Ajay (@ajaysagar99) profile picture

Proud Indian 🇮🇳

India
Joined December 2011
Photo by Ajay (@ajaysagar99)Photo by Ajay (@ajaysagar99)Photo by Ajay (@ajaysagar99)Photo by Ajay (@ajaysagar99)
ISM, MeitY, in collaboration with @IntelIndia have launched the "Introduction to Semiconductors" course to educate learners from all backgrounds to the fundamentals of #semiconductor technology and the strategic importance of semiconductors to #India’s economic and technological growth. The course is available in 26 Indian languages and over 240 languages globally, empowering more citizens to understand the #technology powering our digital world in their own language. Learn at your own pace, in your own language. Access the course 👇 🔗learn.ism.gov.in #ViksitBharat #AtmanirbharBharat #IndiaSemiconductorMission
1
7
13
389
Fought for ORS Now this Real life hero🫡🫡
Hyderabad Doctor Brings Down A Rare Disease Drug's Cost From ₹9 Crore To Just ₹2.5 Lakh Imagine being unable to afford a medicine that could help your child. For families battling Allan-Herndon-Dudley syndrome (AHDS), a year's supply of tiratricol once cost nearly ₹9 crore. Now, thanks to Hyderabad paediatrician Dr. Sivaranjani Santosh and Access to Medicines India, seven children will receive a year's supply of the medicine free of cost on November 30. One doctor's determination is turning an unaffordable treatment into a ray of hope for seven families. ❤️ (TH)
1
209
2,620
12,634
Join the launch of AI models developed under the IndiaAI Mission, alongside new research in speech reasoning and multilingual AI. Discover innovations shaping India's AI ecosystem. 📅 25 November | 1:00–4:30 PM 📍 Bharat Mandapam, New Delhi @AshwiniVaishnaw @JitinPrasada @SecretaryMEITY @sudeep_tweets @GoI_MeitY @_DigitalIndia
Join us at Chariot One. Launching AI models developed with the @OfficialINDIAai IndiaAI Mission, alongside new research in speech reasoning and multilingual AI. 25 Nov | 1–4:30 PM Bharat Mandapam, New Delhi Limited seating. Register: luma.com/samfz45o
1
2
16
1,029
Hon’ble Minister of State for Electronics and Information Technology, Shri @JitinPrasada , interacted with the BharatGen Technology Foundation at IIT Bombay, witnessing demonstrations of indigenous AI models and a briefing on ongoing technological initiatives. The engagement underscores India’s commitment to advancing homegrown AI capabilities, fostering research and innovation, and strengthening the country’s position in the global AI landscape. #IndiaAI #BharatGen #AatmanirbharBharat #AIInnovation
Interaction at BharatGen Technology Foundation, at @iitbombay , covering demonstrations of BharatGen AI models and a briefing on its ongoing technological initiatives. As AI continues to reshape industries, research and society, developing indigenous capabilities and fostering innovation in India will be key to ensuring that we remain at the forefront of this transformative technology.
4
4
14
975
Request for startups: I am looking to invest in frontier entrepreneurs building in any of these areas. 1/ AI mediation: One AI system does the work for both sides of a legal case and comes up with the settlement. 2/ Polymarket for insurance: You should be able to bid on offering insurance to people. For example, many homeowners in the Pacific Palisades can't get fire insurance anymore. Build a platform where people bid to offer that insurance, and let the bids price the risk. 3/ Interspecies Communication: We're looking for AI that helps us talk to animals. Decode what they are saying, and let us talk back. 4/ New countries: Anyone starting a new country with a new model for how it runs. 5/ Electromagnetic rocket launches: Build a system that launches rockets horizontally, then vertically, using electromagnetic force instead of traditional propulsion. These ideas are just early concepts. We're open to anything along these lines, so I'd like you to pitch us why your angle is the right one. Comment below, and send me your pitch at [email protected]
321
148
60
1,446
123,488
An Indian lab just shipped an AI model that runs on a phone. India has nearly 900 million people online, and more than half of them would rather use the internet in their own language. Most AI still isn't built for them. Serving them from the cloud gets expensive fast at that scale. FrontiersMind's NandiBERT goes the other way. It's small enough to run on the phone itself, trained from scratch on 11 Indian languages plus English. In my test, it answered in about 0.2 seconds a sentence without a GPU. NandiBERT is state-of-the-art for Indian languages. On its benchmarks, the 150M version beats IndicBERT-v2, a model almost twice its size, on 5 of 7 Indian-language tasks, and the 450M wins all 7. It holds up in English too. Both sizes rank second of nine encoders on average, ahead of ModernBERT and just behind UltraBERT. So any app built for those users can now have search, support and moderation in their language, running on the device with no cloud bill. It's open source under Apache 2.0, free for commercial use.
16
41
104
5,329
Mark my words, Gemma 5 is gonna be a total and complete monster.
20
13
6
656
21,905
Rockstar Games 周三解雇了我。第二天 GTA 6 就发售了 我是 Rockstar North 的高级引擎工程师,从事 RAGE 性能优化。三个月前,我在遥测模块里发现了一条本不该存在的数据管道,而且我能证明它流向了哪里 此刻每一份正在运行的 GTA 6 都在收集玩家的行为数据。这部分不算新闻。每款游戏都会收集遥测数据。隐私政策里写的是"用于改进未来产品的匿名游戏数据分析"。我发现的不是这个 遥测模块的数据会汇入 Take-Two 内部的一个系统,名为 Lens。Lens 并不会把匿名使用统计数据发送到 Rockstar 开发人员的数据看板。它为每一位玩家建立一份个人行为画像。不是匿名的。与你的主机账号、你的平台 ID、你的 IP 绑定 我在8月14日向我的技术主管报告了这件事。他说这是"用于玩家体验研究的扩展分析",让我别管这事。我要求提供数据处理协议。他说这是在 Take-Two 集团层面处理的,不在 Rockstar 的职责范围之内 我调取了 Lens 数据管道的完整文档。我发现的内容如下: 遥测的内容远远超出游戏时长和崩溃报告。Lens 以毫秒级精度记录每一次输入。按键时间点、摇杆动作、反应速度、PS5 自适应扳机上的扳机压力模式。它实时运行行为分析,并生成他们称之为"玩家行为签名"的东西 - 一份对每位玩家都独一无二的画像,就像指纹对每个人都独一无二一样 Lens 会从你的游戏过程中推导出:冲动控制评分、风险承受度指数、攻击性反应模式、耐心阈值、压力下的决策速度、消费冲动评级。这些不是调试指标。这些是心理行为类别 直到我找到了对外传输的管道,我才明白它们是用来做什么的 这些数据不会留在 Rockstar。也不会留在 Take-Two。Lens 通过内部 API 将行为画像传送给 Take-Two 旗下一家名为 Vantage Analytics 的子公司。Vantage 没有公开网站,没有新闻页面,没有招聘信息。它于2023年在特拉华州注册成立 Vantage 是一家行为 AI 公司。它用玩家画像训练预测模型,并把这些模型授权给商业客户。管道文档中包含一份不完整的客户清单。两家保险集团,一家信用风险建模公司,一个人力资源筛选平台,以及一家广告数据经纪商 你的 GTA 6 游戏过程此刻正在生成一份行为评分。没人追你的时候,你开得有多快。你会不会在红灯前停下。NPC 无端攻击你时,你怎么反应。你行动之前要想多久。你花游戏内的钱有多冲动。每一个选择都在喂养一个模型,企业正在付费查询这个模型,用来对真实的人做决定
2,719
20,926
2,629
118,380
5,429,242
after the announcement of looped transformers, vLLM are rebranding vLLM to vLLLM.
Stay in the loop with vLLM. Day-0 vLLM support for @thinkymachines’ new looped Inkling model has landed. 12 shared core layers, 3 passes, and gated messages from one pass to the next. 1/4
12
17
3
225
8,351
Ajay retweeted
Your model should dream while you sleep. 😴 Today we’re open-sourcing Songe. It’s a local training engine that turns your conversations, instructions, and corrections into training data, then fine-tunes your local model on your own tasks and preferences. It all runs on your own machine, using compute that would otherwise sit idle. Your data stays on your device. The model that gets better is yours. pip install songe souslabs.ai/
69
94
23
1,008
44,359
🧵 NandiBERT is out. Open encoders that are state of the art in Indic 🇮🇳 and among the best in English 🌍 📦 150M (16K context) 📦 450M (32K context) 🔓 Apache 2.0 Here's what's inside 👇
4
9
20
509
Classic government employee vibes. Feeling superior without achieving anything. Running away from accountability, again just like the government :D
This is how @SarvamAI employees speak typically I made a post on their founders calling India “data colonies of the west” btw when their own employees use the same foreign tools They just troll but hardly answer any questions
1
1
98
Werent the previous rounds raised for setting up on prem compute at govt installations ? So are you suggesting that the gpus installed at govt installations are also being used for providing access to token factory and that is how workload has been disturbuted
2
1
1
11
1,468
Ajay retweeted
Is there a thinker behind the thought? Is there a doer behind the deed? Is there an observer behind the observation?
614
423
84
5,424
247,401
someone sent me this. saying that govt gpus are being used for serving inference. I want to keep it anonymous, but I can't verify the info.
2
1
1
9
811
这个开源模型:用 1000 部成人视频第一时间重新训练的 MiniMax H3,而且压缩了 0.62GB 完全免费,而且极致写实。。。。 人人都能做黄果视频的时代要来了。。。。。。 huggingface.co/Sentinel7/h3/…
99
313
42
4,689
492,752
Big Pharma wanted ₹9 crore a year for a drug. A Hyderabad doctor just bypassed them and got it for ₹2.5 lakh. 🤯 She didn't wait for the company to drop the price. She used a legal workaround in the Netherlands. For kids with a rare genetic disorder called AHDS, a year's supply of the medicine (tiratricol) costs up to ₹9 crore in the US. One Indian mother was paying ₹12–13 lakh every month just to import it from Germany. When attempts to get the drug through the manufacturer failed, paediatrician Dr. Sivaranjani Santosh put out a public appeal on social media in September. A Dutch non-profit (Pharmaceutical Accountability Foundation) saw it, teamed up with Transvaal Pharmacy in The Hague, and legally custom-compounded the exact same medicine using the raw active ingredient. • The annual cost crashed from ₹9 crore to ~₹2.5 lakh per child. • Donors in India and the Netherlands pooled ~₹15 lakh to cover the entire bill. • With Access to Medicines India and ORDI clearing the import permits, 7 kids will get a full year’s supply completely free on November 30. The commercial price hasn't changed globally, and Dr. Sivaranjani says importing batches from Europe isn't a permanent fix. She’s now pushing the Indian government to allow local compounding under the National Policy for Rare Diseases so families aren't left at the mercy of ₹9 crore price tags.
117
1,717
92
7,856
192,289
inkling 2 mini is the first small open model I'd actually try for real coding work on my own machine
8
2
87
3,561
Today we’re releasing Inkling 2 Mini, a looped transformer, with open weights. A big reason we're doing this is to let the AI safety community study looping, especially how loops affect chain-of-thought monitorability and how to interpret the computation inside loops. 🧵
Some of you thought that Inkling 2 Small was too big at 594B parameters. We heard you: Inkling 2 Mini is a looped transformer with just 26B parameters, small enough to run on your laptop, powerful enough for agentic work with a score of 20.1 on AA Index. thinkingmachines.ai/news/ink…
17
50
5
562
21,434
Some of you thought that Inkling 2 Small was too big at 594B parameters. We heard you: Inkling 2 Mini is a looped transformer with just 26B parameters, small enough to run on your laptop, powerful enough for agentic work with a score of 20.1 on AA Index. thinkingmachines.ai/news/ink…
521
1,444
879
15,198
2,440,646