Update document formatting for v2 tool use in code examples

[mrmer1]

This PR introduces a new tutorial for Cohere's API, which is split into seven parts. Each part focuses on a different use case, including installation and setup, text generation, chatbots, semantic search, reranking, retrieval-augmented generation (RAG), and agents with tool use. The tutorial is designed to be completed in around 15 minutes.

[ai-yann]

@mrmer1 please check the merge conflicts.

[ai-yann]

@mrmer1 This looks great! Could you please also make the following updates:

1. Add pip install commands as the first code cell in each of the three notebooks
2. Include specific version numbers for all packages (e.g., pip install pandas==2.1.0) to ensure reproducibility

This will help users run the notebooks successfully in the future without version compatibility issues.

[mrmer1]

@ai-yann updated the install versions

[ai-yann]

Thanks for the versioning edits. I was able to run through all the notebooks without issue -- I just added a few more comments, putting myself into the shoes of a customer during a workshop. I hope it's not too much

[ai-yann]

In addition to You can find the [dataset here](https://github.com/cohere-ai/notebooks/blob/main/notebooks/guides/advanced_rag/spotify_dataset.csv). in the markdown, could you please also add to the python cell that follows it:

import pandas as pd
import shutil
from pathlib import Path

def setup_spotify_dataset():
    """
    Loads the Spotify dataset and ensures a local copy exists in the current directory.
    Returns the loaded DataFrame.
    """
    # First try to load from current directory
    local_path = Path('spotify_dataset.csv')
    
    if local_path.exists():
        print("Loading Spotify dataset from local directory...")
        return pd.read_csv(local_path)
    
    # If not found locally, try to find in notebooks directory structure
    try:
        current = Path.cwd()
        while current.name != 'notebooks' and current.parent != current:
            current = current.parent
        if current.name != 'notebooks':
            raise RuntimeError("Could not find notebooks directory")
        
        # Original file path
        original_path = current / 'guides' / 'advanced_rag' / 'spotify_dataset.csv'
        
        if not original_path.exists():
            raise FileNotFoundError(f"Dataset not found at {original_path}")
        
        # Copy file to current directory
        print(f"Copying Spotify dataset to local directory ({local_path})...")
        shutil.copy2(original_path, local_path)
        
        # Load and return the data
        return pd.read_csv(local_path)
        
    except (RuntimeError, FileNotFoundError) as e:
        print(f"Error: {e}")
        print("Please ensure the Spotify dataset is available either locally or in the expected directory structure.")
        raise

# Load the dataset
try:
    spotify_data = setup_spotify_dataset()
    print("\nFirst few rows of the dataset:")
    display(spotify_data.head(3))
except Exception as e:
    print(f"Failed to load dataset: {e}")

This just makes it a smoother experience for folks trying to follow along in a group training setting. This way folks don't need to leave the window, download, upload, etc.

[ai-yann]

Are the final results consistent? Can we add one more cell that outputs the final answers in markdown with clickable citations, like we do in the insert_inline_citations method of the notebooks/agents/Vanilla_Tool_Use_v2.ipynb notebook? Because now it's a lot to scroll through, and a lot looks like

Start: 531 | End: 568 | Text: 'Damián Pacheco (twelve-string guitar)'
Sources:
web_search_ra443ajyz6xj:0
web_search_ra443ajyz6xj:2
web_search_ta7g2cd67jrx:0`

Something that looks like:

Spotify 2023 Top Songs Analysis

Top 3 Most Streamed Songs

1. "Flowers" by Miley Cyrus
2. "Ella Baila Sola" by Eslabon Armado and Peso Pluma
3. "Shakira: Bzrp Music Sessions, Vol. 53" by Shakira and Bizarrap

Artist Details

Miley Cyrus

• Age: 31 years old
• Citizenship: United States
• Sources:
  • Miley Cyrus: Facts & Related Content - Britannica
  • Miley Cyrus - Wikidata

Eslabon Armado

• Type: American regional Mexican group
• Location: Patterson, California
• Formation: 2017
• Members:
  • Pedro Tovar (vocals)
  • Brian Tovar (bass)
  • Ulises González (acoustic guitar)
  • Damián Pacheco (twelve-string guitar)
• Sources:
  • Eslabon Armado - Wikipedia
  • Latin Artist on the Rise: Eslabon Armado - Billboard

Peso Pluma

• Age: 25 years old
• Birthdate: June 15, 1999
• Citizenship: Mexican
• Sources:
  • Who is Peso Pluma? - Wegotthiscovered
  • Peso Pluma Nationality - Guiderweb

Shakira

• Age: 47 years old (born February 2, 1977)
• Citizenship: Colombian
• Location of Birth: Barranquilla, Colombia
• Sources:
  • Shakira - Wikipedia
  • Shakira | Biography, Songs, & Facts - Britannica

Bizarrap

• Age: 25 years old
• Birthdate: August 29, 1998
• Citizenship: Argentine
• Profession: Record producer, songwriter, and DJ
• Sources:
  • Bizarrap - Wikipedia
  • Bizarrap - Wikidata

Methodology

• Data sourced from Spotify's 2023 streaming statistics
• Artist information verified through web searches
• Ages and citizenships confirmed through multiple sources

[ai-yann]

And you can use:

from IPython.display import Markdown
display(Markdown(markdown_response))

[ai-yann]

Since we go to the trouble of creating a markdown table with the results and inline citations, maybe we can display the table as markdown, by adding:

from IPython.display import Markdown
display(Markdown(cited_text))
print("\n" + list_sources(response.message.citations, source_index))

[ai-yann]

can we remove the two newlines at the top of this cell? just for presentation purposes :)