diff --git a/06_gpu_and_ml/sentence_embeddings.py b/06_gpu_and_ml/sentence_embeddings.py
deleted file mode 100644
index c6e93c36b..000000000
--- a/06_gpu_and_ml/sentence_embeddings.py
+++ /dev/null
@@ -1,159 +0,0 @@
-# ---
-# lambda-test: false
-# ---
-"""
-Computes sentence embeddings using Modal.
-This uses huggingface's `transformer` library to calculate the vector
-representations for a collection of sentences.
-
-Example modified from: https://huggingface.co/sentence-transformers/paraphrase-xlm-r-multilingual-v1
-
-Install dependencies before running example:
-
-$ pip3 install torch==1.10.2 tqdm numpy requests tensorboard
-"""
-import tarfile
-from pathlib import Path
-from typing import List
-
-import modal
-
-# dependencies
-dependencies = ["torch==1.10.2", "transformers==4.16.2", "tensorboard"]
-stub = modal.Stub(
-    "example-sentence-embeddings",
-    image=modal.Image.debian_slim().pip_install(*dependencies),
-)
-
-if stub.is_inside():
-    import numpy as np
-    import requests
-    import torch
-    from torch.utils.tensorboard import SummaryWriter
-    from transformers import AutoModel, AutoTokenizer
-
-    TOKENIZER = AutoTokenizer.from_pretrained(
-        "sentence-transformers/paraphrase-xlm-r-multilingual-v1"
-    )
-    MODEL = AutoModel.from_pretrained(
-        "sentence-transformers/paraphrase-xlm-r-multilingual-v1"
-    )
-
-
-def mean_pooling(model_output, attention_mask):
-    token_embeddings = model_output[
-        0
-    ]  # First element of model_output contains all token embeddings
-    input_mask_expanded = (
-        attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
-    )
-    return torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(
-        input_mask_expanded.sum(1), min=1e-9
-    )
-
-
-@stub.function()
-def vectorize(x: str):
-    """Vectorizes a string (sentence) into a PyTorch Tensor."""
-    # encode input and calculate vector
-    encoded_input = TOKENIZER(
-        x, padding=True, truncation=True, return_tensors="pt"
-    )
-    model_output = MODEL(**encoded_input)
-    y = mean_pooling(model_output, encoded_input["attention_mask"])
-
-    sentence = f"{x[:100]} ..."
-    print(f" → calculated vector for: {sentence}")
-    return (sentence, y.detach().numpy())
-
-
-def write_tensorboard_logs(embedding, metadata: List[str]):
-    """Write tensorboard logs."""
-    print(" → writing tensorboard logs")
-    writer = SummaryWriter(log_dir="./tensorboard")
-    writer.add_embedding(np.concatenate(embedding), metadata)
-    writer.close()
-
-
-class MovieReviewsDataset:
-    """Standford's Large Movie Review Dataset."""
-
-    url: str = "http://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz"
-
-    def __init__(
-        self, output_path: Path = Path("./movie_reviews_dataset.tar.gz")
-    ):
-        self.output_path = output_path
-        self.dataset: List[str] = []
-
-    def download(self) -> None:
-        """Downloads dataset if not available locally."""
-        if self.output_path.exists():
-            print(" → using cached dataset")
-            return
-
-        print(f" → downloading dataset from {self.url}")
-        with requests.get(self.url, stream=True) as r:
-            r.raise_for_status()
-            with open(self.output_path, "wb") as f:
-                for chunk in r.iter_content(chunk_size=8192):
-                    # If you have chunk encoded response uncomment if
-                    # and set chunk_size parameter to None.
-                    # if chunk:
-                    f.write(chunk)
-
-        print(" → dataset downloaded")
-
-    def extract(self) -> None:
-        """Extracts dataset package into local path."""
-        if Path("./aclImdb").exists():
-            print(" → using cached extracted dataset")
-            return
-
-        print(f" → extracting file {self.output_path}")
-        with tarfile.open(self.output_path) as tf:
-            tf.extractall()
-
-    def index(self) -> list[Path]:
-        """Creates in-memory index for dataset"""
-        # read only train files
-        dataset_path = Path("./aclImdb/train")
-
-        # combine both negative and positive files
-        negative = [f for f in (dataset_path / "neg").glob("*.txt")]
-        positive = [f for f in (dataset_path / "pos").glob("*.txt")]
-        return negative + positive
-
-    def load(self, sample_size: int = 100) -> None:
-        """Loads dataset into memory"""
-        print(f" → loading dataset with sample size = {sample_size}")
-        for file in np.random.choice(self.index(), size=sample_size):
-            self.dataset.append(file.open().read())  # limit to 500 characters
-
-
-@stub.local_entrypoint()
-def main():
-    # download & prepare dataset
-    dataset = MovieReviewsDataset()
-    dataset.download()
-    dataset.extract()
-
-    # load dataset into memory; use a small samplesize (N=25k)
-    dataset.load(sample_size=100)
-
-    # vectorize the entire dataset
-    embedding = []
-    metadata = []
-    with stub.run():
-        for sentence, vector in vectorize.map(dataset.dataset):
-            embedding.append(vector)
-            metadata.append(sentence)
-
-        write_tensorboard_logs(embedding, metadata)
-
-    # open tensorboard
-    print(" → done!")
-    print(
-        " → to see results in TensorBoard, run: tensorboard --logdir tensorboard/"
-    )
-    print(" → (open http://localhost:6006#projector and wait for it to load)")