This is the 2023 Turkish Wikipedia dataset with added embeddings and token counts. The embeddings were generated using the BAAI/bge-m3 model, which also calculated the token counts. The embeddings are stored in float16 format, and articles exceeding 8000 tokens were excluded due to the context window limit of the embedding model.
Embeddings are only dense, not ColBERT or Lexical.
