Telegram RegisterThe public register of Telegram
Telegram profile photo for Generative AI Insights | 生成式 AI 洞察

Channel

Generative AI Insights | 生成式 AI 洞察

@gainotes

On this record: Topic · Growth · Engagement · What this channel posts · Reactions · Posts · Cite this entry

3,245subscribers

-123 since we began measuring on 7 August 2026

Risers and fallers across the register · movement among entries of 3,162–10,000.

Register entry

Telegram ID-1001646913580
TypeChannel
Username@gainotes
CreatedBetween 1 December 2021 and 15 March 2023 — estimated from Telegram’s id allocation, not measured. How this range is calculated.
First recorded7 August 2026
Last confirmed live9 November 2026
Measurements held25
Confirmed unchanged1 time, most recently 9 November 2026
On Telegramt.me/gainotes

Topic

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 14 September 2026 and assigned it the closest of 31 fixed categories, at 99% 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 the other topics.

Growth

3,2453,3683,306.57 August 2026 — 3,368 subscribers7 August 2026 — 3,368 subscribers10 August 2026 — 3,363 subscribers13 August 2026 — 3,357 subscribers16 August 2026 — 3,352 subscribers19 August 2026 — 3,346 subscribers23 August 2026 — 3,341 subscribers26 August 2026 — 3,340 subscribers30 August 2026 — 3,334 subscribers5 September 2026 — 3,332 subscribers11 September 2026 — 3,322 subscribers14 September 2026 — 3,319 subscribers18 September 2026 — 3,315 subscribers27 September 2026 — 3,310 subscribers5 October 2026 — 3,303 subscribers12 October 2026 — 3,293 subscribers15 October 2026 — 3,289 subscribers18 October 2026 — 3,286 subscribers21 October 2026 — 3,281 subscribers24 October 2026 — 3,276 subscribers28 October 2026 — 3,265 subscribers31 October 2026 — 3,259 subscribers3 November 2026 — 3,257 subscribers5 November 2026 — 3,252 subscribers9 November 2026 — 3,245 subscribers7 August 20269 November 2026
25 measurements spanning 94 days, net -123. Dots are measurements; the straight line between them is drawn to join them, not to claim we know the path taken in between — snapshots are recorded only when a count changes, so gaps mean “no change observed”, never “interpolated”. The vertical axis spans 3,227–3,386 and does not start at zero.
Measurement log — every subscribers count we have recorded, most recent 20 of 25
Measured (UTC)SubscribersChange
9 Nov 2026, 02:173,245-7
5 Nov 2026, 22:593,252-5
3 Nov 2026, 09:203,257-2
31 Oct 2026, 13:063,259-6
28 Oct 2026, 04:183,265-11
24 Oct 2026, 23:483,276-5
21 Oct 2026, 16:273,281-5
18 Oct 2026, 19:063,286-3
15 Oct 2026, 18:263,289-4
12 Oct 2026, 10:183,293-10
5 Oct 2026, 13:163,303-7
27 Sept 2026, 17:173,310-5
18 Sept 2026, 14:153,315-4
14 Sept 2026, 07:553,319-3
11 Sept 2026, 00:013,322-10
5 Sept 2026, 18:383,332-2
30 Aug 2026, 01:373,334-6
26 Aug 2026, 14:543,340-1
23 Aug 2026, 06:173,341-5
19 Aug 2026, 23:333,346first reading

Engagement

20 posts held, back to 15 March 2023 — 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 20 posts for this entry, the most recent from 14 April 2023. An engagement rate over an empty window would be a number about nothing.

What this channel posts

Video runtime
1m 34s
Average length
1m 34s

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.

Reaction mix

367 reactions across 20 posts, in 9 distinct kinds. The most used accounts for 49.0% of them.

Every reaction kind recorded on the sample, most used first
ReactionCountShareShare, drawn
👍18049.0%
❤14539.5%
🥰205.45%
🎉71.91%
👎71.91%
🤯41.09%
🔥20.545%
😁10.272%
😍10.272%

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 367 reactions 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 March 2023 to 14 April 2023, using the newest reading held for each. Telegram Stars are excluded: they are a payment, not a reaction, and they have their own section.

Recent posts

14 Apr 2023, 01:40 UTC≈27,900 views32 reactionsread 7 August 2026
Photo

dolly-15k 的训练集在发布 24 小时就被翻译成了其他语言,这就是开源的力量。 https://huggingface.co/datasets?search=dolly-15k

❤24👍6🥰2

Signed JY

14 Apr 2023, 01:37 UTC≈26,100 views32 reactionsread 7 August 2026
Photo

Sam Altman at MIT 小参数量 + 高质量训练集可能比堆叠参数量要更加重要。 小参数量意味着成本和推理速度更快。

❤19👍10🥰3

Signed JY

14 Apr 2023, 01:28 UTC≈19,900 views11 reactionsread 7 August 2026

Stability AI 发布了新的扩散模型:Stable Diffusion XL Beta 可以使用 API 来进行访问,优化了面部和文字的生成效果。 https://twitter.com/StabilityAI/status/1646672027103707136

👍7❤3😁1

Signed JY

14 Apr 2023, 01:24 UTC≈15,300 views11 reactionsread 7 August 2026
Photo

GPT-4 在标注 twitter 政治相关的推文方面效果要好于人类 虽然 GPT-4 不是分类器,但是可以通过 few-shot / zero-shot 的方式充当分类器,twitter 的这个例子直接使用 zero-shot 就能够很好完成了。 审核工作是非常常见的一种工种,未来被 AI 代替的可能性目前看起来非常高。 https://arxiv.org/pdf/2304.06588.pdf

👍9❤2

Signed JY

14 Apr 2023, 01:18 UTC≈11,100 views10 reactionsread 7 August 2026
Photo

基于富文本的 text-to-image 生成 这个还挺有意思的,把富文本作为特殊的属性标签,例如把 house 这个词改成红色,那么生成的房子就会变成红色。 一定程度上简化了 prompt 的描述,并且可以使用更加具象的信息来作为输入。 https://rich-text-to-image.github.io/

👍7❤3

Signed JY

14 Apr 2023, 01:14 UTC≈9,260 views10 reactionsread 7 August 2026
Photo

一份 ChatGPT 的调查报告 详细介绍了技术、应用、未来挑战这 3 个主题。 https://arxiv.org/abs/2304.06488 https://arxiv.org/pdf/2304.06488.pdf

👍9❤1

Signed JY

13 Apr 2023, 16:45 UTC≈7,910 views18 reactionsread 7 August 2026
Photo

GPT-4 帮助陶哲轩节省了大量繁琐的工作。 https://mathstodon.xyz/@tao/110172426733603359

👍13❤5

Signed JY

13 Apr 2023, 14:56 UTC≈7,150 views17 reactionsread 7 August 2026
Photo

目前看过的最好的 prompt engineering 教程。 已经被翻译成了各种语言,不过看了一下,英文的是最准确的,因为中文的 prompt 只是英译中,非常不 native。 https://learnprompting.org/

❤17

Signed JY

13 Apr 2023, 14:53 UTC≈6,440 views6 reactionsread 7 August 2026

ShareGPT 是一个 chrome 插件,可以分享在 ChatGPT 上有意思的对话。 GitHub Repo 中提供了 API 的形式来调用 ShareGPT 上的数据,一个常见的做法是使用 ShareGPT 上的对话数据来作为其他 pre-train 模型的 instruction tuning 的 input data,可以训练出一个类似 gpt-3.5 的模型(效果略差)。 https://github.com/domeccleston/sharegpt

👍6

Signed JY

13 Apr 2023, 14:45 UTC≈5,770 views1 reactionsread 7 August 2026
Photo

Vicuna-7B 模型(前几天就发布了) 可以直接在 macbook 上跑,使用 pip install fschat 即可安装(但是还是需要自己下载 weights)。 7B 只需要 30GB 的 memory,项目还提供了一个基于 gradio 的 web ui。 开源项目地址:https://github.com/lm-sys/FastChat https://twitter.com/lmsysorg/status/1644060638472470528

👍1

Signed JY

13 Apr 2023, 11:53 UTC≈6,440 views23 reactionsread 7 August 2026

微软在昨天宣布开源了 Deep Speed Chat,能够帮助用户快速训练类 ChatGPT 的大语言模型。 未来可能大模型人手一个了,花个几百到几千刀就可以有一个自己的模型,私有部署。 这会衍生带来很多中间层的机会,比如做 LLM 的 optimization & fine-tuning,model serverless 服务等,甚至是全傻瓜式的一键 fine-tune & 部署的平台。 Copilot Hub 正在做这方面的研究 & 开发,目标是能够让每个人都能训练出自己 Copilot,可以跑在我们平台,也可以自己私有部署。 微软的开源项目地址:https://github.com/microsoft/DeepSpeed

👍16🤯4❤2🥰1

Signed JY

7 Apr 2023, 13:11 UTC≈6,330 views68 reactionsread 7 August 2026

朋友们非常抱歉,最近一周多一直在忙着迭代 Copilot Hub,没有分出精力来更新这个频道,这个周末参加完 Hackathon 后,下周会重新开始更新。 这个频道后面不太想仅仅只做汇总信息,想做点不太一样的(具体还没构思好,评论区也可以讨论讨论)。 另外这个频道永远不会收费(不靠这个赚钱),不当信息差的镰刀,如果想要支持我们,可以去体验我们产品。 https://app.copilothub.co/ 大家周末愉快😄

👍50❤12🎉4🥰2

Signed JY

Showing the 12 most recent of 20 posts we hold for @gainotes. 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.

Cite this entry

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 9 November 2026 — this entry's latest reading, not the date you are reading this.

“Generative AI Insights | 生成式 AI 洞察” (@gainotes), 3,245 subscribers as measured 9 November 2026. Telegram Register, tgregister.com/channel/gainotes.

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.