Deepak Vij

Results 5 comments of Deepak Vij

KVCached support for gpt-oss-20b attention type not supported in SGLang

@jiarong0907 @ivanium , look forward to collaborating with you folks on this effort.

KVCached support for gpt-oss-20b attention type not supported in SGLang

@ivanium , I am going to do that right now and let you know. I was on vacation for a while. Thanks.

KVCached support for gpt-oss-20b attention type not supported in SGLang

@ivanium, it seems SGLang does not provide support for --disable-hybrid-kv-cache-manager

[Feature] Faster torch.compile

Hi @merrymercy , curious to know if you are seeing slowness in the initial compilation process itself. Or you see slowness while using the torch.compile cache, once built. In my...

[Feature] Faster torch.compile

Hi @merrymercy , this week I spent some time on this issue. I was able to observe good performance numbers with **TORCHINDUCTOR_CACHE_DIR** option enabled. The very first time with **no...