← back to paper

Review history

arxiv: 2506.15415 · 2 revisions

Targeted Lexical Injection: Unlocking Latent Cross-Lingual Alignment in Lugha-Llama via Early-Layer LoRA Fine-Tuning

1. 2026-08-15 REJECT MODERATE v1.4.0-alltime-deepseek-medium novelty 5.0
   141288 ms 9076 in 16387 out 2026-08-15T19:34:55.982310+00:00
2. 2026-08-06 REJECT MODERATE v1.4.0-alltime-deepseek-medium novelty 5.0
   122314 ms 9076 in 11371 out 2026-08-06T23:56:36.202860+00:00