On-chain llama2.c - Internet Computer

Run on-chain (Internet Computer) with icpp_llm/llama2_c
Run local with karpathy/llama2.c
Try them out at ICGPT
The models were created with the training procedure outlined in karpathy/llama2.c

TinyStories models

model	tokenizer	notes
stories260Ktok512.bin	tok512.bin	Use this for development & debugging
stories15Mtok4096.bin	tok4096.bin	Fits in canister & works well !
stories42Mtok4096.bin	tok4096.bin	As of April 28, hits instruction limit of canister
stories42Mtok32000.bin (*)	tok32000.bin (*)	As of April 28, hits instruction limit of canister
stories110Mtok32000.bin (*)	tok32000.bin (*)	As of April 28, hits instruction limit of canister

(*) Files with asterix behind them were not trained by us, but simply copied from karpathy/tinyllamas and renamed. We are providing them here under a different name for clarity and ease-of-access.

Setup local git with lfs

See: Getting Started: set-up

# install git lfs
# Ubuntu
git lfs install
# Mac
brew install git-lfs

# install huggingface CLI tools in a python environment
pip install huggingface-hub

# Clone this repo
# https
git clone https://huggingface.co/onicai/llama2_c_canister_models
# ssh
git clone git@hf.co:onicai/llama2_c_canister_models

cd llama2_c_canister_models

# configure lfs for local repo
huggingface-cli lfs-enable-largefiles .

# tell lfs what files to track (.gitattributes)
git lfs track "*.bin"

# add, commit & push as usual with git
git add <file-name>
git commit -m "Adding <file-name>"
git push -u origin main