Hacker News front pageSep 17, 2026, 21:17 UTC+8

I had Gemini train its own replacement for $9

The author paid Gemini 3.1 Pro $9 to label 4,290 Reddit comments with knife brands, models, and steels, then fine-tuned GLiNER large v2.5 on those labels. The resulting model runs locally and hits 0.83 F1 against Gemini's labels after 24 minutes on a Tesla T4. Zero-shot GLiNER scored roughly 0.65 F1. The hardest bug was words_mask: the docs suggest a binary mask, but it's actually a word index; filling it with ones kept loss flat at 70. Five of ten runs produced no usable model—three config failures, two from words_mask. The post doesn't report human accuracy on Gemini's labels, so 0.83 is measured against Gemini, not ground truth.

Why it matters: A hands-on fine-tuning case with real numbers: $9 to distill Gemini labels into a local GLiNER model, lifting F1 from 0.65 to 0.83. Score capped because the domain (knife NER) is narrow, but the method transfers.

Read the original ↗Export Markdown

Take

I clicked because the cost breakdown is refreshingly concrete: $9 to label 4,290 Reddit comments, about $2.50 in T4 GPU time, 24 minutes of training, and a local GLiNER model that jumps from 0.65 zero-shot F1 to 0.83. The debugging log is the useful part—especially the words_mask trap. The docs suggest a binary mask, but it's actually a word index; filling it with ones kept loss flat at 70, and five of ten runs produced nothing usable.

I'd discount the F1 score, though. 0.83 is measured against Gemini's labels, and nobody checked whether Gemini was right. If Gemini hallucinated a brand, the model gets credit for copying the mistake. If the model fixed it, it gets penalized. So 0.83 really means "the model learned to mimic Gemini's judgment," not "the model learned to identify knife brands correctly."

Practically, if Gemini is already accurate enough on your task, this pipeline saves API money—break-even is around comment 4,291. But you'd want to spot-check Gemini's labels first. Bad labels at $0.0021 each are still bad labels.

Source: Hacker News front page · petervijeh.com
Published: Sep 17, 2026, 21:17 UTC+8 · 12 days ago