vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

feat: Add support for qwen tokenizer

Browse source 
Signed-off-by: teleprint-me <77757836+teleprint-me@users.noreply.github.com>
This commit is contained in:
teleprint-me 2024-05-07 21:47:59 -04:00
parent 3ae6c17067
commit 4155e86ba0
No known key found for this signature in database
GPG key ID: B0D11345E65C4D48
1 changed files with 14 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

21
convert-hf-to-gguf-update.py
View file

@ -119,13 +119,20 @@ for model in models:
# model and repo urls are not the same # model and repo urls are not the same
# url = "https://huggingface.co/Qwen/Qwen-tokenizer/raw/main/tokenizer.json" # url = "https://huggingface.co/Qwen/Qwen-tokenizer/raw/main/tokenizer.json"
if name == "qwen": # qwen is an outlier and will raise a FileNotFoundError
# Get the models tokenizer # fetch the qwen tokenizer
download_file_with_auth( download_file_with_auth(
url=f"{url_main}/tokenizer.json", url="https://huggingface.co/Qwen/Qwen-tokenizer/raw/main/tokenizer.json",
token=token, token=token,
save_path=model_tokenizer_path save_path=model_tokenizer_path
) )
else: # Et tu, Brute?
# Get the models tokenizer
download_file_with_auth(
url=f"{url_main}/tokenizer.json",
token=token,
save_path=model_tokenizer_path
)
# Get the models hyper params # Get the models hyper params
download_file_with_auth( download_file_with_auth(