How to get the embedding_matrix of the model and text or image embeddings?

THUDM / VisualGLM-6B

Chinese and English multimodal conversational language model | 多模态中英双语对话语言模型

Apache License 2.0

4.07k stars 414 forks source link

How to get the embedding_matrix of the model and text or image embeddings? #310

Open tcexeexe opened 10 months ago

tcexeexe commented 10 months ago

Hello~ I want to get the embedding matrix and of the model , and the embeddings when l input a text or image. Can you tell me how to get it? l have tried "model.embed_tokens.weight" and “model.model.embed_tokens(input)” but it failed.