Jiahao Li
Jiahao Li
The link should be correct. It contains "50 video from websites like YouTube".
That's cool. I need some time to review and test it. Glad that you're interested in this repo!
It is definitely possible. Any pull request is welcome!
Hi, I finally got time to review this POC and based on it I have implemented an accelerated version supporting both forward and backward energy (#4). Numba jit gives a...
Thanks for your advice. I have implemented an accelerated version using numba jit and gained a 4x end-to-end speedup for forward energy and 2x for backward energy (#4). You may...
Maybe you can try https://github.com/li-plus/chatglm.cpp. It's written in the same way as llama.cpp based on GGML. Quantization does help to improve inference speed.
> @li-plus Great thing, is there a plan to support gpu inference？ Sure there is. I'll make it infer-able on a consumer GPU.
It seems that `assert_jit_shape_analysis=True` is not supported if the op input is a list of tensors. Should I remove it?
Removed `assert_jit_shape_analysis=True` and now it should pass the CI. Would you approve the workflow?
@bdhirsh @eellison hi, is there any update for this?