DeepSeek Harness is so difficult to use! The document is a mess and is not for human use. It relies entirely on AI to query the document!
In the past, when there was no AI, the documentation was very good. Now that there is AI, the documentation is a mess, and we all rely on AI to search to find things. It’s so difficult! It takes a long time to access a LAN machine. Listen to a song you just made to relieve your boredom!
Jev vs Deepseek vs Luna on Connect 4
Jev beats Deepseek and ties with Luna But Deepseek faster than Jev 👀
It seems that Jev is also learning using synthetic data and labels from a large-scale model
It seems that Jev is also learning using synthetic data and labels from a large-scale model, so I distilled and learned the judgments of DeepSeek V4 Flash with a weight of 157GB over 26 hours using DGX Spark, and transferred it to the 4B model. It is 1/20th the size and exceeds the teacher's immediate response mode, approximately 22ms per judgment. I never thought I could do this much with Local LLM alone.
I'm back with another banger demo from DeepSeek-v4.1-flash.
This is the best result i've ever gotten from v4.1-flash Not just visually. This is probably the most useful, polished, and genuinely mind-blowing thing i've seen it build so far. If i had something like this when i
You have reached the end of the archive
All of deepseek