Error in inference #19
Description
Came across error when running inference (https://github.com/ermongroup/TEOChat?tab=readme-ov-file#-inference) and teochat_demo.py.
Similar to #11 (comment) and #11 (comment), I downloaded the weight of LanguageBind_Image and LanguageBind_Video_merge, modified the corresponding lines in config.json, and annotated the line11-12 in videollava/model/multimodal_encoder/builder.py.
But when I run inference as follow,
from videollava.eval.eval import load_model
from videollava.eval.inference import run_inference_single
tokenizer, model, processor = load_model(model_path="/home/yons/projects/TEOChat/TEOChat-ckpt", model_base=None, load_8bit=True, device='cuda')
image_paths = ["/home/yons/data/xView2_matgo/hold/images/guatemala-volcano_00000004_pre_disaster.png", "/home/yons/data/xView2_matgo/hold/images/guatemala-volcano_00000004_post_disaster.png"]
inp = "These are two satellite images in chronological order: <video> Classify the level of damage experienced by the building."
response = run_inference_single(model, processor, tokenizer, inp, image_paths)
print(response)
The error shows as below,
Some weights of the model checkpoint at /home/yons/projects/TEOChat/TEOChat-ckpt were not used when initializing LlavaLlamaForCausalLM: ['model.image_tower.image_tower.encoder.layers.11.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.9.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.21.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.23.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.5.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.13.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.16.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.14.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.16.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.22.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.11.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.13.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.15.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.15.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.4.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.14.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.4.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.9.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.18.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.2.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.11.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.17.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.12.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.10.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.17.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.16.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.16.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.6.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.6.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.15.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.17.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.9.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.21.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.22.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.19.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.7.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.14.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.4.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.4.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.4.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.14.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.2.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.6.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.18.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.21.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.10.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.16.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.5.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.1.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.7.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.23.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.15.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.13.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.22.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.17.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.9.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.19.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.9.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.11.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.0.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.0.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.23.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.18.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.1.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.3.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.0.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.18.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.1.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.8.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.10.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.0.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.17.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.13.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.20.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.12.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.3.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.5.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.15.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.17.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.9.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.23.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.6.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.16.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.17.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.7.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.7.mlp.fc2.bias', 'model.image_tower.image_tower.post_layernorm.weight', 'model.image_tower.image_tower.encoder.layers.19.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.20.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.20.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.21.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.20.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.8.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.8.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.0.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.3.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.12.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.2.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.1.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.3.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.4.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.4.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.23.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.14.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.11.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.12.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.11.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.20.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.5.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.20.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.11.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.10.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.2.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.7.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.10.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.22.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.11.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.16.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.15.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.8.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.10.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.23.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.2.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.21.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.13.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.4.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.22.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.1.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.22.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.21.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.7.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.4.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.22.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.9.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.2.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.21.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.3.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.0.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.1.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.16.self_attn.k_proj.bias', 'model.image_tower.image_tower.embeddings.class_embedding', 'model.image_tower.image_tower.encoder.layers.12.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.5.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.2.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.2.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.2.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.16.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.0.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.9.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.19.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.11.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.22.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.19.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.6.mlp.fc1.weight', 'model.image_tower.image_tower.embeddings.patch_embedding.weight', 'model.image_tower.image_tower.encoder.layers.19.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.13.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.18.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.15.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.21.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.3.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.15.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.7.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.23.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.22.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.19.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.18.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.19.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.5.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.17.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.0.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.15.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.15.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.13.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.9.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.5.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.15.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.0.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.0.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.17.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.18.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.7.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.19.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.10.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.14.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.8.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.7.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.14.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.1.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.15.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.17.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.20.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.5.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.3.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.3.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.1.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.17.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.20.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.6.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.8.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.5.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.1.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.3.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.10.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.18.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.7.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.13.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.18.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.14.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.14.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.16.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.10.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.23.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.3.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.8.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.2.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.19.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.3.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.15.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.22.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.16.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.7.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.10.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.9.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.2.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.13.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.13.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.23.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.4.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.19.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.14.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.11.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.19.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.10.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.9.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.12.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.17.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.23.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.21.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.1.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.8.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.13.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.21.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.16.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.13.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.8.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.18.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.23.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.10.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.0.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.15.mlp.fc2.bias', 'model.image_tower.image_tower.embeddings.position_embedding.weight', 'model.image_tower.image_tower.encoder.layers.14.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.6.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.13.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.7.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.5.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.10.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.20.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.21.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.12.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.12.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.4.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.12.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.6.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.19.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.22.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.3.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.6.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.10.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.3.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.0.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.22.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.13.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.10.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.21.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.19.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.1.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.7.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.9.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.10.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.12.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.6.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.11.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.11.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.23.mlp.fc2.bias', 'model.image_tower.image_tower.pre_layrnorm.bias', 'model.image_tower.image_tower.encoder.layers.21.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.22.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.2.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.22.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.4.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.15.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.8.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.4.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.4.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.19.self_attn.k_proj.weight', 'model.image_tower.image_tower.pre_layrnorm.weight', 'model.image_tower.image_tower.encoder.layers.12.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.14.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.12.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.23.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.9.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.12.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.23.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.2.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.5.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.11.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.5.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.3.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.7.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.5.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.3.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.17.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.0.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.18.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.5.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.13.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.7.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.11.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.14.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.18.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.13.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.8.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.9.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.6.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.1.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.14.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.20.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.16.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.8.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.18.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.20.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.14.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.7.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.19.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.3.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.6.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.20.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.1.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.6.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.20.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.12.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.2.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.20.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.4.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.12.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.0.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.0.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.23.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.22.self_attn.v_proj.weight', 'model.image_tower.image_tower.encoder.layers.23.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.2.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.12.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.15.self_attn.q_proj.bias', 'model.image_tower.image_tower.encoder.layers.17.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.1.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.6.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.8.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.5.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.1.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.21.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.11.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.11.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.4.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.1.mlp.fc2.bias', 'model.image_tower.image_tower.encoder.layers.5.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.16.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.18.self_attn.k_proj.weight', 'model.image_tower.image_tower.encoder.layers.8.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.20.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.21.self_attn.out_proj.weight', 'model.image_tower.image_tower.encoder.layers.22.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.6.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.16.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.17.layer_norm2.weight', 'model.image_tower.image_tower.encoder.layers.9.mlp.fc2.weight', 'model.image_tower.image_tower.post_layernorm.bias', 'model.image_tower.image_tower.encoder.layers.6.self_attn.out_proj.bias', 'model.image_tower.image_tower.encoder.layers.20.mlp.fc2.weight', 'model.image_tower.image_tower.encoder.layers.2.layer_norm2.bias', 'model.image_tower.image_tower.encoder.layers.21.self_attn.k_proj.bias', 'model.image_tower.image_tower.encoder.layers.16.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.0.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.17.layer_norm1.weight', 'model.image_tower.image_tower.encoder.layers.8.mlp.fc1.bias', 'model.image_tower.image_tower.encoder.layers.9.self_attn.v_proj.bias', 'model.image_tower.image_tower.encoder.layers.18.self_attn.q_proj.weight', 'model.image_tower.image_tower.encoder.layers.14.layer_norm1.bias', 'model.image_tower.image_tower.encoder.layers.18.mlp.fc1.weight', 'model.image_tower.image_tower.encoder.layers.8.layer_norm1.bias']
- This IS expected if you are initializing LlavaLlamaForCausalLM from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).
- This IS NOT expected if you are initializing LlavaLlamaForCausalLM from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).
Traceback (most recent call last):
File "/home/yons/projects/TEOChat/TEOChat-main/TEOChat-main/inference.py", line 28, in
response = run_inference_single(model, processor, tokenizer, inp, image_paths)
File "/home/yons/projects/TEOChat/TEOChat-main/TEOChat-main/videollava/eval/inference.py", line 64, in run_inference_single
output_ids = model.generate(
File "/home/yons/anaconda3/envs/teochat/lib/python3.9/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "/home/yons/anaconda3/envs/teochat/lib/python3.9/site-packages/transformers/generation/utils.py", line 1588, in generate
return self.sample(
File "/home/yons/anaconda3/envs/teochat/lib/python3.9/site-packages/transformers/generation/utils.py", line 2642, in sample
outputs = self(
File "/home/yons/anaconda3/envs/teochat/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/home/yons/anaconda3/envs/teochat/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
return forward_call(*args, **kwargs)
File "/home/yons/anaconda3/envs/teochat/lib/python3.9/site-packages/accelerate/hooks.py", line 165, in new_forward
output = module._old_forward(*args, **kwargs)
File "/home/yons/projects/TEOChat/TEOChat-main/TEOChat-main/videollava/model/language_model/llava_llama.py", line 79, in forward
) = self.prepare_inputs_labels_for_multimodal(
File "/home/yons/projects/TEOChat/TEOChat-main/TEOChat-main/videollava/model/llava_arch.py", line 289, in prepare_inputs_labels_for_multimodal
cur_new_input_embeds = torch.cat(cur_new_input_embeds)
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cuda:1! (when checking argument for argument tensors in method wrapper_CUDA_cat)
Similar weights were not used error information when running teochat_demo.py.