라는 것도 있다. Soft Prompt는 모델 가중치는 고정하고, 입력 앞에 붙는 학습 가능한 가상 토큰 임베딩만 최적화한다. P = [p_1, ..., p_m], p_i ∈ R^d 입력 임베딩 X가 있을 때 모델에는 [P; X] 를 넣고, 학습 중에는 P만 업데이트한다. 즉, 사람이 읽는 프롬프트 문장을 만드는 것이 아니라 embedding 공간에서 태스크에 맞는 연속 벡터를 학습한다. 장점 - 학습 파라미터 수: O(md) - 모델 원본을 건드리지 않음 - 태스크별 prompt를 작게 저장하고 빠르게 교체 가능 - 모델이 클수록 성능이 좋아지는 경향 단점 - 학습된 prompt는 사람이 해석하기 어려움 - 모델이 작거나 태스크가 복잡하면 LoRA보다 성능이 떨어질 수 있음 - 입력 길이를 m 토큰만큼 차지함 언제 쓸까? 기본 모델 능력이 충분하고, 같은 모델에 여러 분류·QA·조건부 생성 태스크를 가볍게 붙일 때 적합하다. 행동, 스타일, 도메인 지식을 더 크게 바꿔야 한다면 LoRA가 보통 더 안정적이다.
한국어
0
2
4
41
ACM India retweetledi
KhongkqanShop 📸| ฐานลูกค้า 60k
🌀D I S N E Y+ H o s t a r (จอส่วนตัว) 👀 *ᵎ หาร4 จอไม่ชน 7 days — 55 baht 30 days — 120 baht มีโอกาสจอชน 7 days — 40 baht 30 days — 90 baht 🩰 สั่งซื้อ : lin.ee/wfZYndK ❣️ รีทวิต ลด 5 baht ❣️ #หารดิสนีย์ #หารดิสนีย์พลัสราคาถูก #ดิสนีย์พลัส @ HNZMFY
KhongkqanShop 📸| ฐานลูกค้า 60k tweet media
0
2
2
13
ACM India retweetledi
Noyfb
Noyfb@simonmanto32773·
@LBC @TomSwarbrick1 Its not a change of position. Its -we have no evidence to suggest- and then - we now have evidence to suggest. Its an ongoing situation a d the police are informing us on how its progressing. This guy is blatantly trying to control the narative and undermine the police.
English
0
1
5
121