GitHub Repository

Intelligent training framework that automatically skips mastered samples and gives 5× more compute to hard ones. Up to 80% compute savings on LLM fine-tuning.

1 starsPython

GEKO (up to 80% compute savings on LLM fine-tuning)

Name: GEKO (up to 80% compute savings on LLM fine-tuning)
Availability: InStock
Author: SyedAbdurR2hman

by SyedAbdurR2hman·Feb 28, 2026·1 point·1 comment

Visit Project View on HN

AI Analysis

●●●BangerBig BrainWizardryShip It

Mountain Curriculum routing: 5× compute to hard samples, skip mastered ones.

Strengths

•Per-sample confidence tracking identifies mastery vs. struggle in real time
•Native LoRA/PEFT/BF16 support with torch.compile makes integration frictionless
•80% compute savings claim backed by versioned release with real benchmarks (v0.3.0)

Weaknesses

•Fine-tuning-only scope limits TAM; no pre-training guidance despite mentioning it
•Competitive field: LoRA, QLoRA, curriculum learning exist; novelty is orchestration not invention

Post Description

Hey HN, Most fine-tuning loops waste a huge amount of compute by treating every sample equally every epoch — even the ones the model has already mastered. I built GEKO (Gradient-Efficient Knowledge Optimization) to fix that. It tracks per-sample confidence and correctness in real time and:

Completely skips samples the model has mastered Gives up to 5× more compute to hard/confidently-wrong samples Dynamically adjusts sample weights using a "Mountain Curriculum" Just dropped v0.3.0 with native LoRA/PEFT, BF16, gradient checkpointing, torch.compile, and 8-bit optimizer support. I'm currently building a clean UI for it. I'm a 17-year-old indie dev working on this. Would love honest feedback, especially from people who do a lot of fine-tuning.