Saswat Das @_saswatdas_

that moment when you start typing into @ChatGPTapp and then hydration kicks in and wipes the input field EVERY SINGLE TIME, drives me absolutely insane. do something about it @OpenAI

0 0 0 65 0

Saswat Das @_saswatdas_

3 months ago

the more type-safe a library is, the better it works with agents

0 0 1 55 0

Andrej Karpathy @karpathy

4 months ago

LLMs are chmod a+w artifacts yay

162 177 4K 576K 531

Saswat Das @_saswatdas_

4 months ago

gemini 2.5 Pro is damm good

0 0 0 55 0

Saswat Das @_saswatdas_

4 months ago

Such a nice day! 🇮🇳 #OperationSindoor

0 0 0 190 0

Saswat Das @_saswatdas_

5 months ago

Serverless is damm costly at scale ..

0 0 0 63 0

Saswat Das @_saswatdas_

5 months ago

Seedhe Maut X HIIT

0 0 0 81 0

Harj Taggar @harjtaggar

6 months ago

Autists had a great run, the AI future belongs to ADHD

385 791 10K 934K 2K

dax @thdxr

6 months ago

😂💀

40 9 550 40K 67

Download Video

0 0 0 77 0

Microsoft Developer @msdev

6 months ago

TS is being ported to GO🫡🫡

7 36 241 23K 24

Download Video

0 0 0 53 0

Saswat Das @_saswatdas_

6 months ago

MCP was released on 25 Nov 2024. Why are people hyping it now after 4 months ?? Did I miss something ?

0 0 0 35 0

Introducing Claude 3.7 Sonnet: our most intelligent model to date. It's a hybrid reasoning model, producing near-instant responses or extended, step-by-step thinking. One model, two ways to think. We’re also releasing an agentic coding tool: Claude Code.

1K 3K 19K 3.6M 3K

Download Video

Saswat Das @_saswatdas_

6 months ago

What is Claude 3.7 sonnet ?????

0 0 1 40 0

Saswat Das @_saswatdas_

6 months ago

I will be honest. Grok3 is really good @xai

0 0 0 37 0

Naval @naval

6 months ago

Ultimately, you attract the complement of what you project.

353 2K 20K 1.3M 3K

Saswat Das @_saswatdas_

7 months ago

If memory is tricky in agents, try using a smaller model (e.g., O1 or a distilled R1 around 8B) to extract factual info from chat messages. Then update memory based on those facts.. I would love to perform a benchmark..