
UI engineer at @databricks. Formerly Twitter. Vancouver ☞ Taipei ☞ SF.
San Francisco
Joined December 2008
- Tweets2.8K
- Following594
- Followers546
- Likes6.4K
Mike Lin retweeted
Meet DBRX, a new sota open llm from @databricks. It's a 132B MoE with 36B active params trained from scratch on 12T tokens. It sets a new bar on all the standard benchmarks, and - as an MoE - inference is blazingly fast. Simply put, it's the model your data has been waiting for.
Mike Lin retweeted
Probably the thing I’m most excited about with DBRX, it’s super fast! Easily 150 tokens/s for quality comparable to much slower closed models.
Okay @databricks what're you cooking behind this space its so fast lmao
Mike Lin retweeted
We're launching two comprehensive online courses on building and using Large Language Models! The first is on using LLMs in applications, covering topics like prompt engineering, embeddings, chains, and MLOps. The second teaches you to build your own LLMs.
databricks.com/blog/enroll-o…
Mike Lin retweeted
MLflow just added first-class support for LLMs, including integrations with @huggingface transformers/pipelines, @OpenAI and @langchain! Open source #LLMOps is here. databricks.com/blog/2023/04/…
I’ve definitely had some quality coding sessions in some of those Twitter lounge chairs. Wouldn’t buy one though because of bad energy.
This kind of baked in data lineage is where I originally hoped to get to building EagleEye at twitter 10 years ago. cc @feng
youtube.com/watch?v=8wGUnXhI…
Mike Lin retweeted
So…
Be the person to ask the dumb question.
Be the person who admits they don’t know the answer.
Be the person who chooses to walk a path they suspect might lead to failure.
Mike Lin retweeted
This is unfortunately not an anomaly. The most unethical thing I was asked to do was shortly after becoming head of Vine. I haven’t spoken publicly about this out of embarrassment and fear of retribution, but probably better overall for the universe to air out these mistakes…
My key professional strength is being able to type a question in the time between "any questions?" and "I guess not, thanks for watching".
Mike Lin retweeted
A tiny oil and gas facility in Texas spewed methane into the atmosphere with the same earth-warming power as burning seven tanker trucks of gasoline every day. It was one of 533 so-called “super emitters” found in the biggest U.S. oil field. apne.ws/icMwglN
Late to the party on this, but it really is strange not being able to smell anything. Like coffee is just bitter hot water. Hope it comes back soon.
My Datong steamer just short circuited with a pop and I was like did that just happen? I can’t smell anything burning. Totally burnt inside the wire.
This Tableau + Databricks demo is mind blowing. Instantaneous big data queries on Tableau has been a dream for a long time. Amazing to see it live. #DataAISummit















