I’m trying to finetune the CLIP model for my private dataset. I have written CustomDataset for that, the data loading is working well but when I try to train the model with Trainer, It raises IndexError: index out of range in self
Custom Dataset
```
class ClassifierDataset(Dataset):
def init(self, processor, folds: int, mode=“train”):
super(ClassifierDataset, self).init()
self.mode = mode
self.fold = folds
self.processor = processor
self.max_target_length = 32
self.train_transform = A.Compose(
[
A.CLAHE(),
A.RandomRotate90(),
A.Transpose(),
A.ShiftScaleRotate(
shift_limit=0.0625, scale_limit=0.50, rotate_limit=45, p=0.75
),
A.Blur(blur_limit=3),
A.OpticalDistortion(),
A.GridDistortion(),
A.HueSaturationValue(),
A.VerticalFlip(),
A.HorizontalFlip(),
A.Normalize(),
ToTensorV2(),
]
)
self.val_transform = A.Compose(
[
A.Normalize(),
ToTensorV2(),
]
)
self.df = pd.read_csv(str(Path.cwd() / "data_new.csv"))
self.train_df = self.df[self.df.folds != self.fold].reset_index(drop=True)
self.val_df = self.df[self.df.folds == self.fold].reset_index(drop=True)
self.dff = self.train_df if self.mode == "train" else self.val_df
def __len__(self):
return len(self.dff)
def __getitem__(self, index):
img_name = self.dff.loc[index, "paths"]
text = self.dff.loc[index, "corrected_prompts"]
img = Image.open(img_name).convert("RGB")
pixel_values = self.processor.feature_extractor(img, return_tensors="pt").pixel_values
labels = self.processor.tokenizer(text,
padding="max_length",
max_length=77,
truncation=True).input_ids
labels = [label if label != self.processor.tokenizer.pad_token_id else -100 for label in labels]
return {"input_ids":torch.tensor(labels), "pixel_values":pixel_values.squeeze()}
## Trainer
default_data_collator = DefaultDataCollator()
training_args = TrainingArguments(
output_dir=‘./results’,
num_train_epochs=3,
warmup_steps=500,
weight_decay=0.01,
logging_dir=‘./logs’,
logging_steps=10,
evaluation_strategy=‘epoch’
)
trainer = Trainer(
model=model,
args=training_args,
train_dataset=train_data,
eval_dataset=val_data,
data_collator=default_data_collator
)
trainer.train()
## The Error StackTrace
in <cell line: 22>:22 │
│ │
│ 19 │ data_collator=default_data_collator │
│ 20 ) │
│ 21 │
│ ❱ 22 trainer.train() │
│ 23 │
│ │
│ /opt/conda/envs/blip/lib/python3.8/site-packages/transformers/trainer.py:1543 in train │
│ │
│ 1540 │ │ inner_training_loop = find_executable_batch_size( │
│ 1541 │ │ │ self._inner_training_loop, self._train_batch_size, args.auto_find_batch_size │
│ 1542 │ │ ) │
│ ❱ 1543 │ │ return inner_training_loop( │
│ 1544 │ │ │ args=args, │
│ 1545 │ │ │ resume_from_checkpoint=resume_from_checkpoint, │
│ 1546 │ │ │ trial=trial, │
│ │
│ /opt/conda/envs/blip/lib/python3.8/site-packages/transformers/trainer.py:1791 in │
│ _inner_training_loop │
│ │
│ 1788 │ │ │ │ │ with model.no_sync(): │
│ 1789 │ │ │ │ │ │ tr_loss_step = self.training_step(model, inputs) │
│ 1790 │ │ │ │ else: │
│ ❱ 1791 │ │ │ │ │ tr_loss_step = self.training_step(model, inputs) │
│ 1792 │ │ │ │ │
│ 1793 │ │ │ │ if ( │
│ 1794 │ │ │ │ │ args.logging_nan_inf_filter │
│ │
│ /opt/conda/envs/blip/lib/python3.8/site-packages/transformers/trainer.py:2539 in training_step │
│ │
│ 2536 │ │ │ return loss_mb.reduce_mean().detach().to(self.args.device) │
│ 2537 │ │ │
│ 2538 │ │ with self.compute_loss_context_manager(): │
│ ❱ 2539 │ │ │ loss = self.compute_loss(model, inputs) │
│ 2540 │ │ │
│ 2541 │ │ if self.args.n_gpu > 1: │
│ 2542 │ │ │ loss = loss.mean() # mean() to average on multi-gpu parallel training │
│ │
│ /opt/conda/envs/blip/lib/python3.8/site-packages/transformers/trainer.py:2571 in compute_loss │
│ │
│ 2568 │ │ │ labels = inputs.pop(“labels”) │
│ 2569 │ │ else: │
│ 2570 │ │ │ labels = None │
│ ❱ 2571 │ │ outputs = model(**inputs) │
│ 2572 │ │ # Save past state if it exists │
│ 2573 │ │ # TODO: this needs to be fixed and made cleaner later. │
│ 2574 │ │ if self.args.past_index >= 0: │
│ │
│ /opt/conda/envs/blip/lib/python3.8/site-packages/torch/nn/modules/module.py:1110 in _call_impl │
│ │
│ 1107 │ │ # this function, and just call forward. │
│ 1108 │ │ if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks o │
│ 1109 │ │ │ │ or _global_forward_hooks or _global_forward_pre_hooks): │
│ ❱ 1110 │ │ │ return forward_call(*input, **kwargs) │
│ 1111 │ │ # Do not call functions when jit is used │
│ 1112 │ │ full_backward_hooks, non_full_backward_hooks = , │
│ 1113 │ │ if self._backward_hooks or _global_backward_hooks: │
│ │
│ /opt/conda/envs/blip/lib/python3.8/site-packages/transformers/models/clip/modeling_clip.py:1125 │
│ in forward │
│ │
│ 1122 │ │ │ return_dict=return_dict, │
│ 1123 │ │ ) │
│ 1124 │ │ │
│ ❱ 1125 │ │ text_outputs = self.text_model( │
│ 1126 │ │ │ input_ids=input_ids, │
│ 1127 │ │ │ attention_mask=attention_mask, │
│ 1128 │ │ │ position_ids=position_ids, │
│ │
│ /opt/conda/envs/blip/lib/python3.8/site-packages/torch/nn/modules/module.py:1110 in _call_impl │
│ │
│ 1107 │ │ # this function, and just call forward. │
│ 1108 │ │ if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks o │
│ 1109 │ │ │ │ or _global_forward_hooks or _global_forward_pre_hooks): │
│ ❱ 1110 │ │ │ return forward_call(*input, **kwargs) │
│ 1111 │ │ # Do not call functions when jit is used │
│ 1112 │ │ full_backward_hooks, non_full_backward_hooks = , │
│ 1113 │ │ if self._backward_hooks or _global_backward_hooks: │
│ │
│ /opt/conda/envs/blip/lib/python3.8/site-packages/transformers/models/clip/modeling_clip.py:712 │
│ in forward │
│ │
│ 709 │ │ input_shape = input_ids.size() │
│ 710 │ │ input_ids = input_ids.view(-1, input_shape[-1]) │
│ 711 │ │ │
│ ❱ 712 │ │ hidden_states = self.embeddings(input_ids=input_ids, position_ids=position_ids) │
│ 713 │ │ │
│ 714 │ │ bsz, seq_len = input_shape │
│ 715 │ │ # CLIP’s text model uses causal mask, prepare it here. │
│ │
│ /opt/conda/envs/blip/lib/python3.8/site-packages/torch/nn/modules/module.py:1110 in _call_impl │
│ │
│ 1107 │ │ # this function, and just call forward. │
│ 1108 │ │ if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks o │
│ 1109 │ │ │ │ or _global_forward_hooks or _global_forward_pre_hooks): │
│ ❱ 1110 │ │ │ return forward_call(*input, **kwargs) │
│ 1111 │ │ # Do not call functions when jit is used │
│ 1112 │ │ full_backward_hooks, non_full_backward_hooks = , │
│ 1113 │ │ if self._backward_hooks or _global_backward_hooks: │
│ │
│ /opt/conda/envs/blip/lib/python3.8/site-packages/transformers/models/clip/modeling_clip.py:227 │
│ in forward │
│ │
│ 224 │ │ │ position_ids = self.position_ids[:, :seq_length] │
│ 225 │ │ │
│ 226 │ │ if inputs_embeds is None: │
│ ❱ 227 │ │ │ inputs_embeds = self.token_embedding(input_ids) │
│ 228 │ │ │
│ 229 │ │ position_embeddings = self.position_embedding(position_ids) │
│ 230 │ │ embeddings = inputs_embeds + position_embeddings │
│ │
│ /opt/conda/envs/blip/lib/python3.8/site-packages/torch/nn/modules/module.py:1110 in _call_impl │
│ │
│ 1107 │ │ # this function, and just call forward. │
│ 1108 │ │ if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks o │
│ 1109 │ │ │ │ or _global_forward_hooks or _global_forward_pre_hooks): │
│ ❱ 1110 │ │ │ return forward_call(*input, **kwargs) │
│ 1111 │ │ # Do not call functions when jit is used │
│ 1112 │ │ full_backward_hooks, non_full_backward_hooks = , │
│ 1113 │ │ if self.backward_hooks or global_backward_hooks: │
│ │
│ /opt/conda/envs/blip/lib/python3.8/site-packages/torch/nn/modules/sparse.py:158 in forward │
│ │
│ 155 │ │ │ │ self.weight[self.padding_idx].fill(0) │
│ 156 │ │
│ 157 │ def forward(self, input: Tensor) → Tensor: │
│ ❱ 158 │ │ return F.embedding( │
│ 159 │ │ │ input, self.weight, self.padding_idx, self.max_norm, │
│ 160 │ │ │ self.norm_type, self.scale_grad_by_freq, self.sparse) │
│ 161 │
│ │
│ /opt/conda/envs/blip/lib/python3.8/site-packages/torch/nn/functional.py:2183 in embedding │
│ │
│ 2180 │ │ # torch.embedding_renorm │
│ 2181 │ │ # remove once script supports set_grad_enabled │
│ 2182 │ │ no_grad_embedding_renorm(weight, input, max_norm, norm_type) │
│ ❱ 2183 │ return torch.embedding(weight, input, padding_idx, scale_grad_by_freq, sparse) │
│ 2184 │
│ 2185 │
│ 2186 def embedding_bag( │
╰──────────────────────────────────────────────────────────────────────────────────────────────────╯
IndexError: index out of range in self