Modelpublic

mechramc/kalavai-phase1-1b-code-specialist-seed42

sourceHugging Faceapache-2.0updated 6mo agoView on Hugging Face

0likes22downloads

Model Card

KALAVAI — Code Specialist (pythia-1b, seed 42)

Fine-tuned EleutherAI/pythia-1b on Code data as part of the KALAVAI decentralized cooperative training protocol.

—Paper: KALAVAI: Predicting When Independent Specialist Fusion Works -- A Quantitative Model for Post-Hoc Cooperative LLM Training
—GitHub: https://github.com/mechramc/Kalavai

Paper results

Phase 1 English domains at 1B scale. MoE fusion: +7.49% ±0.01pp over best specialist (3 seeds). Mean divergence 15.28%.

How to use

python

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("mechramc/kalavai-phase1-1b-code-specialist-seed42")
tokenizer = AutoTokenizer.from_pretrained("EleutherAI/pythia-1b")

This model is one specialist in a KALAVAI cooperative. To reproduce the MoE fusion results from the paper, load multiple domain specialists and combine them with a trained MoE router (see the paper and GitHub for details).

Citation

bibtex

@article{kumaresan2026kalavai,
  title     = {{KALAVAI}: Predicting When Independent Specialist Fusion Works
               --- A Quantitative Model for Post-Hoc Cooperative {LLM} Training},
  author    = {Kumaresan, Ramchand},
  journal   = {arXiv preprint arXiv:2603.22755},
  year      = {2026},
  url       = {https://arxiv.org/abs/2603.22755}
}