Skip to content

Simulating systematic biases in neural scaling law fitting procedures, with focus on Chinchilla Approach 2

Notifications You must be signed in to change notification settings

Open-Athena/scaling-law-analysis

Repository files navigation

Scaling Law Analysis

Simulating systematic biases in neural scaling law fitting procedures, with focus on Chinchilla Approach 2.

Overview

This project uses synthetic loss surfaces to study how sampling choices and experimental design affect scaling exponent recovery. By controlling ground truth parameters, we isolate sources of bias in the IsoFLOP parabolic fitting method (Approach 2) without confounding from training noise.

The main output is a self-contained HTML article suitable for publication as a blog post. See the Article section below.

See specs/project.md for the full directory layout and implementation map.

Installation

uv sync

Article

The article is a single self-contained HTML file demonstrating systematic biases in Chinchilla Approach 2 using noise-free synthetic data. It lives at results/article/article.html (source) and results/article/article_standalone.html (deployable, with inlined images).

Sections cover: symmetric baselines, asymmetric surface errors, off-center sampling biases, a robust alternative via variable projection, and evidence from published scaling law studies.

See specs/build.md for the full build workflow, which covers:

  1. Running experiments
  2. Generating article figures and CSV data
  3. Generating the references list
  4. Syncing CSV data with article prose
  5. Building the standalone HTML and supplementary PDF

Experiments

Run all experiments:

uv run python -m scaling_law_analysis.experiments.run_all
Experiment Focus
Exp 1: Empirical Error How sampling range affects exponent recovery
Exp 2: Exponent Imbalance How α/β asymmetry amplifies fitting errors
Exp 3: Drift Sensitivity How systematic sampling center biases affect accuracy
Exp 4: Extrapolation Error How inference degrades when extrapolating beyond fitting range
Exp 5: Parametric Surface Whether direct surface fitting (variable projection) avoids Approach 2 biases
Exp 6: Analytical Error Analytical modeling of inference error (TODO)

About

Simulating systematic biases in neural scaling law fitting procedures, with focus on Chinchilla Approach 2

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •