The Thief’s Almanac
vibe coding and dev journal
TheModestThief
@thief
Aim: launch a MVP bubblewrap popper mini-app on Farcaster. Goal: learn about how the different lego pieces (e.g. vercel, db, fc sdk, frontend) work with each other rather than web design specifically Mini-app functions in MVP: populate grid of unpopped.png, touch bubble to pop, play sound, show popped.png & increase count Progress: 1. Learnt and pushed with Git 2. Learnt about Vercel deployment (and pricing plans) 3. Vibe coded a bubblewrap popper that runs locally Next steps: host webpage + assets on Vercel, and integrate FC SDK (?), launch After MVP (in descending priorities): 1. Save user popped count across sessions, using FC authkit + some db that works with Vercel (i think the db needs api integration) 2. Force touch triggers (dynamic touch surface area to trigger pop) 3. Beautify app
0 reply
1 recast
2 reactions
TheModestThief
@thief
TIL there is a missingno library for data science!
0 reply
0 recast
2 reactions
TheModestThief
@thief
listening to linkin' park to get me through writing code even when i don't to, just hope that in the end it actually matters
0 reply
0 recast
5 reactions
TheModestThief
@thief
finally finished tuning the model after debugging for the whole fucking day must be doing something wrong because debugging the whole shit takes so long (bottlenecked at getting the KNN pipeline to predict shit)
0 reply
0 recast
1 reaction
TheModestThief
@thief
jumping into hyper(parameter search) space
1 reply
0 recast
2 reactions
TheModestThief
@thief
https://farcaster.xyz/thief/0xa8022751
0 reply
0 recast
0 reaction
TheModestThief
@thief
jupyter notebook prepares you to become a explosive ordnance disposal technician
0 reply
0 recast
1 reaction
TheModestThief
@thief
as useful as AI is to my learning so far, every now and then i think "gee this fucking thing is stupid as fuck"
0 reply
0 recast
6 reactions
TheModestThief
@thief
Another day of grinding: today’s aim is further refining the .py scripts to incorporate feedback. And start with the EDA for a new project altogether.
0 reply
0 recast
2 reactions
TheModestThief
@thief
a roundup of the ML-speedrunning course so far (linear regression): ## Jupyter Notebook ## 1. Exploratory Data Analysis (EDA): - gotten a grasp on EDA methods i.e. check/drop duplicates, deal with na values, check/deal with outliers (drop/cap/impute mean) - check distribution of target variable - if skewed (left/right) and apply an appropriate transformation (e.g. log) - univariate analysis/visualisation: trellis plots, scatterplots, using seaborn and plotly (for interactive, e.g. hovering) - bivariate analysis: correlation matrix (any highly correlated features? maybe drop 1 for redundancy) 2. Data-splitting 3. Feature scaling (normalise or standardise) - using standardscaler() to standardise necessary numerical values
1 reply
0 recast
3 reactions
TheModestThief
@thief
I absolutely hate Jupyter Notebook’s erratic behaviour around scrolling, adding new cells. I’ll scroll around and sometimes large cells just jump about. Had to turn my monitor to portrait orientation to combat this. What dog shit is this. Who is the CEO of EDA I’m writing a letter to complain.
1 reply
0 recast
1 reaction
TheModestThief
@thief
for once, thief knows what is what, this self guided learning course is definitely making progress
0 reply
0 recast
2 reactions
TheModestThief
@thief
a win is a win is a win go get it son
0 reply
0 recast
1 reaction
TheModestThief
@thief
realised some portions of machine learning basics i have already learnt it over the course of study, they are just named funny: e.g.: - you call independent variables & dependent variables features and labels. - loss/cost function is basically just derivatives and optimisation (excel's goalseek)
0 reply
0 recast
1 reaction
TheModestThief
@thief
some city pop to help me push through this linear algebra shit
0 reply
0 recast
0 reaction
