llama2 转换权重

最新推荐文章于 2024-08-06 21:40:02 发布

prinTao

最新推荐文章于 2024-08-06 21:40:02 发布

阅读量788

点赞数 2

文章标签： bug

本文链接：https://blog.csdn.net/prinTao/article/details/134132239

版权

文章描述了如何使用`convert_llama_weights_to_hf.py`脚本将LLAMA模型的权重从指定目录加载到7B大小，并将其保存为HuggingFace的兼容格式。提醒用户注意默认行为和新行为设置的差异。

摘要由CSDN通过智能技术生成

python src/transformers/models/llama/convert_llama_weights_to_hf.py --input_dir  D:\code\Github_code\llama-main  --model_size 7B  --output_dir ./output

(llama) D:\code\Github_code\transformers-main>python src/transformers/models/llama/convert_llama_weights_to_hf.py --input_dir  D:\code\Github_code\llama-main  --model
_size 7B  --output_dir ./output
You are using the default legacy behaviour of the <class 'transformers.models.llama.tokenization_llama.LlamaTokenizer'>. This is expected, and simply means that the `
legacy` (previous) behavior will be used so nothing changes for you. If you want to use the new behaviour, set `legacy=False`. This should only be set if you understa
nd what it means, and thouroughly read the reason why this was added as explained in https://github.com/huggingface/transformers/pull/24565
Fetching all parameters from the checkpoint at D:\code\Github_code\llama-main\7B.
Loading the checkpoint in a Llama model.
Loading checkpoint shards: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████| 33/33 [00:06<00:00,  5.00it/s]
Saving in the Transformers format.