Replace notebooks_general.ipynb

%% Cell type:markdown id: tags:

# General Remarks on the Notebooks

**General principle** we try to follow:

1)  We implement small components in more manual fashion (e.g. a tokenizer) to improve *understanding*,

2)  then learn how to use a library (torch, transfomers) to *implement* the same task (more efficiently).

**Structure**

  - Setup & data, including info whether GPU is required.

  - Content of the chapter

There are text (markdown) cells and code cells where you can run python code.

**Code remarks:**

- If a package is not installed, run: pip install packagename

- If a package is not installed, run: *pip install {packagename}*

- If a dataset cannot be loaded: check that the file is available and the path is correct.

- Make sure that imported function files (e.g. utils.py) are located in the same directory. These contain functionalities we want to use, but not focus on.

- Make sure that imported function files (e.g. *utils.py*) are located in the same directory. These contain functionalities we want to use, but not focus on.

**Section remarks**

- Show (and extend) the topic of the lecture part.

- The descriptions are sometimes very short. Check the slides and ask the lecturer for more details.

- Feel free to also change / experiment with codes here!

- A **"*"** denotes optional components (tasks and sections)

- A **"*"** denotes **optional components** (tasks and sections)

**Tasks Sections**

  - Tasks are for you to write code yourself.

  - Other codes from the notebooks and all online help can and should be used to solve them.

  - Simply add your own code cell and start solving. There is no "one correct" solution.

  - There are example solutions provided for most non-optional tasks. Check them only if you are really stuck!

**Note on library usage**

We use libraries because:

- working with libraries is faster and removes many details to focus on the key elements,

- in the course time, we will not be able to combine our self-written functions together to an entire well performing LMs.

%% Cell type:markdown id: tags:

# Examples

%% Cell type:markdown id: tags:

This is an example Markdown Cell

We can write **text** and equations $a^2+b^2 = c^2$ (and many more things) here.

%% Cell type:code id: tags:

``` python

# Example Code Cell

import numpy as np  # imports

a = np.array([1, 2, 3, 4, 5])

for i in a:

    print(i)

```

%% Output

    1

    2

    3

    4

    5