Working with LLMs in existing pipelines can often be bloated, complex, and slow. That's why I created FlashLearn, a streamlined library that mirrors the user experience of scikit-learn. It follows a pipeline-like structure allowing you to "fit" (learn) skills from sample data or instructions, and "predict" (apply) these skills to new data, returning structured results.

High-Level Concept Flow:

Your Data --> Load Skill / Learn Skill --> Create Tasks --> Run Tasks --> Structured Results --> Downstream Steps

Installation:

    pip install flashlearn

Learning a New "Skill" from Sample Data

Just like a fit/predict pattern in scikit-learn, you can quickly "learn" a custom skill from minimal (or no!) data. Here's an example where we create a skill to evaluate the likelihood of purchasing a product based on user comments:

  from flashlearn.skills.learn_skill import LearnSkill
  from flashlearn.client import OpenAI

  # Instantiate your pipeline "estimator" or "transformer", similar to a scikit-learn model
  learner = LearnSkill(model_name="gpt-4o-mini", client=OpenAI())
  data = [
    {"comment_text": "I love this product, it's everything I wanted!"},
    {"comment_text": "Not impressed... wouldn't consider buying this."},
    # ...
  ]

  # Provide instructions and sample data for the new skill
  skill = learner.learn_skill(
    data,
    task=(
      "Evaluate how likely the user is to buy my product based on the sentiment in their comment, "
      "return an integer 1-100 on key 'likely_to_buy', "
      "and a short explanation on key 'reason'."
    ),
  )

    # Save skill to use in pipelines

skill.save("evaluate_buy_comments_skill.json")

Input Is a List of Dictionaries

Simply wrap each record into a dictionary, much like feature dictionaries in typical ML workflows:

  user_inputs = [
    {"comment_text": "I love this product, it's everything I wanted!"},
    {"comment_text": "Not impressed... wouldn't consider buying this."},
    # ...
  ]

Run in 3 Lines of Code - Concurrency Built-in up to 1000 calls/min

  # Suppose we previously saved a learned skill to "evaluate_buy_comments_skill.json".
  skill = GeneralSkill.load_skill("evaluate_buy_comments_skill.json")
  tasks = skill.create_tasks(user_inputs)
  results = skill.run_tasks_in_parallel(tasks)
  print(results)

Get Structured Results

Here's an example of structured outputs mapped to indexes of your original list:

  {
  "0": {
    "likely_to_buy": 90,
    "reason": "Comment shows strong enthusiasm and positive sentiment."
  },
  "1": {
    "likely_to_buy": 25,
    "reason": "Expressed disappointment and reluctance to purchase."
  }
  }

Pass on to the Next Steps

You can use each record’s output for downstream tasks such as storing results in a database or filtering high-likelihood leads:

  # Suppose 'flash_results' is the dictionary with structured LLM outputs
  for idx, result in flash_results.items():
      desired_score = result["likely_to_buy"]
      reason_text = result["reason"]
      # Now do something with the score and reason, e.g., store in DB or pass to next step
      print(f"Comment #{idx} => Score: {desired_score}, Reason: {reason_text}")