I directly used the code from QuickStart in the model card, but encountered the following error.
Traceback (most recent call last):
File "example_4khd_chat.py", line 47, in <module>
response, his = model.chat(tokenizer, query=text, image=image, hd_num=55, history=[], do_sample=False, num_beams=3)
File "/home/my/.conda/envs/internlm/lib/python3.8/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "/root/.cache/huggingface/modules/transformers_modules/internlm-xcomposer2-4khd-7b/modeling_internlm_xcomposer2.py", line 556, in chat
outputs = self.generate(
File "/home/my/.conda/envs/internlm/lib/python3.8/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
return func(*args, **kwargs)
File "/home/my/.conda/envs/internlm/lib/python3.8/site-packages/transformers/generation/utils.py", line 1611, in generate
return self.beam_search(
File "/home/my/.conda/envs/internlm/lib/python3.8/site-packages/transformers/generation/utils.py", line 2909, in beam_search
outputs = self(
File "/home/my/.conda/envs/internlm/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/home/my/.conda/envs/internlm/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
return forward_call(*args, **kwargs)
File "/root/.cache/huggingface/modules/transformers_modules/internlm-xcomposer2-4khd-7b/modeling_internlm_xcomposer2.py", line 408, in forward
outputs = self.model(
File "/home/my/.conda/envs/internlm/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/home/my/.conda/envs/internlm/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
return forward_call(*args, **kwargs)
File "/root/.cache/huggingface/modules/transformers_modules/internlm-xcomposer2-4khd-7b/modeling_internlm2.py", line 960, in forward
layer_outputs = decoder_layer(
File "/home/my/.conda/envs/internlm/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/home/my/.conda/envs/internlm/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
return forward_call(*args, **kwargs)
File "/root/.cache/huggingface/modules/transformers_modules/internlm-xcomposer2-4khd-7b/modeling_internlm2.py", line 659, in forward
hidden_states, self_attn_weights, present_key_value = self.attention(
File "/home/my/.conda/envs/internlm/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "/home/my/.conda/envs/internlm/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
return forward_call(*args, **kwargs)
File "/root/.cache/huggingface/modules/transformers_modules/internlm-xcomposer2-4khd-7b/modeling_internlm2.py", line 488, in forward
query_states, key_states = apply_rotary_pos_emb(query_states, key_states, cos, sin, position_ids)
File "/root/.cache/huggingface/modules/transformers_modules/internlm-xcomposer2-4khd-7b/modeling_internlm2.py", line 235, in apply_rotary_pos_emb
q_embed = (q * cos) + (rotate_half(q) * sin)
RuntimeError: The size of tensor a (8302) must match the size of tensor b (8303) at non-singleton dimension 2
../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [4151,0,0], thread: [32,0,0] Assertion `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` failed.
../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [4151,0,0], thread: [33,0,0] Assertion `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` failed.
../aten/src/ATen/native/cuda/IndexKernel.cu:92: operator(): block: [4151,0,0], thread: [34,0,0] Assertion `-sizes[i] <= index && index < sizes[i] && "index out of bounds"` failed.
And I find the position_ids.shape is changing when processing the same image. Is there anyone else facing this error?
I directly used the code from QuickStart in the model card, but encountered the following error.
And I find the
position_ids.shape
is changing when processing the same image. Is there anyone else facing this error?