MakotoUwu commented on PR #19771: URL: https://github.com/apache/tvm/pull/19771#issuecomment-4732450569
Hi @tqchen, just checking whether the latest head addresses your concern about the chunking motivation and runtime scope. One extra bit of context: the large record here is not specific to a local conversion artifact. Google's current Gemma 4 model card documents the E2B/E4B Per-Layer Embedding design: the small models keep large per-layer token embedding tables while exposing a lower "effective" parameter count. In the WebLLM artifact this shows up as valid tensor-cache records such as the 1120 MiB per-layer embedding weight, which is the concrete case this runtime staging change handles. The PR itself remains scoped to the Web runtime path, required CI is green on `a2359c159cb3fb94766335beea33f281fb9a7bba`, and the latest follow-up also includes the downstream WebLLM smoke fix for the `makeShapeTuple` scope issue. Please let me know if you would prefer an additional runtime test or a smaller adjustment before re-review. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
