PyTorch
llama
alignment-handbook
Generated from Trainer
Mamba2InLlama_0_50 / configs.yaml
Junxiong Wang
add models
5aa60eb
raw
history blame
447 Bytes
Mamba2InLlama_0_50:
prompt_template: "templates/llama3.txt"
fn_completions: "huggingface_local_completions"
completions_kwargs:
model_name: "JunxiongWang/Mamba2InLlama_0_50"
model_kwargs:
torch_dtype: 'bfloat16'
max_new_tokens: 2048
temperature: 0.7
top_p: 1.0
do_sample: True
pretty_name: "Mamba2 0 5 From meta-llama/Meta-Llama-3-8B-Instruct"
link: "https://huggingface.co/JunxiongWang/Mamba2InLlama_0_50"