Channel
@aihubvk
On this record: Topic · Growth · Engagement · What this channel posts · Reactions · Posts · Citations · Telegram's recommendations · Cite this entry
2,488subscribers
+85 since we began measuring on 6 August 2026
Risers and fallers across the register · movement among entries of 1,000–3,162.
| Telegram ID | -1001696542591 |
|---|---|
| Type | Channel |
| Username | @aihubvk |
| Created | Between 1 December 2021 and 30 April 2023 — estimated from Telegram’s id allocation, not measured. How this range is calculated. |
| First recorded | 6 August 2026 |
| Last confirmed live | 16 September 2026 |
| Measurements held | 12 |
| Confirmed unchanged | 1 time, most recently 16 September 2026 |
| On Telegram | t.me/aihubvk |
Technology — a classification, not a measurement. An on-box language model (Qwen3.6-35B-A3B-FP8, prompt version 1) read this channel’s own recent posts on 15 September 2026 and assigned it the closest of 31 fixed categories, at 100% confidence. This is a model’s judgement about what the channel is likely to be about, not a fact this register measured the way a subscriber count or a view count is measured — it can be revised on a later pass, and it carries no weight anywhere else on this page. How this classification works, and why it has no browse page of its own yet.
| Measured (UTC) | Subscribers | Change |
|---|---|---|
| 16 Sept 2026, 01:36 | 2,488 | +1 |
| 12 Sept 2026, 04:16 | 2,487 | +11 |
| 31 Aug 2026, 02:24 | 2,476 | +3 |
| 28 Aug 2026, 00:04 | 2,473 | -3 |
| 25 Aug 2026, 05:22 | 2,476 | +4 |
| 22 Aug 2026, 11:28 | 2,472 | +41 |
| 18 Aug 2026, 14:59 | 2,431 | +8 |
| 15 Aug 2026, 15:04 | 2,423 | +10 |
| 12 Aug 2026, 00:09 | 2,413 | +7 |
| 9 Aug 2026, 07:52 | 2,406 | +3 |
| 6 Aug 2026, 05:50 | 2,403 | no change |
| 6 Aug 2026, 05:49 | 2,403 | first reading |
11 posts held, back to 14 July 2026 — the reader has not yet reached the start of this channel’s public history, so older posts may sit further back, unread. Read across 2 pages of Telegram’s post history, 20 posts per page.
Nothing published in the last 30 days. ERR and ER are rolling 30-day measures, so there is nothing to compute — we hold 11 posts for this entry, the most recent from 6 August 2026. An engagement rate over an empty window would be a number about nothing.
Measured directly from 1 video with a duration reading, out of the posts we hold for this channel — not this channel’s whole posting history, only the sample this register has actually read. An exact reading to the second, taken from the post itself rather than from Telegram’s own rounded chrome, so it carries no ≈ mark.
201 reactions across 9 posts, in 8 distinct kinds. The most used accounts for 32.8% of them.
| Reaction | Count | Share | Share, drawn |
|---|---|---|---|
| 🔥 | 66 | 32.8% | |
| ❤ | 53 | 26.4% | |
| 👏 | 22 | 10.9% | |
| 👍 | 21 | 10.4% | |
| 💅 | 21 | 10.4% | |
| 🎉 | 12 | 5.97% | |
| 🤩 | 4 | 1.99% | |
| 😁 | 2 | 0.995% |
No sentiment is inferred, and none should be read in. This table is ordered by count and by nothing else. Emoji do not carry stable meaning across languages or communities — 🙏 is thanks in one channel and mourning in another — so we publish which ones were pressed and how often, and pass no judgement on what an audience meant by them.
Precision. Telegram publishes reaction counts per emoji and short-forms each one — 4.34K, 1.2M — so any single kind at or above 1,000 reaches us at three significant figures, and only counts below 1,000 are exact. The shares above are ratios of those figures and carry the same error. This is also why the total here can differ slightly from a reaction total printed elsewhere on the page: both are sums of the same rounded parts, taken over samples with different edges.
Coverage. Reactions were read on 11 of the 11 sampled posts in this sample. Summed by Telegram’s own count on each post — not by adding up the per-emoji breakdown above — those same posts carry 233 reactions in total: the kind of figure the paragraph above means by “a reaction total printed elsewhere on the page”.
Measured over the 11 most recent posts we hold, published 14 July 2026 to 6 August 2026, using the newest reading held for each. Telegram Stars are excluded: they are a payment, not a reaction, and they have their own section.
Легковесная full-band модель для шумоподавления и дереверберации Большинство моделей speech enhancement решают только шумоподавление и работают с аудио 16 кГц. Исследователи НИУ ВШЭ и VK адаптировали архитектуру FSPEN для одновременного шумоподавления и дереверберации full-band аудио (48 кГц) — 95 тыс. параметров, RTF 0.11, качество шумоподавления на уровне ground truth. В карточках кратко разбираем, что изменили в…
👍4❤3🔥2
🟣Alibaba: 3 релиза за 2 недели 3 августа компания анонсировала флагманскую Qwen3.8-Max, MoE на 2.4 трлн параметров (~95 млрд активных), контекст 1M токенов, модель позиционируют как «вторую после Fable 5». 27 июля тихо, без техотчёта и бенчмарков, вышла бюджетная vision-language Qwen3.7 Flash ($ 0,03 — $ 0,13 за 1M токенов, контекст 1M). 21 июля представили Qwen-Image-3.0, генерация изображений с промптом до 4 500 т…
👍2❤1🔥1
Наш коллега, ведущий исследователь AI VK Research Владимир Байкалов, вернулся из Мельбурна с конференции SIGIR 2026 и подготовил обзор интересных работ, по которым виден главный тренд 2026 года: генеративные end-to-end модели не отменили каскадные пайплайны, а нашли своё место на стадии отбора. Ранжирование всё ещё делает отдельная модель. Помимо этого, видны ещё три тенденции: рекомендательные системы уходят в мульт…
💅10❤8👏5🔥3👍2
Исследователи AI VK Research представили большой датасет VK-LSVD на The Web Conference 2026 — одной из ключевых международных конференций в сфере web science. О датасете можно почитать подробнее в нашем канале и на Хабре. Мы подготовили обзор нескольких интересных статей, отмеченных программным комитетом конференции. 🏆 Best paper award From Retrieval to Generation: Unifying External and Parametric Knowledge for Me…
🔥8🎉7❤2👍1👏1
⚡️ AI VK Research на SIGIR Прямо сейчас в Мельбурне проходит SIGIR 2026 — одна из ключевых конференций в области информационного поиска, ML и RecSys. Ведущий исследователь AI VK Владимир Байкалов вместе с коллегами из ИТМО представляет работу Mitigating Collaborative Semantic ID Staleness in Generative Retrieval. Подробнее о статье — здесь. Скоро мы разберем самые интересные исследования с SIGIR 2026. #aivkhub #re…
❤14🔥8👏5💅5🎉1
Generative Recommender Systems Классические рекомендательные системы работают по двухстадийной схеме: retrieval сужает множество кандидатов, ranking ранжирует их по релевантности. Генеративные рекомендательные системы (Generative RecSys) переосмысливают эту задачу: вместо скоринга готовых кандидатов модель генерирует представление следующего релевантного объекта. Semantic ID — основа подхода Обычно идентификаторы …
🔥6❤4👍3
Несмотря на взрывной рост рекомендательных трансформеров, генеративных рекомендаций и так далее, классические методы на основе матричных факторизаций всё ещё применяются в рекомендательных системах. Преимущество современных подходов в том, что они позволяют работать с пользователем в долгосрочной перспективе и учитывать её при построении рекомендаций. Так делают, например, PinnerFormer, OneRec. При этом матричные фа…
👍9🔥9❤7💅1😁1
➡️ Смотрим, как прошёл RecSys Meetup. 🟣Рассказали об исследованиях и научных разработках 🟣Представили широкой аудитории команду AI VK Research 🟣Рассказали, как работает рекомендательная платформа Discovery AI 📹 Подробности и атмосфера мероприятия — в репортаже. Включите уведомления, чтобы не пропустить анонсы следующих митапов🤘 #aivkhub #recsys #meetup
🔥21👏10🤩4🎉2😁1
ICML 2026: как бигтех переосмысливает рекомендательные системы Продолжаем обзор конференции ICML 2026, на которой исследователь Александр Тараканов (AI VK) и коллеги из НИУ ВШЭ представили фреймворк PIEFS. Мы подробно писали об этой работе. Мы уже рассказывали об общем тренде в развитии AI-систем, заметном по основному пулу презентаций на ICML 2026. Сегодня на примере трёх работ исследователей из Kuaishou, Meta* и …
❤14🔥8💅5🎉2👏1
✨ С 6 по 11 июля в Сеуле прошла ICML 2026 — крупная международная конференция по машинному обучению. В программе были доклады, воркшопы и индустриальная выставка по самым разным направлениям — от теории обучения и оптимизации до мультимодальных моделей, AI for Science, рекомендательных систем и LLM. На воркшопе исследователь Александр Тараканов из AI VK и коллеги из НИУ ВШЭ представили фреймворк PIEFS для генерации …
Методы адаптации LLM За несколько лет методы адаптации LLM прошли путь от прямого обновления весов до надстройки над замороженной моделью. Подходов много, и все они отличаются по стоимости, сложности внедрения и качеству результата. В карточках рассмотрим семь групп методов — от дообучения до агентов. #aivkhub #llm
Showing the 11 most recent of 11 posts we hold for @aihubvk. View and reaction counts are the latest single reading for each post, not a live figure, and a recent post is still accumulating both. A view count marked ≈ was rounded by Telegram before we ever saw it — t.me prints views in full below 1,000 and to three significant figures above, so ≈1,200,000 means somewhere between 1,150,000 and 1,249,999. Unmarked counts are exact. Text is reproduced from the public post preview and truncated for length.
Republished by
Channels on the register that have forwarded this channel's posts into their own feed.
Built only from forwarded posts we have actually read, on both sides. Coverage is early and deliberately incomplete: a missing link means we have not read the post that would prove it, never that the relationship does not exist. Counts are distinct forwarded posts observed, so they only ever go up as we read more.
Named by 17 registered channels — every channel on the register whose own posts have named this one, by its current username or any other username it currently holds, merged from two separately captured readings of the same fact so a namer caught by only one of them is not missed and a namer both caught is not counted twice. A username this channel has since dropped is not matched — that handle may belong to someone else now, and crediting today’s namer to yesterday’s owner would misattribute it.
Named by
Channels on the register whose posts name this channel's handle.
Names
Channels on the register whose handles appear in this channel's posts.
A mention is a weaker signal than a forward and is counted separately for that reason — naming a channel is not republishing it, and a handle in a post body is easy to place deliberately. The post counts beside each row below are distinct posts in which the handle appeared, from posts we have read on both sides — the “Named by N registered channels” figure above is a different count, of distinct NAMING CHANNELS rather than posts, and is not the sum of the rows under it.
The reverse of the list above, and a different kind of signal. This does not require this channel to have ever been asked about directly — each row below is a channel we DID ask Telegram about, whose Telegram-generated list happened to include this one. A channel can appear here with an empty list above it, because being named by someone else’s query is independent of having been queried itself.
This channel appears in 6 seed channels' Telegram-generated recommendation lists in total. Each is Telegram’s list for THAT channel, not this one — see how this is measured.
A live page changes as we take new readings, so a citation should name the measurement it is based on, not just the URL. The line below cites the subscriber count as measured 16 September 2026 — this entry's latest reading, not the date you are reading this.
“AI VK Hub” (@aihubvk), 2,488 subscribers as measured 16 September 2026. Telegram Register, tgregister.com/channel/aihubvk.
Full measurement history, CC BY 4.0. Every reading this register holds for this entry, not just the latest one, as a dated, downloadable record: CSV · JSON. Free to use with attribution to tgregister.com. Each file carries its own generation timestamp, which is the figure to cite for exactly when the data was retrieved.