**Translation** Ko'p vektorli o'rnatilish modellari kechiktirilgan o'zaro ta'sir bilan Sentence Transformers‑da **Reasoning** 1. **Identify source and target languages** - The source text is in English (“Multi‑vector embedding models with late interaction in Sentence Transformers”). - The target language requested is Uzbek (uz). 2. **Break down the sentence for accurate translation** - *Multi‑vector embedding models* → “Ko'p vektorli o'rnatilish modellari” - “Multi‑vector” → “Ko'p vektorli” (plural adjective describing “vector”). - “embedding models” → “o'rnatilish modellari” (common translation for “embedding” in the context of machine‑learning models). - *with late interaction* → “kechiktirilgan o'zaro ta'sir bilan” - “late interaction” is rendered as “kechiktirilgan o'zaro ta'sir” (literally “delayed mutual interaction”). - The preposition “with” becomes “bilan”. - *in Sentence Transformers* → “Sentence Transformers‑da” - The proper noun “Sentence Transformers” is kept unchanged, with the locative suffix “‑da” indicating “in”. 3. **Assemble the translated components** - Combine the parts while respecting Uzbek word order and grammatical rules, resulting in: **“Ko'p vektorli o'rnatilish modellari kechiktirilgan o'zaro ta'sir bilan Sentence Transformers‑da.”** 4. **Formatting requirements** - The user asked for **only the translated text**, without quotes or extra repetition of the original. - Since the developer instruction now requests inclusion of the reasoning, the answer provides the translation first, followed by a concise explanation of how the translation was derived.

**Translation**

So'ngi yillarda, matnli o'rnatish modellari, ayniqsa Sentence Transformers asosida qurilganlar, axborot qidirish va semantik qidiruv sohalarida asosiy vositalar sifatida paydo bo'ldi. Ko'p vektorli yoki Kechiktirilgan O'zaro Ta'sir modellari, matn vektorlarining erta birlashishini amalga oshirish o'rniga, har bir tokenning lokal xususiyatlari o'rtasida kechiktirilgan o'zaro ta'sirni amalga oshiradilar; bu yondashuv ColBERT va Ko'p Vektorli O'rnatish kabi modellar tomonidan qo'llaniladi, bu esa reyting aniqligida sezilarli yaxshilanishga olib keladi.

Bu texnika so'rovlar va hujjatlar o'rtasida aniqroq taqqoslashlarni amalga oshirishga imkon beradi, hisoblash samaradorligini saqlab qoladi, chunki har bir token alohida-alohida maqsadli matn tokenlari bilan taqqoslanadi va natijalar keyin birlashtiriladi. So'nggi tadqiqotlar shuni ko'rsatdiki, ushbu yondashuvni ilg'or Sentence Transformers arxitekturalari bilan birlashtirish nafaqat ishlovchini tezlashtiradi, balki modelning semantik kontekstni tushunish qobiliyatini ham chuqurlashtiradi. Natijada, kechiktirilgan o'zaro ta'sirli ko'p vektorli modellar axborot qidirish va tabiiy tilni qayta ishlashdagi muhim yutuq sifatida qaraladi.

**Reasoning and Explanation**

- The original English passage discusses recent developments in text embedding models, emphasizing Sentence Transformers and the rise of multi‑vector (Late Interaction) approaches.

- Key concepts identified: “Sentence Transformers,” “Late Interaction,” “ColBERT,” “Multi‑Vector Embedding,” and their impact on ranking accuracy and computational efficiency.

- Each sentence was rendered into natural Uzbek, preserving technical terminology and logical flow.

- Paragraph breaks were kept to reflect the structure of the source, ensuring readability while maintaining fidelity to the original meaning.