Adityam Ghosh⚡

Released a new @OpenAlex_org S3 wrapper for efficient processing of large OpenAlex datasets. PyPI: pypi.org/project/pyalex…

@abhi1thakur This agentic thing has me all confused. The only agents I previously knew where RL agents. But this new thing has me all confused.

🤣

@sarahookr You can check out Petals by @BigscienceW : github.com/bigscience-wor…

what is the best open source code framework for decentralized compute? btw no requirement for private -- I'm just interested in easy to use decentralized coordination. also is there an approachable UI for a community to opt into sharing their compute.

Bought an @Easyfone_Roy for my grandfather via online store. They gave me a defective one and hasn't replaced the phone even after thousands of complaints and delivery charges that I had to pay (extra) and send to their "technicians"

@huggingface any plans for the open-source version of o1?

Did you notice how most video AIs are commercial, closed-source & secret? It sucks and a big reason for the lack of open-source & transparency is that there are very few high-quality open video datasets. The @huggingface team is changing this today by releasing FineVideo, a high-quality annotated video dataset with a creative commons license. Time to build!

@polynoamial @OpenAI @rao2z Do you guys have any plans to release a paper on this?

@OpenAI @rao2z You can read more about the research here: openai.com/index/learning…

Today, I’m excited to share with you all the fruit of our effort at @OpenAI to create AI models capable of truly general reasoning: OpenAI's new o1 model series! (aka 🍓) Let me explain 🧵 1/

@itsfoss I use it even on my latest gaming pc.

Save old PCs with Linux 🐧 💪🏽

The "underlying factor" that pushed me to pursue a career in Computer Science and AI. "Still taking me back to school!"

@bozo10n Hey so did mine. To even this day, I watch this movie when I am feeling down and not feel like working. Get's me back to my beast mode 💪

This 50 sec scene single handedly shaped my entire career 16 years ago

@abhi1thakur I am still working on mine, as it's not based on LLMs and VLMs. It's more about a researcher's hub where they can connect and collaborate with each other and exchange ideas.

So, how many startups did apple kill today? 🤔

@alex_conneau @OpenAI Congrats! wishing you all the best in your future endeavors.

Career update: After an amazing journey at @OpenAI building #Her, I’ve decided to start a new company.

@barrywinata @Hamptonism Try EurekAI, and you can actively collaborate with researchers and learn from them. It's like a GitHub for Researcher. eurekai.tech

@Hamptonism This pretty neat. I think this would be so useful for general articles and content on the web.

Students at Stanford have built alphaXiv, an open discussion forum for arXiv papers. Think of it as 𝕏 for research.

@Hamptonism I got something better, try EurekAI you can even collaborate with the researchers in that platform: eurekai.tech It's the GitHub for researchers

Being a big fan of @allen_ai, just read their SPECTER paper recently. Was literally amazed how a simple concept can create a SOTA model. Also, it was fascinating for me to see that they trained the whole model on a single Titan V GPU.

@tunguz Lol looks like she committed every single change every day of every month for 2 years straight 😂

Scarlett Johansson GitHub contributions timeline.

Just added some more features to EurekAI. Trying to get this app slowly off the ground and help researchers in their research journey. eurekai.tech #artificalintelligence #startup #research

There is OpenAI and then there is AllenAI both working on great applications of NLP but one post their models on Huggingface and share it with the rest of the world for free.

AI21Labs just released Jamba 1.5 Mini & Large - MoE, permissively licensed, 256K context, Multilingual, JSON model & Tool use 🔥 > Jamba 1.5 Large - (94B active, 398B total) > Jamba 1.5 Mini - (12B active/ 52B total) > Arena Hard: Large (65.4), Mini (46.1) > MMLU: large (81.2), Mini (69.7) > Architecture: Joint Attention with Mamba + Mixture of Experts > Context length: 256K > Languages supported: English, Spanish, French, Portuguese, Italian, Dutch, German, Arabic and Hebrew > Permissive license for orgs < 50M USD in revenue > Upto 2.5x faster inference with models comparable in size > Allows for tool use & feedback > Supports grounded generation for taking in a list of documents > Comes with an explict JSON mode > Integrated with Transformers 🤗 > Introduce a new ExpertInt8 quantization to quantize 85% of the MoE in int8 > Finetune with Transformers + Axolotl + FSDP on a single H100 node Congrats AI21 Labs for such a brilliant open release and sharing this with us! 🐐

AI21 released 🥁 Jamba 1.5 - Mini (52B - 12B active) + Large (398B - 94B active) - Instruct models - Long context (256k) - Strong quality - Multilingual - Function call, JSON output, document understanding - Hybrid transformer/Mamba arch 🔥Read more at huggingface.co/collections/ai…