Number of topics

[AbhayGoyal]

Hi,
Can you tell me how to change the number of topics?

Why does the lvl2 keep on running for me?

[anne-kreuter]

Hey, I probably had the same issue. When running lvl2 it seemed to get stuck while collecting all top-level topic assigned documents. My solution was processing the documents in fixed-size batches rather than one-by-one. This is achieved by introducing a new parameter (batch_size) and modifying the prompt construction so that each prompt is built from a batch of documents (in my case 5). This prevents the uncontrolled growth of prompts and improves performance.

[AbhayGoyal]

Ok yeah makes sense

…

On Tue, Mar 4, 2025 at 10:29 AM anne-kreuter ***@***.***> wrote: Hey, I probably had the same issue. When running lvl2 it seemed to get stuck while collecting all top-level topic assigned documents. My solution was processing the documents in fixed-size batches rather than one-by-one. This is achieved by introducing a new parameter (batch_size) and modifying the prompt construction so that each prompt is built from a batch of documents (in my case 5). This prevents the uncontrolled growth of prompts and improves performance. — Reply to this email directly, view it on GitHub <#17 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEMF2JQG344M33HYEZTXPSD2SXIFXAVCNFSM6AAAAABXALWCB6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMOJYGI2DONZYGE> . You are receiving this because you authored the thread.Message ID: ***@***.***> [image: anne-kreuter]*anne-kreuter* left a comment (chtmp223/topicGPT#17) <#17 (comment)> Hey, I probably had the same issue. When running lvl2 it seemed to get stuck while collecting all top-level topic assigned documents. My solution was processing the documents in fixed-size batches rather than one-by-one. This is achieved by introducing a new parameter (batch_size) and modifying the prompt construction so that each prompt is built from a batch of documents (in my case 5). This prevents the uncontrolled growth of prompts and improves performance. — Reply to this email directly, view it on GitHub <#17 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEMF2JQG344M33HYEZTXPSD2SXIFXAVCNFSM6AAAAABXALWCB6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDMOJYGI2DONZYGE> . You are receiving this because you authored the thread.Message ID: ***@***.***>