# The exact same perplexity chart, from data mix alone

> Source: <https://micjm.com/blog/same-perplexity-chart-data-mix/>
> Published: 2026-11-22 23:05:36+00:00

**Recommendation:** Control for the percentage of code in your pre-training mix before crediting perplexity gains on code benchmarks to scale or method.
More in the playbook "Evals and testing for LLM apps": <https://wpnews.pro/playbooks/evals-testing>

# The exact same perplexity chart, from data mix alone

Earlier today, I published [a blog post](https://micjm.com/blog/lowering-perplexity-without-scaling/) whose point was that OpenAI’s recent post showing pre-training progress was confounded by the percent of code in their pre-training corpus.

In the post, I demonstrated that if you vary the percent of code in your pre-training mix and you only evaluate log perplexity loss on a set of code, then you almost by definition get a reduction in log perplexity loss.

After publishing, I realized I could do one better and generate something in the shape of the exact same chart:
