@<1523701118159294464:profile|ExasperatedCrab78> Sorry only saw this now,
Thanks for checking it!
Glad to see you found the issue, hope you find a way to fix the second one. for now we will continue using the previous version.
Would be glad if you can post when everything is fixed so we can advance our version.

@<1523701118159294464:profile|ExasperatedCrab78>
Ok. bummer to hear that it won't be included automatically in the package.

I am now experiencing a bug with the patch, not sure it's to blame... but i'm unable to save models in the pipeline.. checking if it's related

I appreciate it!

@<1523701118159294464:profile|ExasperatedCrab78>
Here is an example that reproduces the second error

from clearml.automation import PipelineDecorator
from clearml import TaskTypes

@PipelineDecorator.component(task_type=TaskTypes.data_processing, cache=True)
def run_demo():
    from transformers import AutoTokenizer, DataCollatorForTokenClassification, AutoModelForSequenceClassification, TrainingArguments, Trainer
    from datasets import load_dataset
    import numpy as np
    import evaluate
    from pathlib import Path

    dataset = load_dataset("yelp_review_full")

    tokenizer = AutoTokenizer.from_pretrained("bert-base-cased")


    def tokenize_function(examples):
        return tokenizer(examples["text"], padding="max_length", truncation=True)
    
    
    def compute_metrics(eval_pred):
        logits, labels = eval_pred
        predictions = np.argmax(logits, axis=-1)
        return metric.compute(predictions=predictions, references=labels)

    
    small_train_dataset = dataset["train"].shuffle(seed=42).select(range(10))
    small_eval_dataset = dataset["test"].shuffle(seed=42).select(range(10))
    
    small_train_dataset = small_train_dataset.map(tokenize_function, batched=True)
    small_eval_dataset = small_eval_dataset.map(tokenize_function, batched=True)

    model = AutoModelForSequenceClassification.from_pretrained("bert-base-cased", num_labels=5)

    training_args = TrainingArguments(
        output_dir="test_trainer", 
        evaluation_strategy="epoch",
        # num_train_epoch=1,
    )
    
    metric = evaluate.load("accuracy")
    
    trainer = Trainer(
        model=model,
        args=training_args,
        train_dataset=small_train_dataset,
        eval_dataset=small_eval_dataset,
        compute_metrics=compute_metrics,
    )
    
    trainer.train()
    
    return Path('test_trainer')

@PipelineDecorator.component(task_type=TaskTypes.data_processing, cache=True)
def second_step(some_param):
    print("Success!")
    
@PipelineDecorator.pipeline(name="StuffToDelete", project=".Dev", version="0.0.2", pipeline_execution_queue="aws_cpu")
def pipeline():
    data = run_demo()
    second_step(data)

if __name__ == '__main__':
    PipelineDecorator.set_default_execution_queue("aws_cpu")
    
    PipelineDecorator.run_locally()
    
    pipeline()

SmugDolphin23 BTW, this is using clearml and huggingface’s automatic logging… didn’t log something manual

The version is v1.9.1

Nothing that i think is relevant, I'm using latest from master. It might be a new bug on their side, wasn't sure.

@<1523701435869433856:profile|SmugDolphin23>
Hey 🙂
Any update?

We are having more issues with transformers and clearml in their new version.
The step that has transformers 4.25.1 isn’t able to upload artifacts.
If we downgrade transformers==4.21.3 it works

for now we downgraded to 1.7.2, but of course prefer not to stay that way

Hey 🙂 Thanks for the update!

what i’m missing the is the point where you report to clearml between cast and casting back 🤔

Saw it was merged 🙂 One down, one to go

When creating it, I found that this hack should be on our side, not on Huggingface's. So I'm only going to fix issue 1 with the PR, issue 2 is ours 🙂

Good to hear 🙏

However, I actually do think I can already open the Huggingface PR in the meantime. It has actually relatively little to do with the second bug.

Allright, a bit of searching later and I've found 2 things:

• You were right about the task! I've staged a fix here . It basically detects whether a task is already running (e.g. from the pipelinedecorator component) and if so, uses that task instead. We should probably do this for all of our integrations.
• But then I found another bug. Basically the pipeline decorator task would mess up the internal nested dict of the label mapping inside of the model config. You will probably have the same issue if you run the pipeline with my fix above.
  So for now, we're looking into the 2nd bug, because it breaks with Hugging Face models in a pipeline. Until we sort that out, I'm going to hold off on opening a PR to HF with the first fix. Makes sense?

Thanks a lot for the example, it helped tons to be able to reproduce!

Hey @<1523701949617147904:profile|PricklyRaven28> , So as discussed above there were 2 issues. The first one is still waiting on the second, it's on the backlog of our devs and should be done soon(tm).

That said, in the meantime I also wanted to do fun stuff with transformers, so I've written a quick hack that deals with the bug. It's bascially 2 functions that keep track of which types of keys are in the dict.

def cast_keys_to_string(d, changed_keys=dict()):
    nd = dict()
    for key in d.keys():
        if not isinstance(key, str):
            casted_key = str(key)
            changed_keys[casted_key] = key
        else:
            casted_key = key
        if isinstance(d[key], dict):
            nd[casted_key], changed_keys = cast_keys_to_string(d[key], changed_keys)
        else:
            nd[casted_key] = d[key]
    return nd, changed_keys

def cast_keys_back(d, changed_keys):
    nd = dict()
    for key in d.keys():
        if key in changed_keys:
            original_key = changed_keys[key]
        else:
            original_key = key
        if isinstance(d[key], dict):
            nd[original_key], changed_keys = cast_keys_back(d[key], changed_keys)
        else:
            nd[original_key] = d[key]
    return nd, changed_keys

You can then use them like this:

        training_args = TrainingArguments(
            output_dir="my_awesome_model",
            learning_rate=2e-5,
            per_device_train_batch_size=16,
            per_device_eval_batch_size=16,
            dataloader_num_workers=0,
            num_train_epochs=2,
            weight_decay=0.01,
            evaluation_strategy="epoch",
            save_strategy="epoch",
            load_best_model_at_end=True
        )

        # Allow ClearML access to the training args and allow it to override the arguments for remote execution
        args_class = type(training_args)
        args, changed_keys = cast_keys_to_string(training_args.to_dict())
        training_args = args_class(**cast_keys_back(args, changed_keys)[0])

        self.trainer = Trainer(
            model=self.model,
            args=training_args,
            train_dataset=tokenized_dataset["train"],
            eval_dataset=tokenized_dataset["test"],
            tokenizer=self.tokenizer,
            data_collator=data_collator,
            compute_metrics=self.compute_metrics,
        )

        self.trainer.train()

This "hack" in combination with the patch to Huggingface from above should work 🙂 That said, it is a hack, so a production version of this should be there soon. I'll let you know when that happens!

No worries! And thanks for putting in the time.

i believe this is because of this code
None

Which initialized the task if clearml is installed… but a task already exists (because of the pipeline), it will replace it

` args.py #504:
for k, v in dictionary.items():
# if key is not present in the task's parameters, assume we didn't get this far when running
# in non-remote mode, and just add it to the task's parameters
if k not in parameters:
self._task.set_parameter((prefix or '') + k, v)
continue

task.py #1266:
def set_parameter(self, name, value, description=None, value_type=None):
# type: (str, str, Optional[str], Optional[Any]) -> ()
"""
Set a single Task parameter. This overrides any previous value for this parameter.

    :param name: The parameter name.
    :param value: The parameter value.
    :param description: The parameter description.
    :param value_type: The type of the parameters (cast to string and store)
    """
    if not Session.check_min_api_version('2.9'):
        # not supported yet
        description = None
        value_type = None

    self._set_parameters(
        {name: value}, __update=True,
        __parameters_descriptions={name: description},
        __parameters_types={name: value_type}
    )

task.py #1227:
def create_description():
if org_param and org_param.description:
return org_param.description
created_description = ""
if org_k in descriptions:
created_description = descriptions[org_k]
if isinstance(v, Enum):
# append enum values to description
if created_description:
created_description += "\n"
created_description += "Values:\n" + ",\n".join(
[enum_key for enum_key in type(v).dict.keys() if not enum_key.startswith("_")]
)
return created_description `We can see from this code that the description will always be None (because copy_to_dict never passes a description, it defaults to None and is always put in the descriptions dict as None), and if the arg is an Enum it will always throw the exception

Hey @<1523701949617147904:profile|PricklyRaven28> , about the S3 loading issue. The path to the model in the artifact tab, is it an S3 bucket or a local path?

Just for reference, the main issue is that ClearML does not allow non-string types as dict keys for its configuration. Usually the labeling mapping does have ints as keys. Which is why we need to cast them to strings first, then pass them to ClearML then cast them back.

will check

BTW the code above is from clearml github so it’s the latest

Hi @<1523701949617147904:profile|PricklyRaven28> ! We released ClearmlSDK 1.9.1 yesterday. Can you please try it?

Looks like the first issue has been solved 🙂

i think the second one still consists, still checking

@<1523701118159294464:profile|ExasperatedCrab78>
Hey 🙂
Any updates on this? We need to use a new version of transformers because of another bug they have in an old version. so we can’t use the old transformers version anymore.

Hi PricklyRaven28 , can you try with 1.9.1rc0?

This is the next step not being able to find the output of the last step

ValueError: Could not retrieve a local copy of artifact return_object, failed downloading 

sounds good 🙂 I’ll soon check if this fixes our issue and update you

Hi PricklyRaven28 ! What dict do you connect? Do you have a small script we could use to reproduce?