search

OpenGVLab / Ask-Anything

[CVPR2024 Highlight][VideoChatGPT] ChatGPT with video understanding! And many more supported LMs such as miniGPT4, StableLM, and MOSS.
https://vchat.opengvlab.com/
MIT License
2.86k stars 230 forks source link

weight-related errors occur when deploying the demo locally #80

Closed XuecWu closed 6 months ago

XuecWu commented 6 months ago

在按照说明文档的要求把相关权重准备好，并进行转换之后，我运行 demo.py，返回如下的信息： Initializing VideoChat Loading checkpoint shards: 100%|██████████████████████████████████████| 3/3 [00:01<00:00, 2.40it/s] Some weights of LlamaForCausalLM were not initialized from the model checkpoint at /home/gsp/cc/Ask-Anything/video_chat2/model_data/vicuna-7b-v0 and are newly initialized: ['model.layers.24.self_attn.rotary_emb.inv_freq', 'model.layers.20.self_attn.rotary_emb.inv_freq', 'model.layers.30.self_attn.rotary_emb.inv_freq', 'model.layers.31.self_attn.rotary_emb.inv_freq', 'model.layers.2.self_attn.rotary_emb.inv_freq', 'model.layers.16.self_attn.rotary_emb.inv_freq', 'model.layers.25.self_attn.rotary_emb.inv_freq', 'model.layers.3.self_attn.rotary_emb.inv_freq', 'model.layers.23.self_attn.rotary_emb.inv_freq', 'model.layers.13.self_attn.rotary_emb.inv_freq', 'model.layers.5.self_attn.rotary_emb.inv_freq', 'model.layers.8.self_attn.rotary_emb.inv_freq', 'model.layers.15.self_attn.rotary_emb.inv_freq', 'model.layers.29.self_attn.rotary_emb.inv_freq', 'model.layers.6.self_attn.rotary_emb.inv_freq', 'model.layers.10.self_attn.rotary_emb.inv_freq', 'model.layers.7.self_attn.rotary_emb.inv_freq', 'model.layers.11.self_attn.rotary_emb.inv_freq', 'model.layers.18.self_attn.rotary_emb.inv_freq', 'model.layers.14.self_attn.rotary_emb.inv_freq', 'model.layers.4.self_attn.rotary_emb.inv_freq', 'model.layers.1.self_attn.rotary_emb.inv_freq', 'model.layers.17.self_attn.rotary_emb.inv_freq', 'model.layers.19.self_attn.rotary_emb.inv_freq', 'model.layers.0.self_attn.rotary_emb.inv_freq', 'model.layers.12.self_attn.rotary_emb.inv_freq', 'model.layers.26.self_attn.rotary_emb.inv_freq', 'model.layers.9.self_attn.rotary_emb.inv_freq', 'model.layers.27.self_attn.rotary_emb.inv_freq', 'model.layers.22.self_attn.rotary_emb.inv_freq', 'model.layers.28.self_attn.rotary_emb.inv_freq', 'model.layers.21.self_attn.rotary_emb.inv_freq'] You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference. _IncompatibleKeys(missing_keys=['llama_model.base_model.model.model.embed_tokens.weight', 'llama_model.base_model.model.model.layers.0.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.0.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.0.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.0.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.0.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.0.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.0.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.0.input_layernorm.weight', 'llama_model.base_model.model.model.layers.0.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.1.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.1.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.1.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.1.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.1.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.1.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.1.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.1.input_layernorm.weight', 'llama_model.base_model.model.model.layers.1.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.2.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.2.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.2.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.2.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.2.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.2.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.2.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.2.input_layernorm.weight', 'llama_model.base_model.model.model.layers.2.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.3.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.3.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.3.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.3.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.3.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.3.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.3.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.3.input_layernorm.weight', 'llama_model.base_model.model.model.layers.3.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.4.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.4.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.4.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.4.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.4.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.4.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.4.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.4.input_layernorm.weight', 'llama_model.base_model.model.model.layers.4.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.5.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.5.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.5.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.5.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.5.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.5.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.5.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.5.input_layernorm.weight', 'llama_model.base_model.model.model.layers.5.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.6.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.6.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.6.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.6.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.6.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.6.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.6.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.6.input_layernorm.weight', 'llama_model.base_model.model.model.layers.6.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.7.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.7.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.7.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.7.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.7.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.7.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.7.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.7.input_layernorm.weight', 'llama_model.base_model.model.model.layers.7.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.8.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.8.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.8.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.8.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.8.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.8.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.8.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.8.input_layernorm.weight', 'llama_model.base_model.model.model.layers.8.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.9.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.9.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.9.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.9.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.9.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.9.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.9.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.9.input_layernorm.weight', 'llama_model.base_model.model.model.layers.9.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.10.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.10.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.10.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.10.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.10.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.10.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.10.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.10.input_layernorm.weight', 'llama_model.base_model.model.model.layers.10.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.11.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.11.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.11.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.11.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.11.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.11.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.11.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.11.input_layernorm.weight', 'llama_model.base_model.model.model.layers.11.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.12.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.12.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.12.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.12.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.12.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.12.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.12.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.12.input_layernorm.weight', 'llama_model.base_model.model.model.layers.12.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.13.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.13.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.13.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.13.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.13.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.13.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.13.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.13.input_layernorm.weight', 'llama_model.base_model.model.model.layers.13.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.14.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.14.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.14.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.14.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.14.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.14.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.14.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.14.input_layernorm.weight', 'llama_model.base_model.model.model.layers.14.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.15.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.15.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.15.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.15.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.15.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.15.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.15.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.15.input_layernorm.weight', 'llama_model.base_model.model.model.layers.15.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.16.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.16.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.16.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.16.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.16.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.16.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.16.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.16.input_layernorm.weight', 'llama_model.base_model.model.model.layers.16.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.17.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.17.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.17.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.17.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.17.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.17.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.17.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.17.input_layernorm.weight', 'llama_model.base_model.model.model.layers.17.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.18.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.18.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.18.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.18.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.18.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.18.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.18.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.18.input_layernorm.weight', 'llama_model.base_model.model.model.layers.18.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.19.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.19.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.19.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.19.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.19.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.19.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.19.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.19.input_layernorm.weight', 'llama_model.base_model.model.model.layers.19.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.20.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.20.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.20.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.20.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.20.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.20.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.20.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.20.input_layernorm.weight', 'llama_model.base_model.model.model.layers.20.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.21.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.21.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.21.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.21.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.21.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.21.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.21.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.21.input_layernorm.weight', 'llama_model.base_model.model.model.layers.21.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.22.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.22.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.22.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.22.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.22.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.22.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.22.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.22.input_layernorm.weight', 'llama_model.base_model.model.model.layers.22.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.23.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.23.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.23.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.23.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.23.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.23.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.23.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.23.input_layernorm.weight', 'llama_model.base_model.model.model.layers.23.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.24.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.24.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.24.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.24.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.24.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.24.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.24.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.24.input_layernorm.weight', 'llama_model.base_model.model.model.layers.24.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.25.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.25.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.25.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.25.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.25.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.25.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.25.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.25.input_layernorm.weight', 'llama_model.base_model.model.model.layers.25.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.26.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.26.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.26.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.26.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.26.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.26.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.26.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.26.input_layernorm.weight', 'llama_model.base_model.model.model.layers.26.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.27.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.27.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.27.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.27.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.27.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.27.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.27.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.27.input_layernorm.weight', 'llama_model.base_model.model.model.layers.27.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.28.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.28.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.28.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.28.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.28.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.28.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.28.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.28.input_layernorm.weight', 'llama_model.base_model.model.model.layers.28.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.29.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.29.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.29.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.29.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.29.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.29.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.29.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.29.input_layernorm.weight', 'llama_model.base_model.model.model.layers.29.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.30.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.30.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.30.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.30.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.30.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.30.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.30.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.30.input_layernorm.weight', 'llama_model.base_model.model.model.layers.30.post_attention_layernorm.weight', 'llama_model.base_model.model.model.layers.31.self_attn.q_proj.weight', 'llama_model.base_model.model.model.layers.31.self_attn.k_proj.weight', 'llama_model.base_model.model.model.layers.31.self_attn.v_proj.weight', 'llama_model.base_model.model.model.layers.31.self_attn.o_proj.weight', 'llama_model.base_model.model.model.layers.31.mlp.gate_proj.weight', 'llama_model.base_model.model.model.layers.31.mlp.down_proj.weight', 'llama_model.base_model.model.model.layers.31.mlp.up_proj.weight', 'llama_model.base_model.model.model.layers.31.input_layernorm.weight', 'llama_model.base_model.model.model.layers.31.post_attention_layernorm.weight', 'llama_model.base_model.model.model.norm.weight', 'llama_model.base_model.model.lm_head.weight'], unexpected_keys=[]) Initialization Finished

请问这是哪里出现了问题？

llama-7b-hf的文件截图如下： image

vicuna-7b-delta-v0的文件截图如下： image

使用的apply_delta.py来自https://huggingface.co/CarperAI/stable-vicuna-13b-delta/raw/main/apply_delta.py

转换后的vicuna-7b-v0的文件截图如下： image

期待您的回复！

Andy1621 commented 6 months ago

Hi! The warnings are normal since we load different parts of models iteratively, except that you meet the unexpected keys.