This is Qwen 3.6 running in only 11GB ram although its 21.6GB on Disk, using an optimized file format for model streaming on MLX and still processing prompts at 1390 Tok/s ...... yes it is possible with a file format that is perfectly optimized for paging/swapping experts from disk.
We’ve received notice that the Department of Commerce has lifted export controls on Claude Fable 5 and Mythos 5.
We'll begin restoring access tomorrow, and will share an update soon.
We’re grateful to our users for their patience, and to everyone who worked with us on redeploying the models.
@NCGardening BTW @NCGardening - do you have any updates on the parthenocarpic smyrna fig experiment? There was a video showing what seemed like unexpected female fruiting ratio. Did that continue to hold?
@NCGardening Thank you for the amount of information you've been sharing. I enjoyed watching the fig series, and experimenting with persistence. That's amazing. Did any of your experiments include Smyrna figs? That would be a breakthrough for east cost. I'm in zone 7 - no wasps :(
Great work @aosama & Matt Gillet from @FINRA_Tech on presenting “Machine Learning At Scale” at the “Apache Spark Interactive” meetup hosted at @FINRA. More than 75 in attendance. Great job guys!
"We ...applaud our all-volunteer community, and thank the billions of users who benefit from their Herculean efforts." –20 Years of Open Source Innovation, The Apache Way https://t.co/p3wgr48zGE #Apache#Open#Innovation#Community#Leadership#TheApacheWay#ASF20
join us for a hands-on machine learning session using Spark MLIB and Databricks. Will share our experience @FINRA_Tech using ML algorithms to conduct market surveillance on 135bn events per day. https://t.co/iAolazc8yg