valiantone

Wavess Guide to Fine-tuning LLMs

Wavess is a play to explore funding opportunities with Astrik in the marketing co-pilot AI B2B service space.

Ahead AI specializes in Machine Learning & AI research and is read by tens of thousands of researchers and practitioners who want to stay ahead in the ever-evolving field.

General Fine-tuning

Resources

Fine-tuning large language models (LLMs) in 2024 Fine-tuning open source large language models (LLMs)

valiantone

def s3_to_pandas(client, bucket, key, header=None):

    # get key using boto3 client
    obj = client.get_object(Bucket=bucket, Key=key)
    gz = gzip.GzipFile(fileobj=obj['Body'])
    
    # load stream directly to DF
    return pd.read_csv(gz, header=header, dtype=str)

def s3_to_pandas_with_processing(client, bucket, key, header=None):

valiantone

Weekly, Monthly Active Emails

WITH
    -- this is your original query, with the ISO week and month number added.
    members_log_aggr(login_date,  year_nbr, iso_week_nbr, month_nbr, email_count) AS
    (
        SELECT
            CAST(ml.login AS Date),
            DATEPART(YEAR, ml.login),
            DATEPART(ISO_WEEK, ml.login),

valiantone

Transitioning from Tag Clouds to Tag Trees

In the packages paradigm - each packaged selection/offering is a bundle of semi-unique characteristics - lets term these as attributes. When a package bundle is selected it instantly informs our controller that certain atrributes are ground truth for this move. The current process introduces assumptions rather than ground truth. For instance let's observe Use Case 1.

Use Case 1: Studio Package Bundle

Facts:
    - Two Bellhops on the move
    - Duration of ~ 2hours
    - 16 Foot Moving truck required

valiantone

Installing Cool-Retro-Term on Windows10

First of all, this document is just a recompilation of different resources that already existed on the web previously that I personally tested some ones did work and other not. I liked the idea to make a full guide from start to end so all of you could also enjoy playing with cool-retro-term on windows 10. Personally I installed it on a windows 10 pro version. Fingers crossed!

valiantone

train = pd.DataFrame([
    {"Name": "Olyphant", "FamilySize": 1},
    {"Name": "Rodent", "FamilySize": 3},
    {"Name": "Possum", "FamilySize": 1},
])

sub = train[train["FamilySize"] == 1]
sub["isAlone"] = 1
train

valiantone

Statistical Digressions

• T-test, ANOVA or Regression, what's the difference?

Hypothesis-driven testing

• A/B testing by Julian Shapiro
• A/B Testing with Machine Learning - A Step-by-Step Tutorial
• Beyond A/B Testing: Multi-armed Bandit Experiments

Handling Network Effects

• How to Optimize your Switchback A/B Test Configuration
• Design and Analysis of Switchback Experiments

valiantone

Developer Tools

osx setups

• Mac Setup for Web Development [2022]
• My dev setup 2022
• macOS Setup after 15 Years of Linux
• My 2021 New Mac Setup
• Don't fall in love with your Mac—automate it!
• macOS Setup Guide

valiantone

# @Author: xiewenqian <int>
# @Date:   2016-11-28T20:35:09+08:00
# @Email:  wixb50@gmail.com
# @Last modified by:   int
# @Last modified time: 2016-12-01T19:32:48+08:00


import pandas as pd
from pymongo import MongoClient

valiantone

import pandas as pd
from pymongo import MongoClient
import json

def mongoimport(csv_path, db_name, coll_name, db_url='localhost', db_port=27000)
    """ Imports a csv file at path csv_name to a mongo colection
    returns: count of the documants in the new collection
    """
    client = MongoClient(db_url, db_port)
    db = client[db_name]