您好，我在跑训练的时候遇见了一个维度错误，不知道如何解决，我使用的是下载的数据集，clip是指定连接下载的clip #9
Comments
|
Could you give more information including the minimal code to reproduce this problem and the environment configuration. |
|
Thank you for your reply.I have addressed this question.I replaced the zero-3,jsonl with zero-2,jsonl.However,I want to repeat your result of the experiment.Can you provide the config of your training please?Thank you! |
|
You can refer to our released paper for more details of reproduction. Feel free to ask if you need further help. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
RuntimeError: weight should have at least three dimensions
return forward_call(*args, **kwargs)
return forward_call(args, **kwargs)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/transformers/models/clip/modeling_clip.py", line 182, in forward
Traceback (most recent call last):
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/conv.py", line 460, in forward
patch_embeds = self.patch_embedding(pixel_values.to(dtype=target_dtype)) # shape = [, width, grid, grid]
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
File "/running_package/code_package/./muffin/train/train_llava15.py", line 338, in
return self._conv_forward(input, self.weight, self.bias)
return self._call_impl(*args, **kwargs)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
train(attn_implementation="flash_attention_2")
File "/running_package/code_package/./muffin/train/train_llava15.py", line 313, in train
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/conv.py", line 456, in _conv_forward
return forward_call(*args, **kwargs)
model, data_module, tokenizer = init_model(
File "/running_package/code_package/./muffin/train/train_llava15.py", line 279, in init_model
return F.conv2d(input, weight, bias, self.stride,
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/conv.py", line 460, in forward
data_module = make_dpo_data_module(tokenizer, data_args=data_args, reference_model=copy.deepcopy(model).cuda())
File "/running_package/code_package/./muffin/train/train_llava15.py", line 149, in make_dpo_data_module
RuntimeError: weight should have at least three dimensions
return self._conv_forward(input, self.weight, self.bias)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/conv.py", line 456, in _conv_forward
train_dataset = DPODataset(tokenizer=tokenizer,
File "/running_package/code_package/./muffin/train/train_llava15.py", line 133, in init
return F.conv2d(input, weight, bias, self.stride,
self.list_data_dict = RLAIFVDataset(data_dir, reference_model, tokenizer,multimodal_cfg['image_token_len'], multimodal_cfg['image_processor'], multimodal_cfg['use_im_start_end'], is_llava15=True)
RuntimeError: weight should have at least three dimensions
File "/running_package/code_package/muffin/data/datasets.py", line 45, in init
inference_logp(reference_model, tokenizer, hf_data, self.data_path,
File "/running_package/code_package/muffin/eval/muffin_inference_logp.py", line 326, in inference_logp
outputs = get_multimodal_sample_logps(model, dataloader, tokenizer, is_llava15=is_llava15) # win_logp_list, win_avg_logp_list, win_per_token_logp_list, rej_logp_list, rej_avg_logp_list, rej_per_token_logp_list
File "/running_package/code_package/muffin/eval/muffin_inference_logp.py", line 243, in get_multimodal_sample_logps
) = model.prepare_inputs_labels_for_multimodal(
File "/running_package/code_package/llava/model/llava_arch.py", line 207, in prepare_inputs_labels_for_multimodal
image_features = self.encode_images(images)
File "/running_package/code_package/llava/model/llava_arch.py", line 146, in encode_images
image_features = self.get_model().get_vision_tower()(images)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "/running_package/code_package/llava/model/multimodal_encoder/clip_encoder.py", line 57, in forward
image_forward_outs = self.vision_tower(images.to(device=self.device, dtype=self.dtype), output_hidden_states=True)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/transformers/models/clip/modeling_clip.py", line 917, in forward
return self.vision_model(
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/transformers/models/clip/modeling_clip.py", line 841, in forward
hidden_states = self.embeddings(pixel_values)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
return forward_call(args, **kwargs)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/transformers/models/clip/modeling_clip.py", line 182, in forward
patch_embeds = self.patch_embedding(pixel_values.to(dtype=target_dtype)) # shape = [, width, grid, grid]
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
return forward_call(*args, **kwargs)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/conv.py", line 460, in forward
return self._conv_forward(input, self.weight, self.bias)
File "/miniconda3/envs/llava_cu122/lib/python3.10/site-packages/torch/nn/modules/conv.py", line 456, in _conv_forward
return F.conv2d(input, weight, bias, self.stride,
RuntimeError: weight should have at least three dimensions
The text was updated successfully, but these errors were encountered: