Channel
@artyfishl
On this record: Growth · Engagement · Reactions · Posts · Citations · Cite this entry
103subscribers
+0 since we began measuring on 3 November 2026
Risers and fallers across the register · movement among entries of Under 1,000.
| Telegram ID | -1004373103380 |
|---|---|
| Type | Channel |
| Username | @artyfishl |
| Created | 8 October 2026 — measured — dated from the channel’s first post |
| First recorded | 3 November 2026 |
| Last confirmed live | 4 November 2026 |
| Measurements held | 2 |
| Confirmed unchanged | 1 time, most recently 4 November 2026 |
| On Telegram | t.me/artyfishl |
| Measured (UTC) | Subscribers | Change |
|---|---|---|
| 4 Nov 2026, 11:36 | 103 | no change |
| 3 Nov 2026, 12:22 | 103 | first reading |
13 posts held, back to 8 October 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 1 page of Telegram’s post history, 20 posts per page.
ERR is average views per post over the last 30 days divided by subscribers, the definition TGStat uses, so this figure is comparable with the one you will see elsewhere. It falls structurally as a channel grows: a high ERR on a small channel and a low one on a large channel describe reach mathematics, not quality. We publish the figure and the sample it came from and pass no verdict on it.
ER is defined industry-wide as (forwards + reactions + comments) ÷ views — note the denominator is views, not subscribers. Telegram’s public web preview carries views and reactions but not forward or comment counts, so the reaction rate above is the reactions term only and is therefore a floor: the true ER for this channel is higher by an amount we have not measured and will not estimate. It is computed over the 4 of 7 measured posts that carry a reaction reading, and over those same posts' views.
| Window | Rolling 30 days · latest post in window 26 October 2026 |
|---|---|
| Posts held | 13 (8 October 2026 – 26 October 2026) |
| Views total | 1,099 |
| Reactions total | 20 |
| Forwards / comments | not exposed by the public surface — not measured, not estimated |
| Readings taken | 3 Nov 2026, 12:22 UTC |
Views are a single reading per post, taken at the time above. A post published in the last day or two is still accumulating views, which pulls the 30-day average down slightly. That is a property of the standard definition rather than a fault in it, so we keep the definition rather than “correcting” the number into something nobody can reproduce.
Precision. Telegram publishes view counts on its public widget in short form — 8.12K, 3.7M — so any reading at or above 1,000 reaches us rounded to three significant figures, and only counts below 1,000 are exact. Averages and rates derived from them are shown to the same precision rather than to the unit: a figure like 3,701,250 would assert digits nobody measured.
Reaction counts are published per emoji and rounded the same way, so a total below 1,000 is exact and a larger one is a sum that may carry a rounded component from each emoji above 1,000. Because it is a sum, it does not look rounded — read a large reaction total as three significant figures per contributing emoji rather than as the figure it prints.
32 reactions across 7 posts, in 6 distinct kinds. The most used accounts for 59.4% of them.
| Reaction | Count | Share | Share, drawn |
|---|---|---|---|
| 👍 | 19 | 59.4% | |
| 🔥 | 5 | 15.6% | |
| ❤ | 4 | 12.5% | |
| 🤩 | 2 | 6.25% | |
| ✍ | 1 | 3.13% | |
| 👏 | 1 | 3.13% |
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 7 of the 13 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 32 reactions in total: the kind of figure the paragraph above means by “a reaction total printed elsewhere on the page”.
Measured over the 13 most recent posts we hold, published 8 October 2026 to 26 October 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.
Telegram’s own ‘edited’ marker: 1 of 7 posts held from the last 30 days (14.3%) carried it when this register last read them. The marker says a post was changed after publishing, not what it said before — this register holds the earlier wording only where an edit fell between two of its own reads (see the edit ledger and how it is measured).
Насколько Jev хорош как реранкер? Qdrant и Hindsight показывают, что Jev реранжирует лучше маленьких кросс-энкодеров. Я решил это тоже проверить, тем более что спрашивали ;) Взял MIRACL-ru из MTEB: 300 случайных запросов, на каждый по 100 кандидатов. Одни и те же кандидаты я дал пяти реранкерам: Jev, BGE-reranker-v2-m3, Qwen3-Reranker, KaLM-Reranker и FRIDA-Decisions. По исходной разметке Jev оказался последним. …
UltraBERT потратил 97% претрейна впустую. У крошечного эмбеддера похожая история Принято считать, что претрейна много не бывает. Больше токенов, лучше модель. Решил проверить это на своем крошечном эмбеддере для поиска. Вышло, что 89% пользы он получил с первых 6% токенов. Повод дала статья про UltraBERT. Энкодеры на 150M и 1.3B там учили до 8.8 трлн токенов, а качество поиска на BEIR встало уже после 275 млрд (ло…
❤3✍1👏1
Posted without readable text
Марио пройден, следующие на очереди — Русы против ящеров
🤩2
Rank1 проигрывал 10 пунктов. Потом авторы прочитали выдачу Знакомься: Rank1, реранкер, который перед оценкой рассуждает. Его учили почти на 400 тысячах цепочек рассуждений DeepSeek-R1 по запросам MS MARCO, и на задачах, где надо подумать, это заметно помогало. А потом авторы прогнали его на TREC DL19, классическом бенчмарке для задач поиска, на котором меряют все. DL19 славится разметкой: на каждый из 43 запросов …
👍4🔥3❤1
Индекс строит 9B, запрос кодирует 0.6B Perplexity выложили pplx-embed-v2-late: две ColBERT-модели, 0.6B и 9B. Каждый токен кодируется вектором на 128 чисел. Самое интересное не размер, а общее пространство. Документы можно один раз проиндексировать тяжелой 9B, а запросы кодировать маленькой. Цена запроса не меняется, потому что документы все равно в оффлайне считать. На текстовых доменных задачах такая схема дает …
👍3🔥2
Обрезать эмбеддинги можно и без MRL Чем больше эмбеддер, тем лучше он обычно ищет. Но с моделью растет и размер вектора, а с ним индекс: миллион документов при 1536 измерениях в float32 занимает ~6 ГБ, при 256 — ~1 ГБ. Самый простой способ сжать индекс — обрезка: берем первые N компонент и нормируем заново. Под это даже есть метод MRL: модель специально учат, чтобы короткие префиксы вектора работали сами по себе. …
Что не видно в анонсе FRIDA-Decisions Если вы пришли со стрима, что такое FRIDA-Decisions, вы уже знаете: энкодер на 823M, который выбирает вариант, ставит оценку и отвечает «да» или «нет» без генерации. Мне в ней интереснее всего три вывода, которые она подтверждает. 1. Даже с окном в 512 токенов можно закрыть много практических задач. Маршрутизация обращений, модерация, проверка релевантности — это короткие текст…
👍4
Две вещи, которые эмбеддер не вытянет одним вектором Эмбеддер сжимает запрос в один вектор и ищет ближайшие тексты. На фактоидных вопросах вроде «кто подписывает акт приемки» это работает отлично. Но есть две ситуации, где одного вектора не хватает, и под каждую мы сделали свое решение. 1. Многошаговые вопросы «В каком году основали компанию, которая купила X?» Чтобы ответить, сначала нужно найти покупателя и тольк…
👍2
Артифишл pinned «Привет, это «Артифишл» Я Артём, техлид ML R&D. Шесть лет учу машины находить нужное в текстах на русском. Если вы собирали RAG или поиск по документам, возможно, вы уже пользовались нашей работой. С командой мы сделали эмбеддер FRIDA и бенчмарк ruMTEB, на…»
Привет, это «Артифишл» Я Артём, техлид ML R&D. Шесть лет учу машины находить нужное в текстах на русском. Если вы собирали RAG или поиск по документам, возможно, вы уже пользовались нашей работой. С командой мы сделали эмбеддер FRIDA и бенчмарк ruMTEB, на котором сравнивают русскоязычные эмбеддеры. Сейчас делаем поискового агента SMITH и FRIDA-Decisions, модель, которая принимает решения по тексту без генерации. Вс…
👍6
Channel photo updated
Showing the 12 most recent of 13 posts we hold for @artyfishl. 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.
Republishes
Channels on the register whose posts this channel has forwarded.
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 1 registered channel — 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.
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 4 November 2026 — this entry's latest reading, not the date you are reading this.
“Артифишл” (@artyfishl), 103 subscribers as measured 4 November 2026. Telegram Register, tgregister.com/channel/artyfishl.
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.