Unable to see the full context inserted in the memmory and no semantic search is happening to retrieve the actual context

[hariumeshuk]

Hello,
I having some queries related to how mem0 stores the context ?
i have inserted below conversation result in the mem0
id: 123 for all insertion

1st insert
 "response": "The cheapest product we have is listed with a product ID of 70, and it is priced at $12.99. Could you please share more about what you're specifically looking for in a product? For instance, are you looking for specific categories like electronics, clothing, or home items? This will help me understand how this product might meet your needs."

2nd insert
The cheapest product we have is priced at $12.99. If you are looking for a budget-friendly option, this product could be a great choice. 
 Could you share what specific features or types of products you are interested in? This way, I can provide more tailored recommendations!

3rd insert
"The cheapest product we have is priced at $12.99. To provide you with more relevant details, could you specify the type of product you are interested in? For example, are you looking for electronics, clothing, or health products?"

However when i am trying to retrieve the context for the same id: it return 3 above 3 result with some changes in the context

so my question is,

1. does mem0 use any semantic method to insert the context in the memmory that only store relevant context and not all context?
2. Does it allow to store duplicate context or similar context ? if so how should restrict such insertion making non-duplicate insertion against the same id
   3.) How does it perform the semantic search? Does it require prompting to retrieve the actual context against the query passed?

[TRON4R]

Not sure, which installation you're using (cloud or local), but my local installation couldn't retrieve anything until I realized that the modell I has used (ollama 3.2 3B) was incapable of dealing with another language than English. As soon as I used english keywords in my openclaw mem0 test queries, the answers worked.

So it's important to use a decent model to store information (such as ChatGPT-4.1-nano) and a good model for embedding (such as openai/text-embedding-3-large). I also greatly used the help of Opus 4.6 to set things up.