Tianqi Chen comments

Results 637 comments of


                                            Tianqi Chen

[Question] Can we build a Mac or Universal version of the iOS app?

Can you provide more details about the apple silicon prebuild and what does not work? Would love to help digging into it

[Bug] iOS app no longer available

thanks for bring this to the attention, we will work on recovering the app

[Doc] how to install on android?

You can just instal APK as per https://llm.mlc.ai/docs/get_started/quick_start.html

[Backend][Relax] Add Intel GNA backend for NPU support

Thanks for the contribution, given GNA is archived, it perhaps does not make sense to maintain it in the main tree, adding ci will also add extra overhead here. However,...

[Backend][Relax] Add Intel GNA backend for NPU support

i think starting as contrib is fine, and we can have a tutorial explaination point to the code

Wasm file generating but not compatibility issue.

we are working on bringing up the webllm to latest tvm, cc @CharlieFRuan for tracking

Added RISC-V V extension intrinsics for LLVM

cc @cbalint13 can you help to take a look

Handling of streams for new API layer

I think there are two kinds of design patterns in libraries we see so far - D0: Stream life-cycle management is tied to array life-cycle management, something discussed in this...

Handling of streams for new API layer

Thanks for the further discussions > Should JAX / Tensorflow pass an internally used stream through the optional_out_env_stream even though there's no ability for the dlpack deleter or the user...

Handling of streams for new API layer

I feel likely we need to go for the conservative recommendation (maintaining the status quo): - JAX make sure data is synchronized to a known stream(default stream) - JAX pass...