Bernardo Ramos

I suspect that currently only `clang` compiles to Apple Silicon, requiring Xcode 12.2 on Mac. `gcc` has a [suspended ticket](https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96168) for its support, but there is a work-in-progress implementation [here](https://github.com/iains/gcc-darwin-arm64#readme).

Can someone review this? The current README is bad formatted, not clear to understand

@andpor Done. I think that the README is still confusing because it mix instructions for many versions of RN. In my opinion the README should contain instructions for only the...

Lets do it in steps. You can merge this one first. Maybe someone else that understands React Native could do it better than me. My world is C :P

It is not possible to run OpenDevin as root. Check #969 and #887

Good job! I suspect this would be better as a separate repo, as it may have different instructions to run it, and other people may create different implementations Suggestions: *...

@ankan-ban Would not be better to make the computations in FP16 as well? Currently it has lots of conversions BTW, I am learning a lot with your code. Thank you!

Same with me. It is failing somewhere: ``` [email protected]:~/llama2.cu$ ./llama2 stories110M.bin Model params:- dim: 768 hidden_dim: 2048 n_heads: 12 n_kv_heads: 12 n_layers: 12 seq_len: 1024 vocab_size: 32000 [email protected]:~/llama2.cu$ echo $?...

It works when we use a `tokenizer.bin` from previous commit like [this](568a651c45038712859b606b51176b33061fd353) But the output is gibberish: ``` hidden_dim: 2048 n_heads: 12 n_kv_heads: 12 n_layers: 12 seq_len: 1024 vocab_size: 32000...

The instructions I followed: ``` git clone https://github.com/ankan-ban/llama2.cu cd llama2.cu/ wget https://huggingface.co/karpathy/tinyllamas/resolve/main/stories110M.bin nvcc llama2.cu -o llama2 ./llama2 stories110M.bin ```