Deepanshu Sharma
@deepanshusharmx
Followers
31K
Following
31K
Media
2K
Statuses
14K
write about LLMs, MCPs and AI Agents⚡
Earth
Joined July 2023
this is beyond imagination, it will be very addictive if chatgpt could initiate conversations by itself, especially when a user instructs it to behave in a certain way. there are many possibilities that arise with it. one is, people can instruct it to continuously remind them
It's inevitable that ChatGPT will initiate conversations. Their head of model behavior said it's in the "realm of possibility". Notifications can be RLHF'd to be maximally engaging. "You haven't left the house all day. Did you take your meds?" Many users will be addicted.
1
0
2
Tony Stark vibes
"Chatting" with LLM feels like using an 80s computer terminal. The GUI hasn't been invented, yet but imo some properties of it can start to be predicted. 1 it will be visual (like GUIs of the past) because vision (pictures, charts, animations, not so much reading) is the 10-lane
0
0
0
😅😅
0
0
0
This weekend. Tablet's Hanukkah Bazaar is back and better than ever. Come shop and schmooze with us on December 6th and 7th on the Lower East Side. Get your tickets today.
0
7
41
comments to this post are insane, open-source heat is rising.. its good i guess
goodbye, GPT-4. you kicked off a revolution. we will proudly keep your weights on a special hard drive to give to some historians in the future.
0
0
0
So, its officially started..
aI wOnT rEpLaCe mE. i aM imPorTaNt "Duolingo will “gradually stop using contractors to do work that AI can handle,” according to an all-hands email sent by co-founder and CEO Luis von Ahn announcing that the company will be “AI-first.” The email was posted on Duolingo’s
0
0
0
we are living in crazy times, anything can be manipulated rn, text, images, videos, now live interviews also
Online interviews are dead, This guy literally built AI interview glasses, that clones your voice and delivers perfect answers in real time. https://t.co/63pjVyKnLj
0
0
3
Google is not stopping at all🔥
Google just dropped TxGemma, A new open-source model to accelerate drug and therapeutic development. Here's everything you need to know:
0
0
2
Google's Agent to Agent communication protocol, we are accelerating at unprecedented speed🔥
I've just been reading the Agent2Agent technical docs - Google's new protocol for agent to agent communication. You know what that means. Let's diagram them:
0
0
1
$AREB “Massachusetts is the cradle of American liberty — Lexington, Concord, the Minutemen, the shot heard ’round the world — this is where the fire of freedom was lit. The people here still carry that same defiant, patriotic heartbeat, and they damn sure know great beer when
17
11
70
another great feature from OpenAI🔥
we have greatly improved memory in chatgpt--it can now reference all your past conversations! this is a surprisingly great feature imo, and it points at something we are excited about: ai systems that get to know you over your life, and become extremely useful and personalized.
0
0
2
these results are extraordinary..
Gemini Deep Research now uses Gemini 2.5 Pro 🔥🔥 what Google is doing is not normal. it's available to all Advanced users.
0
0
1
All eyes on Google🔥
Should be a good week 👍
0
0
0
0
0
1
there's a difference between "10m context window" and "10m window" they knew you couldn’t verify the “context.”
1
0
0
this is the most unacceptable thing they could’ve done.
If Meta actually did this for Llama 4 training to maximize benchmark scores, it's fucked.
0
0
2
Finally confirmed. > GPT-4: 1.76T > GPT-4o: 200B > GPT-4o-mini: 8B > o1-preview: 300B > o1-mini: 200B > Claude 3.5 Sonnet: 175B
The latest Microsoft paper finally reveals the model size of known LLM models. > GPT-4o-mini: 8B > Claude 3.5 Sonnet: 175B > GPT-4: 1.76T > GPT-4o: 200B > o1-preview: 300B > o1-mini: 200B Wonder how they figured it out..
0
0
1
Aider Polygot benchmark
0
0
1
open-source is the way to go🔥
0
0
1
Intelligence is too cheap to meter
Introducing our first set of Llama 4 models! We’ve been hard at work doing a complete re-design of the Llama series. I’m so excited to share it with the world today and mark another major milestone for the Llama herd as we release the *first* open source models in the Llama 4
0
0
2
10M context window, is it PRACTICALLY possible?
0
1
2