Next
In hand: nothing
Reproduction notes for this level →Teaching simulation · no online model
Open a concept for Chinese, Japanese and English terms and an explanation.
Finding candidate material relevant to a question. Evidence sufficiency still needs to be checked.
In this lesson：Finding the one page
Open a concept for Chinese, Japanese and English terms and an explanation.
Finding candidate material relevant to a question. Evidence sufficiency still needs to be checked.
In this lesson：Choosing among different pages
Open a concept for Chinese, Japanese and English terms and an explanation.
Finding candidate material relevant to a question. Evidence sufficiency still needs to be checked.
In this lesson：Understanding a different way of asking
Finding material by semantic relevance. Similarity does not imply factual equivalence. Level 8 uses hand-authored result lists.
Open a concept for Chinese, Japanese and English terms and an explanation.
Finding candidate material relevant to a question. Evidence sufficiency still needs to be checked.
In this lesson：Giving the question and the page to the LLM to generate an answer
A model that generates language from input. Replies in this game are prewritten teaching content, not live model calls.
Retrieve relevant material, then provide it with the question to a model for generation. Retrieval does not guarantee a correct answer.
Open a concept for Chinese, Japanese and English terms and an explanation.
Finding candidate material relevant to a question. Evidence sufficiency still needs to be checked.
In this lesson：Retrieval candidates and ranking
Reordering retrieved candidates; it cannot add material that was never retrieved. Level 5 uses manual ordering.
Open a concept for Chinese, Japanese and English terms and an explanation.
Changing the retrieval query while preserving the intent of the original question, which the answer should still address.
In this lesson：Improving how the query is phrased
Retrieving with multiple query formulations and merging distinct candidates. Level 6 demonstrates rewriting, not automatic multi-query execution.
Open a concept for Chinese, Japanese and English terms and an explanation.
Splitting documents into retrieval units. Separating subjects or conditions can leave a fragment insufficient for an answer.
In this lesson：Recovering a fragment’s context
Retrieving a small fragment and expanding to surrounding or parent content to recover subjects and conditions.
Open a concept for Chinese, Japanese and English terms and an explanation.
Finding material through lexical matches, useful for exact identifiers and names.
In this lesson：Keyword and semantic search
Finding material by semantic relevance. Similarity does not imply factual equivalence. Level 8 uses hand-authored result lists.
Combining retrieval routes such as keyword and semantic search. Level 8 interleaves lists and removes duplicates.
Open a concept for Chinese, Japanese and English terms and an explanation.
A model that generates language from input. Replies in this game are prewritten teaching content, not live model calls.
In this lesson：Spotting a gap and searching again
A system choosing subsequent actions such as retrieval based on task state. Level 9 illustrates a fixed evidence-gap loop, not an autonomous agent.
Open a concept for Chinese, Japanese and English terms and an explanation.
Organizing and retrieving evidence using entities and relations. Level 10 demonstrates paths in a hand-built graph, not automated construction or a full GraphRAG system.
In this lesson：Answering with relations between entities