search

epfl-nlp / kogito

A Python Commonsense Knowledge Inference Toolkit
https://kogito.readthedocs.io
Apache License 2.0
60 stars 6 forks source link

RuntimeError: element 0 of tensors does not require grad and does not have a grad_fn #2

Open ZornWang opened 9 months ago

ZornWang commented 9 months ago

Hi, thanks for the fantastic work. I have a problem here. I want to train COMET on my own dataset, and I follow the user guide from here docs. My codes are as follows:

import os

from kogito.core.knowledge import KnowledgeGraph
from kogito.models.bart.comet import COMETBART, COMETBARTConfig, AutoTokenizer

path = 'data/csv'

config = COMETBARTConfig(
    output_dir="bart",
    num_workers=16,
    learning_rate=1e-5,
    gpus=1,
    sortish_sampler=True,
    atomic=True,
    max_epochs=56,
    pretrained_model="facebook/bart-large",
)
model = COMETBART(config)
model.tokenizer = AutoTokenizer.from_pretrained("mismayil/comet-bart-ai2")

train_graph = KnowledgeGraph.from_csv(os.path.join(path, 'train.csv'))
val_graph = KnowledgeGraph.from_csv(os.path.join(path, 'val.csv'))
test_graph = KnowledgeGraph.from_csv(os.path.join(path, 'test.csv'))

model.train(train_graph=train_graph, val_graph=val_graph, test_graph=test_graph)

# Save as a pretrained model
model.save_pretrained("comet-bart/v1")

However, i got following errors when i run the code:

  File "/home/wzh/work/comet_ft/cometbart_train.py", line 25, in <module>
    model.train(train_graph=train_graph, val_graph=val_graph, test_graph=test_graph)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/kogito/models/bart/comet.py", line 109, in train
    trainer: pl.Trainer = generic_train(
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/kogito/models/bart/utils.py", line 262, in generic_train
    trainer.fit(model)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 740, in fit
    self._call_and_handle_interrupt(
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 685, in _call_and_handle_interrupt
    return trainer_fn(*args, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 777, in _fit_impl
    self._run(model, ckpt_path=ckpt_path)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 1199, in _run
    self._dispatch()
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 1279, in _dispatch
    self.training_type_plugin.start_training(self)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/plugins/training_type/training_type_plugin.py", line 202, in start_training
    self._results = trainer.run_stage()
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 1289, in run_stage
    return self._run_train()
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 1319, in _run_train
    self.fit_loop.run()
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/loops/base.py", line 145, in run
    self.advance(*args, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/loops/fit_loop.py", line 234, in advance
    self.epoch_loop.run(data_fetcher)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/loops/base.py", line 145, in run
    self.advance(*args, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/loops/epoch/training_epoch_loop.py", line 193, in advance
    batch_output = self.batch_loop.run(batch, batch_idx)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/loops/base.py", line 145, in run
    self.advance(*args, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/loops/batch/training_batch_loop.py", line 88, in advance
    outputs = self.optimizer_loop.run(split_batch, optimizers, batch_idx)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/loops/base.py", line 145, in run
    self.advance(*args, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/loops/optimization/optimizer_loop.py", line 215, in advance
    result = self._run_optimization(
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/loops/optimization/optimizer_loop.py", line 266, in _run_optimization
    self._optimizer_step(optimizer, opt_idx, batch_idx, closure)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/loops/optimization/optimizer_loop.py", line 378, in _optimizer_step
    lightning_module.optimizer_step(
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/core/lightning.py", line 1652, in optimizer_step
    optimizer.step(closure=optimizer_closure)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/core/optimizer.py", line 164, in step
    trainer.accelerator.optimizer_step(self._optimizer, self._optimizer_idx, closure, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/accelerators/accelerator.py", line 339, in optimizer_step
    self.precision_plugin.optimizer_step(model, optimizer, opt_idx, closure, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/plugins/precision/precision_plugin.py", line 163, in optimizer_step
    optimizer.step(closure=closure, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/torch/optim/lr_scheduler.py", line 68, in wrapper
    return wrapped(*args, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/torch/optim/optimizer.py", line 140, in wrapper
    out = func(*args, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/torch/autograd/grad_mode.py", line 27, in decorate_context
    return func(*args, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/transformers/optimization.py", line 457, in step
    loss = closure()
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/plugins/precision/precision_plugin.py", line 148, in _wrap_closure
    closure_result = closure()
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/loops/optimization/optimizer_loop.py", line 160, in __call__
    self._result = self.closure(*args, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/loops/optimization/optimizer_loop.py", line 155, in closure
    self._backward_fn(step_output.closure_loss)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/loops/optimization/optimizer_loop.py", line 327, in backward_fn
    self.trainer.accelerator.backward(loss, optimizer, opt_idx)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/accelerators/accelerator.py", line 314, in backward
    self.precision_plugin.backward(self.lightning_module, closure_loss, *args, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/plugins/precision/precision_plugin.py", line 91, in backward
    model.backward(closure_loss, optimizer, *args, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/pytorch_lightning/core/lightning.py", line 1434, in backward
    loss.backward(*args, **kwargs)
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/torch/_tensor.py", line 488, in backward
    torch.autograd.backward(
  File "/home/wzh/anaconda3/envs/kogito/lib/python3.10/site-packages/torch/autograd/__init__.py", line 197, in backward
    Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
RuntimeError: element 0 of tensors does not require grad and does not have a grad_fn

Here is my environment：

I do not kown why, any reply will be appreciated. Thanks.

ZornWang commented 9 months ago

BTW, I can't run the examples 'examples/train_comet_bart.py' directly, it will report the following error:

ValueError: Usecols do not match columns, columns expected but not found: ['relation', 'tails', 'head']

Here is the code that went wrong:

train_graph = KnowledgeGraph.from_csv(
    "data/atomic2020/train.tsv", header=None, sep="\t"
)

It seems like header=None doesn't work

ZornWang commented 9 months ago

It seems like kogito or source code of COMET use the old version of pytorch which is not support the Ampere architecture GPUs, the code can run on V100 which belongs to Volta architecture. But it still has some problem with numpy==1.24.1, it will got following errors:

ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (2026,) + inhomogeneous part.

Replace numpy==1.24.1 it with older version like numpy==1.21.2 can solve this problem but still have warrings like this:

<__array_function__ internals>:5: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.

Still, it works.

mismayil commented 9 months ago

BTW, I can't run the examples 'examples/train_comet_bart.py' directly, it will report the following error:

ValueError: Usecols do not match columns, columns expected but not found: ['relation', 'tails', 'head']

Here is the code that went wrong:

train_graph = KnowledgeGraph.from_csv(
    "data/atomic2020/train.tsv", header=None, sep="\t"
)

It seems like header=None doesn't work

hi Zorn,

Thanks for pointing this out. It indeed seems to be a bug and I pushed a bug to fix it.

mismayil commented 9 months ago

It seems like kogito or source code of COMET use the old version of pytorch which is not support the Ampere architecture GPUs, the code can run on V100 which belongs to Volta architecture. But it still has some problem with numpy==1.24.1, it will got following errors:

ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (2026,) + inhomogeneous part.

Replace numpy==1.24.1 it with older version like numpy==1.21.2 can solve this problem but still have warrings like this:

<__array_function__ internals>:5: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.

Still, it works.

Regarding the other errors, unfortunately, since we do not actively maintain this repo, it might not support dependency updates. Please, try to use the original dependencies as much as possible and in case you can't and you find a solution, feel free to open a pull request.

ZornWang commented 9 months ago

It seems like kogito or source code of COMET use the old version of pytorch which is not support the Ampere architecture GPUs, the code can run on V100 which belongs to Volta architecture. But it still has some problem with numpy==1.24.1, it will got following errors:

ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (2026,) + inhomogeneous part.

Replace numpy==1.24.1 it with older version like numpy==1.21.2 can solve this problem but still have warrings like this:

<__array_function__ internals>:5: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.

Still, it works.

Regarding the other errors, unfortunately, since we do not actively maintain this repo, it might not support dependency updates. Please, try to use the original dependencies as much as possible and in case you can't and you find a solution, feel free to open a pull request.

Very grateful for your reply, btw, may i ask for a requirements with specific python version and package version list? Use pip install kogito can't work directly, it will automatic install numpy==1.24.1 when python==3.10

mismayil commented 9 months ago

It seems like kogito or source code of COMET use the old version of pytorch which is not support the Ampere architecture GPUs, the code can run on V100 which belongs to Volta architecture. But it still has some problem with numpy==1.24.1, it will got following errors:

ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (2026,) + inhomogeneous part.

Replace numpy==1.24.1 it with older version like numpy==1.21.2 can solve this problem but still have warrings like this:

<__array_function__ internals>:5: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.

Still, it works.

Regarding the other errors, unfortunately, since we do not actively maintain this repo, it might not support dependency updates. Please, try to use the original dependencies as much as possible and in case you can't and you find a solution, feel free to open a pull request.

Very grateful for your reply, btw, may i ask for a requirements with specific python version and package version list? Use pip install kogito can't work directly, it will automatic install numpy==1.24.1 when python==3.10

I see, in that case it seems like we need to resolve the issue so that new numpy version is supported on this architecture. Could you maybe share the full stack trace of the error you get with numpy==1.24.1?

ZornWang commented 9 months ago

I see, in that case it seems like we need to resolve the issue so that new numpy version is supported on this architecture. Could you maybe share the full stack trace of the error you get with numpy==1.24.1?

I try two versions of python on my device, and unfortunately, the subject of this issue is occurred, here is the problem of run the scripts examples/train_comet_bart.py: Here is my hardware environment:

Ubuntu 20.04
Nvidia Tesla V100
Driver Version: 535.54.03
Cuda 12.2
  1. Just use conda create -n env python=3.10 and pip install kogito to configure the environment. The cuda versoon of torch which is automatically installed : 
    Python 3.10.13 (main, Sep 11 2023, 13:44:35) [GCC 11.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> torch.version.cuda
'11.7'

    Here is the output when run pip list:

    absl-py                  2.0.0
aiohttp                  3.9.1
aiosignal                1.3.1
annotated-types          0.6.0
async-timeout            4.0.3
attrs                    23.1.0
bert-score               0.3.13
blis                     0.7.11
cachetools               5.3.2
catalogue                2.0.10
certifi                  2023.11.17
charset-normalizer       3.3.2
click                    8.1.7
cloudpathlib             0.16.0
colorama                 0.4.6
confection               0.1.4
contourpy                1.2.0
cycler                   0.12.1
cymem                    2.0.8
docker-pycreds           0.4.0
et-xmlfile               1.1.0
filelock                 3.13.1
fonttools                4.46.0
frozenlist               1.4.0
fsspec                   2023.12.2
future                   0.18.3
gitdb                    4.0.11
GitPython                3.1.40
google-auth              2.25.2
google-auth-oauthlib     1.2.0
grpcio                   1.51.3
huggingface-hub          0.19.4
idna                     3.6
inflect                  5.6.2
Jinja2                   3.1.2
joblib                   1.3.2
kiwisolver               1.4.5
kogito                   0.6.2
langcodes                3.3.0
lightning-utilities      0.10.0
lxml                     4.9.3
Markdown                 3.5.1
MarkupSafe               2.1.3
matplotlib               3.8.2
multidict                6.0.4
murmurhash               1.0.10
nltk                     3.8.1
numpy                    1.26.2
nvidia-cublas-cu11       11.10.3.66
nvidia-cuda-nvrtc-cu11   11.7.99
nvidia-cuda-runtime-cu11 11.7.99
nvidia-cudnn-cu11        8.5.0.96
oauthlib                 3.2.2
openai                   0.18.1
openpyxl                 3.1.2
packaging                23.2
pandas                   1.5.3
pandas-stubs             2.1.1.230928
pathtools                0.1.2
Pillow                   10.1.0
pip                      23.3.1
portalocker              2.8.2
preshed                  3.0.9
promise                  2.3
protobuf                 3.20.3
psutil                   5.9.6
pyasn1                   0.5.1
pyasn1-modules           0.3.0
pydantic                 2.5.2
pydantic_core            2.14.5
pyDeprecate              0.3.1
pyparsing                3.1.1
python-dateutil          2.8.2
pytorch-lightning        1.5.10
pytz                     2023.3.post1
PyYAML                   6.0.1
regex                    2023.10.3
requests                 2.31.0
requests-oauthlib        1.3.1
rouge-score              0.0.4
rsa                      4.9
sacrebleu                2.4.0
safetensors              0.4.1
sentencepiece            0.1.99
sentry-sdk               1.39.1
setproctitle             1.3.3
setuptools               59.5.0
shortuuid                1.0.11
six                      1.16.0
smart-open               6.4.0
smmap                    5.0.1
spacy                    3.7.2
spacy-legacy             3.0.12
spacy-loggers            1.0.5
srsly                    2.4.8
tabulate                 0.9.0
tensorboard              2.15.1
tensorboard-data-server  0.7.2
thinc                    8.2.1
tokenizers               0.15.0
torch                    1.13.1
torchmetrics             1.2.1
tqdm                     4.66.1
transformers             4.36.1
typer                    0.9.0
types-pytz               2023.3.1.1
typing_extensions        4.9.0
urllib3                  2.1.0
wandb                    0.12.21
wasabi                   1.1.2
weasel                   0.3.4
Werkzeug                 3.0.1
wheel                    0.41.2
yarl                     1.9.4

    And here is the full stack trace:

    Global seed set to 42
/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/trainer/connectors/callback_connector.py:147: LightningDeprecationWarning: Setting `Trainer(checkpoint_callback=<pytorch_lightning.callbacks.model_checkpoint.ModelCheckpoint object at 0x7f1634d0ff40>)` is deprecated in v1.5 and will be removed in v1.7. Please consider using `Trainer(enable_checkpointing=<pytorch_lightning.callbacks.model_checkpoint.ModelCheckpoint object at 0x7f1634d0ff40>)`.
rank_zero_deprecation(
/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/trainer/connectors/callback_connector.py:167: LightningDeprecationWarning: Setting `Trainer(weights_summary=None)` is deprecated in v1.5 and will be removed in v1.7. Please set `Trainer(enable_model_summary=False)` instead.
rank_zero_deprecation(
GPU available: True, used: False
TPU available: False, using: 0 TPU cores
IPU available: False, using: 0 IPUs
/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py:1584: UserWarning: GPU available but not used. Set the gpus flag in your trainer `Trainer(gpus=1)` or script `--gpus=1`.
rank_zero_warn(
/root/miniconda3/envs/peak/lib/python3.10/site-packages/transformers/optimization.py:429: FutureWarning: This implementation of AdamW is deprecated and will be removed in a future version. Use the PyTorch implementation torch.optim.AdamW instead, or set `no_deprecation_warning=True` to disable this warning
warnings.warn(
Validation sanity check: 0it [00:00, ?it/s]/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/trainer/data_loading.py:132: UserWarning: The dataloader, val_dataloader 0, does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` (try 52 which is the number of cpus on this machine) in the `DataLoader` init to improve performance.
rank_zero_warn(
Validation sanity check:   0%|                                                                                                                                                                                                                                                         | 0/2 [00:00<?, ?it/s]/root/miniconda3/envs/peak/lib/python3.10/site-packages/transformers/generation/utils.py:1355: UserWarning: Using the model-agnostic default `max_length` (=20) to control the generation length. We recommend setting `max_new_tokens` to control the maximum length of the generation.
warnings.warn(
Global seed set to 42                                                                                                                                                                                                                                                                                        
/root/miniconda3/envs/peak/lib/python3.10/site-packages/kogito/models/bart/utils.py:490: UserWarning: All learning rates are 0
warnings.warn("All learning rates are 0")
/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/trainer/data_loading.py:132: UserWarning: The dataloader, train_dataloader, does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` (try 52 which is the number of cpus on this machine) in the `DataLoader` init to improve performance.
rank_zero_warn(
Epoch 0:   0%|                                                                                                                                                                                                                                                                      | 0/3126 [00:00<?, ?it/s]Traceback (most recent call last):
File "/root/autodl-tmp/kogito-main/examples/train_comet_bart.py", line 39, in <module>
model.train(
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/kogito/models/bart/comet.py", line 109, in train
trainer: pl.Trainer = generic_train(
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/kogito/models/bart/utils.py", line 262, in generic_train
trainer.fit(model)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 740, in fit
self._call_and_handle_interrupt(
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 685, in _call_and_handle_interrupt
return trainer_fn(*args, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 777, in _fit_impl
self._run(model, ckpt_path=ckpt_path)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 1199, in _run
self._dispatch()
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 1279, in _dispatch
self.training_type_plugin.start_training(self)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/plugins/training_type/training_type_plugin.py", line 202, in start_training
self._results = trainer.run_stage()
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 1289, in run_stage
return self._run_train()
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 1319, in _run_train
self.fit_loop.run()
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/loops/base.py", line 145, in run
self.advance(*args, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/loops/fit_loop.py", line 234, in advance
self.epoch_loop.run(data_fetcher)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/loops/base.py", line 145, in run
self.advance(*args, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/loops/epoch/training_epoch_loop.py", line 193, in advance
batch_output = self.batch_loop.run(batch, batch_idx)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/loops/base.py", line 145, in run
self.advance(*args, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/loops/batch/training_batch_loop.py", line 88, in advance
outputs = self.optimizer_loop.run(split_batch, optimizers, batch_idx)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/loops/base.py", line 145, in run
self.advance(*args, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/loops/optimization/optimizer_loop.py", line 215, in advance
result = self._run_optimization(
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/loops/optimization/optimizer_loop.py", line 266, in _run_optimization
self._optimizer_step(optimizer, opt_idx, batch_idx, closure)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/loops/optimization/optimizer_loop.py", line 378, in _optimizer_step
lightning_module.optimizer_step(
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/core/lightning.py", line 1652, in optimizer_step
optimizer.step(closure=optimizer_closure)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/core/optimizer.py", line 164, in step
trainer.accelerator.optimizer_step(self._optimizer, self._optimizer_idx, closure, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/accelerators/accelerator.py", line 339, in optimizer_step
self.precision_plugin.optimizer_step(model, optimizer, opt_idx, closure, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/plugins/precision/precision_plugin.py", line 163, in optimizer_step
optimizer.step(closure=closure, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/torch/optim/lr_scheduler.py", line 68, in wrapper
return wrapped(*args, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/torch/optim/optimizer.py", line 140, in wrapper
out = func(*args, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/torch/autograd/grad_mode.py", line 27, in decorate_context
return func(*args, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/transformers/optimization.py", line 457, in step
loss = closure()
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/plugins/precision/precision_plugin.py", line 148, in _wrap_closure
closure_result = closure()
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/loops/optimization/optimizer_loop.py", line 160, in __call__
self._result = self.closure(*args, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/loops/optimization/optimizer_loop.py", line 155, in closure
self._backward_fn(step_output.closure_loss)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/loops/optimization/optimizer_loop.py", line 327, in backward_fn
self.trainer.accelerator.backward(loss, optimizer, opt_idx)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/accelerators/accelerator.py", line 314, in backward
self.precision_plugin.backward(self.lightning_module, closure_loss, *args, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/plugins/precision/precision_plugin.py", line 91, in backward
model.backward(closure_loss, optimizer, *args, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/pytorch_lightning/core/lightning.py", line 1434, in backward
loss.backward(*args, **kwargs)
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/torch/_tensor.py", line 488, in backward
torch.autograd.backward(
File "/root/miniconda3/envs/peak/lib/python3.10/site-packages/torch/autograd/__init__.py", line 197, in backward
Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
RuntimeError: element 0 of tensors does not require grad and does not have a grad_fn
  2. Just use conda create -n env python=3.8 and pip install kogito to configure the environment. The cuda versoon of torch which is automatically installed is also 11.7 Here is the output when run pip list: 
    absl-py                  2.0.0
aiohttp                  3.9.1
aiosignal                1.3.1
annotated-types          0.6.0
async-timeout            4.0.3
attrs                    23.1.0
bert-score               0.3.13
blis                     0.7.11
cachetools               5.3.2
catalogue                2.0.10
certifi                  2023.11.17
charset-normalizer       3.3.2
click                    8.1.7
cloudpathlib             0.16.0
colorama                 0.4.6
confection               0.1.4
contourpy                1.1.1
cycler                   0.12.1
cymem                    2.0.8
docker-pycreds           0.4.0
et-xmlfile               1.1.0
filelock                 3.13.1
fonttools                4.46.0
frozenlist               1.4.0
fsspec                   2023.12.2
future                   0.18.3
gitdb                    4.0.11
GitPython                3.1.40
google-auth              2.25.2
google-auth-oauthlib     1.0.0
grpcio                   1.51.3
huggingface-hub          0.19.4
idna                     3.6
importlib-metadata       7.0.0
importlib-resources      6.1.1
inflect                  5.6.2
Jinja2                   3.1.2
joblib                   1.3.2
kiwisolver               1.4.5
kogito                   0.6.2
langcodes                3.3.0
lightning-utilities      0.10.0
lxml                     4.9.3
Markdown                 3.5.1
MarkupSafe               2.1.3
matplotlib               3.7.4
multidict                6.0.4
murmurhash               1.0.10
nltk                     3.8.1
numpy                    1.24.4
nvidia-cublas-cu11       11.10.3.66
nvidia-cuda-nvrtc-cu11   11.7.99
nvidia-cuda-runtime-cu11 11.7.99
nvidia-cudnn-cu11        8.5.0.96
oauthlib                 3.2.2
openai                   0.18.1
openpyxl                 3.1.2
packaging                23.2
pandas                   1.5.3
pandas-stubs             2.0.3.230814
pathtools                0.1.2
Pillow                   10.1.0
pip                      23.3.1
portalocker              2.8.2
preshed                  3.0.9
promise                  2.3
protobuf                 3.20.3
psutil                   5.9.6
pyasn1                   0.5.1
pyasn1-modules           0.3.0
pydantic                 2.5.2
pydantic_core            2.14.5
pyDeprecate              0.3.1
pyparsing                3.1.1
python-dateutil          2.8.2
pytorch-lightning        1.5.10
pytz                     2023.3.post1
PyYAML                   6.0.1
regex                    2023.10.3
requests                 2.31.0
requests-oauthlib        1.3.1
rouge-score              0.0.4
rsa                      4.9
sacrebleu                2.4.0
safetensors              0.4.1
sentencepiece            0.1.99
sentry-sdk               1.39.1
setproctitle             1.3.3
setuptools               59.5.0
shortuuid                1.0.11
six                      1.16.0
smart-open               6.4.0
smmap                    5.0.1
spacy                    3.7.2
spacy-legacy             3.0.12
spacy-loggers            1.0.5
srsly                    2.4.8
tabulate                 0.9.0
tensorboard              2.14.0
tensorboard-data-server  0.7.2
thinc                    8.2.2
tokenizers               0.15.0
torch                    1.13.1
torchmetrics             1.2.1
tqdm                     4.66.1
transformers             4.36.1
typer                    0.9.0
types-pytz               2023.3.1.1
typing_extensions        4.9.0
urllib3                  2.1.0
wandb                    0.12.21
wasabi                   1.1.2
weasel                   0.3.4
Werkzeug                 3.0.1
wheel                    0.41.2
yarl                     1.9.4
zipp                     3.17.0

    And the problem is as same as the first situation

Actually, I'm trying to reproduce PeaCoK, follow the guide I recreate an environment use python=3.10 and install the requirements. This time when I use this environment to run examples/train_comet_bart.py, the problem is like this:

Global seed set to 42
Traceback (most recent call last):
  File "/root/autodl-tmp/kogito-main/examples/train_comet_bart.py", line 39, in <module>
    model.train(
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/kogito/models/bart/comet.py", line 108, in train
    trainer: pl.Trainer = generic_train(
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/kogito/models/bart/utils.py", line 249, in generic_train
    trainer = pl.Trainer(
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/utilities/argparse.py", line 348, in insert_env_defaults
    return fn(self, **kwargs)
TypeError: Trainer.__init__() got an unexpected keyword argument 'weights_summary'

Then I replace the pytorch-lightning==1.9.5 with pytorch-lightning==1.6.5, this error which due to numpy==1.24.1 occurred:

Global seed set to 42
/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/connectors/callback_connector.py:151: LightningDeprecationWarning: Setting `Trainer(checkpoint_callback=<pytorch_lightning.callbacks.model_checkpoint.ModelCheckpoint object at 0x7f13358ebee0>)` is deprecated in v1.5 and will be removed in v1.7. Please consider using `Trainer(enable_checkpointing=<pytorch_lightning.callbacks.model_checkpoint.ModelCheckpoint object at 0x7f13358ebee0>)`.
  rank_zero_deprecation(
/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/connectors/callback_connector.py:171: LightningDeprecationWarning: Setting `Trainer(weights_summary=None)` is deprecated in v1.5 and will be removed in v1.7. Please set `Trainer(enable_model_summary=False)` instead.
  rank_zero_deprecation(
GPU available: True, used: False
TPU available: False, using: 0 TPU cores
IPU available: False, using: 0 IPUs
HPU available: False, using: 0 HPUs
/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py:1814: PossibleUserWarning: GPU available but not used. Set `accelerator` and `devices` using `Trainer(accelerator='gpu', devices=1)`.
  rank_zero_warn(
/root/miniconda3/envs/kgto/lib/python3.10/site-packages/transformers/optimization.py:306: FutureWarning: This implementation of AdamW is deprecated and will be removed in a future version. Use the PyTorch implementation torch.optim.AdamW instead, or set `no_deprecation_warning=True` to disable this warning
  warnings.warn(
Sanity Checking: 0it [00:00, ?it/s]/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/connectors/data_connector.py:240: PossibleUserWarning: The dataloader, val_dataloader 0, does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` (try 52 which is the number of cpus on this machine) in the `DataLoader` init to improve performance.
  rank_zero_warn(
Sanity Checking DataLoader 0:   0%|                                                                                                    | 0/2 [00:00<?, ?it/s]/root/miniconda3/envs/kgto/lib/python3.10/site-packages/transformers/generation/utils.py:1387: UserWarning: Neither `max_length` nor `max_new_tokens` has been set, `max_length` will default to 20 (`self.config.max_length`). Controlling `max_length` via the config is deprecated and `max_length` will be removed from the config in v5 of Transformers -- we recommend using `max_new_tokens` to control the maximum length of the generation.
  warnings.warn(
/root/miniconda3/envs/kgto/lib/python3.10/site-packages/kogito/models/bart/utils.py:487: UserWarning: All learning rates are 0                               
  warnings.warn("All learning rates are 0")
/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/connectors/data_connector.py:240: PossibleUserWarning: The dataloader, train_dataloader, does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` (try 52 which is the number of cpus on this machine) in the `DataLoader` init to improve performance.
  rank_zero_warn(
Epoch 0:   0%|                                                                                                                      | 0/3126 [00:00<?, ?it/s]Traceback (most recent call last):
  File "/root/autodl-tmp/kogito-main/examples/train_comet_bart.py", line 39, in <module>
    model.train(
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/kogito/models/bart/comet.py", line 108, in train
    trainer: pl.Trainer = generic_train(
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/kogito/models/bart/utils.py", line 262, in generic_train
    trainer.fit(model)
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 770, in fit
    self._call_and_handle_interrupt(
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 723, in _call_and_handle_interrupt
    return trainer_fn(*args, **kwargs)
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 811, in _fit_impl
    results = self._run(model, ckpt_path=self.ckpt_path)
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 1236, in _run
    results = self._run_stage()
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 1323, in _run_stage
    return self._run_train()
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py", line 1353, in _run_train
    self.fit_loop.run()
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/loops/base.py", line 204, in run
    self.advance(*args, **kwargs)
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/loops/fit_loop.py", line 266, in advance
    self._outputs = self.epoch_loop.run(self._data_fetcher)
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/loops/base.py", line 199, in run
    self.on_run_start(*args, **kwargs)
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/loops/epoch/training_epoch_loop.py", line 153, in on_run_start
    _ = iter(data_fetcher)  # creates the iterator inside the fetcher
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/utilities/fetching.py", line 179, in __iter__
    self._apply_patch()
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/utilities/fetching.py", line 115, in _apply_patch
    apply_to_collections(self.loaders, self.loader_iters, (Iterator, DataLoader), _apply_patch_fn)
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/utilities/fetching.py", line 154, in loader_iters
    loader_iters = self.dataloader_iter.loader_iters
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/supporters.py", line 545, in loader_iters
    self._loader_iters = self.create_loader_iters(self.loaders)
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/supporters.py", line 585, in create_loader_iters
    return apply_to_collection(loaders, Iterable, iter, wrong_dtype=(Sequence, Mapping))
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/utilities/apply_func.py", line 99, in apply_to_collection
    return function(data, *args, **kwargs)
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 435, in __iter__
    return self._get_iterator()
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 381, in _get_iterator
    return _MultiProcessingDataLoaderIter(self)
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 1072, in __init__
    self._reset(loader, first_iter=True)
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 1105, in _reset
    self._try_put_index()
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 1339, in _try_put_index
    index = self._next_index()
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 618, in _next_index
    return next(self._sampler_iter)  # may raise StopIteration
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/torch/utils/data/sampler.py", line 254, in __iter__
    for idx in self.sampler:
  File "/root/miniconda3/envs/kgto/lib/python3.10/site-packages/kogito/core/utils.py", line 125, in __iter__
    np.concatenate(np.random.permutation(ck_idx[1:]))
  File "mtrand.pyx", line 4703, in numpy.random.mtrand.RandomState.permutation
ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 1 dimensions. The detected shape was (3121,) + inhomogeneous part.
Epoch 0:   0%|          | 0/3126 [00:01<?, ?it/s]

Replace numpy==1.24.1 it with older version like numpy==1.21.2 can solve this error, but with another warnings like this:

Global seed set to 42
/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/connectors/callback_connector.py:151: LightningDeprecationWarning: Setting `Trainer(checkpoint_callback=<pytorch_lightning.callbacks.model_checkpoint.ModelCheckpoint object at 0x7f5b436913c0>)` is deprecated in v1.5 and will be removed in v1.7. Please consider using `Trainer(enable_checkpointing=<pytorch_lightning.callbacks.model_checkpoint.ModelCheckpoint object at 0x7f5b436913c0>)`.
  rank_zero_deprecation(
/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/connectors/callback_connector.py:171: LightningDeprecationWarning: Setting `Trainer(weights_summary=None)` is deprecated in v1.5 and will be removed in v1.7. Please set `Trainer(enable_model_summary=False)` instead.
  rank_zero_deprecation(
GPU available: True, used: False
TPU available: False, using: 0 TPU cores
IPU available: False, using: 0 IPUs
HPU available: False, using: 0 HPUs
/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/trainer.py:1814: PossibleUserWarning: GPU available but not used. Set `accelerator` and `devices` using `Trainer(accelerator='gpu', devices=1)`.
  rank_zero_warn(
/root/miniconda3/envs/kgto/lib/python3.10/site-packages/transformers/optimization.py:306: FutureWarning: This implementation of AdamW is deprecated and will be removed in a future version. Use the PyTorch implementation torch.optim.AdamW instead, or set `no_deprecation_warning=True` to disable this warning
  warnings.warn(
Sanity Checking: 0it [00:00, ?it/s]/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/connectors/data_connector.py:240: PossibleUserWarning: The dataloader, val_dataloader 0, does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` (try 52 which is the number of cpus on this machine) in the `DataLoader` init to improve performance.
  rank_zero_warn(
Sanity Checking DataLoader 0:   0%|                                                                                                    | 0/2 [00:00<?, ?it/s]/root/miniconda3/envs/kgto/lib/python3.10/site-packages/transformers/generation/utils.py:1387: UserWarning: Neither `max_length` nor `max_new_tokens` has been set, `max_length` will default to 20 (`self.config.max_length`). Controlling `max_length` via the config is deprecated and `max_length` will be removed from the config in v5 of Transformers -- we recommend using `max_new_tokens` to control the maximum length of the generation.
  warnings.warn(
/root/miniconda3/envs/kgto/lib/python3.10/site-packages/kogito/models/bart/utils.py:487: UserWarning: All learning rates are 0                               
  warnings.warn("All learning rates are 0")
/root/miniconda3/envs/kgto/lib/python3.10/site-packages/pytorch_lightning/trainer/connectors/data_connector.py:240: PossibleUserWarning: The dataloader, train_dataloader, does not have many workers which may be a bottleneck. Consider increasing the value of the `num_workers` argument` (try 52 which is the number of cpus on this machine) in the `DataLoader` init to improve performance.
  rank_zero_warn(
Epoch 0:   0%|                                                                                                                      | 0/3126 [00:00<?, ?it/s]/root/miniconda3/envs/kgto/lib/python3.10/site-packages/kogito/core/utils.py:125: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.
  np.concatenate(np.random.permutation(ck_idx[1:]))
<__array_function__ internals>:5: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.

It seems like the problem is about the conflict between pytorch-lightning used numpy version and the repo used numpy version.

The cuda versoon of torch which is automatically is 11.7