/home/ubuntu/Project/retrieval_projects/csr-v2/CSR_train_and_inference/codebase_for_release/Qwen3-Embedding-4B-retrieval-finetune
#1
by yversley-ebay - opened
The adapter config mentions a path to Qwen3-Embedding-4B-retrieval-finetune - is this the vanilla Qwen3 embedding model? If not, where can it be found?
Yes, this is based on the vanilla Qwen3-4B model (https://huggingface.co/Qwen/Qwen3-Embedding-4B). This path points to the model to be finetuned on and the adapter_model.safetensors file contains the LoRA fine-tuning weights, just as our other models, such as CSRv2-classification (https://huggingface.co/Y-Research-Group/CSRv2-classification). Thank you for your comment and we will modify the config file soon.