Update: Web Service Speed Improvements, Communist Ringo Learning, Dictionary Content Additions
I've been steadily updating, but it's been a while since I last wrote a post.
In the meantime, I've been deeply immersed in Soviet history and worked hard on the 'Communist Ringo' menu, adding a person dictionary, a term dictionary, and a historical events database.
It still needs to be reinforced going forward, but the materials are quite rich and I've put effort into the UI/UX, so please stop by the Communist Ringo dictionary.
As for the database, I was using Supabase SaaS, but the communication delay was frustrating (almost like traveling around the world), so I moved it to the local DB of the web server. Thanks to that, the page load speed of this site has improved a lot.
Also, when querying VectorDB, I was wasting 30 seconds on reranking with a local VM CPU, but by removing the reranking step, I reduced it to under 1 second, improving the chat service speed.
I've reverted the web chat AI model from DeepSeek v4 Flash to Pro, so the response quality is actually higher.
In the meantime, I've been deeply immersed in Soviet history and worked hard on the 'Communist Ringo' menu, adding a person dictionary, a term dictionary, and a historical events database.
It still needs to be reinforced going forward, but the materials are quite rich and I've put effort into the UI/UX, so please stop by the Communist Ringo dictionary.
As for the database, I was using Supabase SaaS, but the communication delay was frustrating (almost like traveling around the world), so I moved it to the local DB of the web server. Thanks to that, the page load speed of this site has improved a lot.
Also, when querying VectorDB, I was wasting 30 seconds on reranking with a local VM CPU, but by removing the reranking step, I reduced it to under 1 second, improving the chat service speed.
I've reverted the web chat AI model from DeepSeek v4 Flash to Pro, so the response quality is actually higher.