huggingface/candle
1
0
Fork 0
mirror of https://github.com/huggingface/candle.git synced 2025-06-16 10:38:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

Support both llama v1 and llama v2. (#272)

Browse Source
This commit is contained in:
Laurent Mazare
2023-07-28 18:40:59 +01:00
committed by GitHub
parent 7513a5e005
commit 50d8273ae4
2 changed files with 20 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
candle-examples/examples/llama/main.rs
View File

@ -127,7 +127,11 @@ fn main() -> Result<()> {
let args = Args::parse();
let device = candle_examples::device(args.cpu)?;
let config = Config::config_7b(args.use_flash_attn);
let config = if args.v1 {
Config::config_7b_v1(args.use_flash_attn)
} else {
Config::config_7b_v2(args.use_flash_attn)
};
let dtype = if args.use_f32 { DType::F32 } else { DType::F16 };
let cache = model::Cache::new(!args.no_kv_cache, dtype, &config, &device)?;
let (llama, tokenizer_filename) = match args.npy {