← Build a Jev-Style Decision Model from Scratch

🔒

Sign in to check your access, or get the course to unlock all lectures.

Section 5 · Lecture 3 · 8:30

train longer and temperature

← harder questions the permutation test →

Curriculum

0/23 done

0. Welcome

welcome1m

1. Reference Demonstration

2. Architecture Analysis

Materials for this section

1 notebook

01_the_picture.ipynb

3. Implementation

Materials for this section

1 notebook · 1 code file

jevlite.py
02_building_it.ipynb

4. Training

Materials for this section

1 notebook · 1 code file

data.py
03_training.ipynb

5. Calibration and Evaluation

Materials for this section

1 notebook · 9 code files · 6 files

experiments.py
experiments_epochs.py
experiments_shift.py
04_calibration.ipynb

Trained checkpoints (three epochs)6 code files · 6 files · 121.7 MB · in the course ZIP

6. Production Comparison and Serving

Materials for this section

2 code files

sdk_demo.py
serve.py

7. Deployment

Materials for this section

1 notebook · 1 code file

not_applicable.py
05_shipping.ipynb

Course-wide files

Applies to every section

1 doc · 1 code file

Sources for every claim
Python requirements

Progress is remembered in this browser only; a lecture ticks at 90% watched. Materials download after purchase.

train longer and temperature — Build a Jev-Style Decision Model from Scratch | SOTAAZ Blog