How to set up a training job in sagemaker ?

Question

I'm following a blog/sample here - https://kedion.medium.com/fine-tuning-nlp-models-with-hugging-face-f92d55949b66. How can i set up something similar with just sagemaker and using aws cli? (sample code below from the example ) . in the example, it uses distilbert-base-uncased model and it is loaded via this code ->  tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased")

where does the model gets downloaded from and if one were to set up similar training job via boto3/cli, can we pass model location somewhere in a s3 bucket?

```
# Importing necessary tools
from transformers import AutoTokenizer, TFAutoModelForSequenceClassification
from datasets import load_dataset
import tensorflow as tf
import numpy as np

# Loading our dataset
tweet_dataset = load_dataset(path="tweet_eval", name="emotion")

# Instantiating our DistilBERT tokenizer and model
tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased")
model = TFAutoModelForSequenceClassification.from_pretrained("distilbert-base-uncased", num_labels=4

```

Answer

Hi clouduser,

If you are looking to set up something similar with just SageMaker and using the AWS CLI, [here](https://huggingface.co/docs/sagemaker/train) is an article that shows how you can directly set up a training job using a Hugging Face model and Amazon SageMaker.

[Here](https://docs.aws.amazon.com/sagemaker/latest/dg/training-compiler-enable-pytorch.html) is another example setup with PyTorch Training Jobs.

[Here](https://docs.aws.amazon.com/sagemaker/latest/dg/training-compiler-enable-tensorflow.html) is another example setup with TensorFlow Training jobs.

I would recommend following one of these three blogs to set up your Amazon SageMaker training job based on which model you decide to go with.

How to set up a training job in sagemaker ?

関連するコンテンツ