Channel
@soroushsarabiblog
On this record: Growth · Engagement · Reactions · Posts · Citations · Cite this entry
760subscribers
-4 since we began measuring on 6 August 2026
Risers and fallers across the register · movement among entries of Under 1,000.
| Telegram ID | -1003882788292 |
|---|---|
| Type | Channel |
| Username | @soroushsarabiblog |
| Created | Between 1 February 2026 and 15 June 2026— estimated from Telegram’s id allocation, not measured. How this range is calculated. |
| First recorded | 6 August 2026 |
| Last confirmed live | 14 August 2026 |
| Measurements held | 3 |
| Confirmed unchanged | 1 time, most recently 14 August 2026 |
| On Telegram | t.me/soroushsarabiblog |
| Measured (UTC) | Subscribers | Change |
|---|---|---|
| 14 Aug 2026, 10:35 | 760 | -4 |
| 6 Aug 2026, 15:50 | 764 | no change |
| 6 Aug 2026, 09:23 | 764 | first reading |
20 posts held, back to 15 June 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 pageof 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.
| Window | Rolling 30 days · latest post in window 22 July 2026 |
|---|---|
| Posts held | 20 (15 June 2026 – 22 July 2026) |
| Views total | 3,782 |
| Reactions total | 37 |
| Forwards / comments | not exposed by the public surface — not measured, not estimated |
| Readings taken | 6 Aug 2026, 15:50 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.
282 reactions across 20 posts, in 6 distinct kinds. The most used accounts for 91.8% of them.
| Reaction | Count | Share | Share, drawn |
|---|---|---|---|
| ❤ | 259 | 91.8% | |
| 👍 | 9 | 3.19% | |
| 🔥 | 5 | 1.77% | |
| 🤯 | 5 | 1.77% | |
| 👎 | 3 | 1.06% | |
| ⚡ | 1 | 0.355% |
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 20 of the 20 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 282reactions in total: the kind of figure the paragraph above means by “a reaction total printed elsewhere on the page”.
Measured over the 20 most recent posts we hold, published 15 June 2026 to 22 July 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.
محدودیتهای EEG هرگز ناشی از ساختار الکترودها نبوده بلکه مشکل اساسی در مدلسازی بوده است. چهار مقالهی تازه نشان میدهند که اگر سیگنال مغزی را بهجای موجی که باید آستانهگذاری شود، مثل یک تابع جعبهسیاه یا توزیعی قابلیادگیری ببینیم، به نتایجی میرسیم که پیشتر دور از دسترس بودند. در واقع سختافزار هم دیگر مانع نیست. 📖 https://soroushsarabi.com/blog/posts/2026-07-22-eeg-decoded-what-the-model-actually-buys-you/fa …
❤14⚡1
عامل هوشمند برای فیبریلاسیون دهلیزی: چرا مرز مهمتر از مقیاس است مقالهای تازه در npj Digital Medicine عاملی مبتنی بر مدل زبانی برای مدیریت فیبریلاسیون دهلیزی معرفی میکند. ارزش واقعی این کار به خود بیماری برنمیگردد، بلکه به یک تصمیم معماری بازمیگردد: دامنهی مرزدار همراه با دانش سازمانیافته، بهتر از مقیاس خام کار میکند. 📖 https://soroushsarabi.com/blog/posts/2026-07-20-what-a-knowledge-enhanced-clinical-agent-…
❤8👍1
وقتی دستور ساخت مدل بنیادی به دادهی پرنویز نمیخورد دستور استاندارد آموزش مدلهای بنیادی یک فرض پنهان دارد: داده باید متراکم و کمنویز باشد. سیگنال مغزی این فرض را نقض میکند و همینجاست که یک انتخاب معماری متفاوت به کار میآید. 📖 https://soroushsarabi.com/blog/posts/2026-07-17-why-the-masked-autoencoder-recipe-fails-on-brain-signals/fa #applied_ai Soroush Sarabi Blogs
❤13
وقتی مدل ویدیو میسازد، دقیقاً چه چیزی یاد میگیرد؟ مدل برای ساختن فریم بعدی چارهای ندارد جز اینکه ساختار جهان را درون خودش رمزگذاری کند. ادعای GenCeption این است که همین ساختار را میتوان برای ادراک بازیافت و بودجهی پروژههای بینایی را دگرگون کرد. 📖 https://soroushsarabi.com/blog/posts/2026-07-14-what-a-video-generator-knows-about-the-world/fa #ai_research Soroush Sarabi Blogs
❤12
مدل جهان، یک تصمیم زیرساختی است، نه برندهی بنچمارک Qwen-AgentWorld را نباید در جدول رتبهبندی سنجید. ارزش واقعیاش در دو تصمیم مهندسی است: کاهش هزینهی آموزش تقویتی و گرمکردن اولیهی عامل. اینجا میگویم کدام را زودتر برداریم و مرزش کجاست. 📖 https://soroushsarabi.com/blog/posts/2026-07-05-when-the-environment-is-a-model-reading-qwen-agentworld-as-an-infra-decision/fa #applied_ai Soroush Sarabi Blogs
❤16
چرا در یادگیری ماشین بالینی، پیشپردازش مهمتر از طبقهبند است مقالهای تازه دربارهی تشخیص زودهنگام آلزایمر میپرسد کدام مدل بهتر است. اما درس اصلی جای دیگری نهفته است؛ در همان مراحل خستهکنندهای که پیش از رسیدن داده به مدل انجام میشوند. 📖 https://soroushsarabi.com/blog/posts/2026-07-03-the-ensemble-is-not-the-story-what-an-alzheimer-s-pipeline-teaches-about-product/fa #applied_ai Soroush Sarabi Blogs
❤15
agentic یا agentive؟ مرزی که میگوید هوش در وزنهاست یا در پوشش ارکستراسیون (داخل کد به شکل مهندسی الگوریتم) تمایز میان دو نوع عاملیت یک پرسش بازاریابی نیست؛ آزمونی مهندسی است که نشان میدهد توانایی سیستم در وزنهای مدل نشسته یا در کدی که دور آن پیچیدهایم. همین آزمون است که آنچه را میسنجیم و آنچه را صادقانه میتوانیم تحویل دهیم دگرگون میکند. 📖 https://soroushsarabi.com/blog/posts/2026-06-30-agentic-vs-agentiv…
❤14🤯1
وقتی انتشار یک مدل اول باید از فیلتر دولت بگذرد OpenAI مدل GPT-5.6 را به درخواست دولت آمریکا فقط به حدود بیست شرکتِ تأییدشده داد. نکتهی اصلی خودِ مدل نیست؛ سازوکار انتشار است. 📖 https://soroushsarabi.com/blog/posts/2026-06-27-when-the-model-release-goes-through-the-government-first/fa #applied_ai Soroush Sarabi Blogs
❤10🔥3
وقتی عامل یاد میگیرد محیط را در ذهنش بسازد مقالهی Qwen-AgentWorld این پرسش را مطرح میکند که آیا یک مدل زبانی میتواند نقش «مدل جهان» را برای یک عامل نرمافزاری بازی کند؛ شبیهسازی که با متن نوشته میشود و گزارش میدهد آموزش درون این محیطِ خیالی گاهی از آموزش در محیط واقعی هم بهتر است. 📖 https://soroushsarabi.com/blog/posts/2026-06-26-when-the-agent-learns-to-imagine-the-environment/fa #ai_research Soroush Sara…
❤12👍1
Unlimited OCR و هزینهٔ بهخاطر سپردنِ همهچیز مدلِ جدیدِ OCR از Baidu حافظهٔ روبهرشدِ رمزگشا را با یک حافظهٔ کاری ثابت عوض میکند تا دهها صفحه را در یک گذر بخواند. نکتهٔ جذاب خودِ OCR نیست؛ ترفندِ توجهی است که زیرِ آن نشسته. 📖 https://soroushsarabi.com/blog/posts/2026-06-25-unlimited-ocr-and-the-cost-of-remembering-everything/fa #ai_research Soroush Sarabi Blogs
❤13
مدلهای پایهی EEG چه چیزی دربارهی دادههای کوچک شما میگویند سه مقالهی تازه نشان میدهند که دستور پخت آشنای مدلهای پایه روی سیگنال مغزی بهخوبی متن جواب نمیدهد؛ و درس مشترکشان این است که برد از همترازی مدل با داده میآید، نه از مقیاس. 📖 https://soroushsarabi.com/blog/posts/2026-06-23-what-eeg-foundation-models-tell-us-about-your-own-data/fa #applied_ai Soroush Sarabi Blogs
❤11
آیا مدل میتواند خوببودن را در یک حوزه یاد بگیرد و همهجا خوب بماند؟ OpenAI گزارش میدهد که یادگیریِ تقویتی روی مجموعهای کوچک از گفتوگوهای «صفاتِ سودمند» به دهها بنچمارکِ بیربط هم سرایت میکند — و زیرِ فشارِ خصمانه دوام میآورد. ادعای جالب، همین تعمیم است نه تیترِ خبری. 📖 https://soroushsarabi.com/blog/posts/2026-06-21-can-a-model-learn-to-be-good-in-one-domain-and-stay-good-everywhere/fa #ai_research Sorous…
❤9👍4
Showing the 12 most recent of 20 posts we hold for @soroushsarabiblog. 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.
Citation-graph rank — 594,590 of 1,340,412entries in the measured graph. A weighted position computed from the forward and mention edges below — republished posts weigh more than named mentions — and recomputed periodically, over the whole graph. Published only as this ordinal position, never as a score: a position is a fact, and a score printed beside one channel’s name would read as a verdict this register does not make. The two counts beneath stay separate for the same reason mentions are never summed with forwards anywhere else on this page — a named-by count costs nothing to manufacture. The top 100 by this measure, or how it is computed.
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.
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 14 August 2026 — this entry's latest reading, not the date you are reading this.
“سروش سارابی | بلاگ تخصصی” (@soroushsarabiblog), 760 subscribers as measured 14 August 2026. Telegram Register, tgregister.com/channel/soroushsarabiblog.
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.