Dataset

NTL ATLAS Qwen3 8B v1

Contains 18 source-specific adapter weights for frozen Qwen3-8B-Base, trained on 2,699 official competition rows, plus inference code and config.

Kaggle ↗Oct 20, 2026Paper ↗

  • llm
  • adapter-weights
  • source-attribution
  • kaggle-competition
Training/inference pipeline diagram showing per-source LoRA method for source attribution
Training/inference pipeline diagram showing per-source LoRA method for source attribution

What it is

ATLAS adapter package for Qwen3-8B-Base, trained on 2,699 Name That LLM rows with 18 source-specific adapters plus inference code and provenance. 18 source-specific adapters for frozen Qwen3-8B-Base, trained 5 epochs on 2,699 official Name That LLM rows; includes weights, inference code, config, and provenance.

Key results

  • 18 source-specific adapters
  • Trained on 2,699 official training rows
  • Weights CC BY-SA 4.0; code Apache 2.0

Useful for reproducible LLM source attribution experiments in the Name That LLM competition.

One useful ML email a week.

Worth your time, not all of your time.

We track new ML work across papers, code, and data, then keep only the pieces worth a closer look. Selected, not collected.

Papers.Data.Code

The Weekly ML Signal

Selected, not collected.

One useful ML email a week: the papers, repos and datasets worth your attention.

Worth your time, not all of your time.