Anup
@anupk24
Followers
313
Following
610
Media
11
Statuses
114
research @openai
San Francisco, CA
Joined November 2009
this is so cool.
New Anthropic Research: Project Vend. We had Claude run a small shop in our office lunchroom. Here’s how it went.
0
0
3
It is actually wild that Siri is so bad but multimodal models are getting so good at realism, intelligence and latency. The delta in capabilities will soon unlock new patterns for how we communicate with devices. Time really is ticking for Apple.
1
0
7
It was a fun (and intense) sprint working on the early versions of our codex (agent) and training the codex (mini-latest) model with @hwchung27 @fouadmatin @rohancalum, optimized for codex (cli)! 😅.
We've made some improvements to Codex CLI, based on your feedback:. ⬥ Sign in with ChatGPT to quickly connect your API org.⬥ New model, codex-mini, optimized for low-latency code Q&A and editing
1
2
24
why read Meditations when “just put the fries in the bag lil bro” gets you 90% there.
3
1
21
RT @Felipe_Millon: Today, we at OpenAI launched Deep Researcher and I wanted to share a deeply personal story about how amazing this tool i….
0
833
0
love the academic equivalent of a patent troll (currency = social credit).
DeepSeek [1] uses elements of the 2015 reinforcement learning prompt engineer [2] and its 2018 refinement [3] which collapses the RL machine and world model of [2] into a single net through the neural net distillation procedure of 1991 [4]: a distilled chain of thought system.
0
0
2
RT @dylan522p: So this is like 2T in loss market cap for a $6M training run (ignoring cost of research, ablations, distilled data from GPT,….
0
126
0
more evidence of investors not knowing what they’re talking about.
This report is long but very good. “With R1, DeepSeek essentially cracked one of the holy grails of AI: getting models to reason step-by-step without relying on massive supervised datasets. Their DeepSeek-R1-Zero experiment showed something remarkable: using pure reinforcement.
0
0
3
1) What.
Wow. This, if true, raises serious questions about yesterday’s announcement.
0
0
5
this will age poorly.
Today Gemini 2.0 DESTROYED everyone on lmsys: .- kills o1 in math and coding ???.- handily beats Claude 3.5 even with Style Control ???. Meanwhile: .- Meta: "new LLaMa3.3-70B model go brrrrrrrr, you guys care right? please don't use Qwen2.5 72B".- OpenAI: "here's some mostly
1
0
3