Llama Data Science 🦙🦙🦙
🦙🦙🦙 Llama Data Science🦙🦙🦙 I created a simple Python command line tool that leverages Ollama Client and llm-sandbox to test how well the latest tool wielding open-weights models can do data science! Full code and usage at GregariousEngineering/llama-datascience In short, ChatGPT’s brand new OPEN-WEIGHTS model gpt-oss does data science better and faster than many people :) Runner up, Mistral Small 3.2 was also quite competent! gregarious@hal9000 : ~/code/llama-datascience $ ./llama_datascience.py --data-file traffic_volume_000256_2004_2024.csv "What is the best time to travel South through the intersection?" Expert: gpt-oss (temp: 0.5) --- Expert Reasoning Step 1/20 --- > Expert wants to use tool: 'python_datascience' > Reason for tool selection: Load and inspect data > Code: import pandas as pd import numpy as np # Load data try: df = pd.read_csv('/sandbox/data.csv') except Exception as e: print('Error loadin...