Fenua model training
Stage B bilingual adapter
A clearer view of the model run: loss, generation evals, dataset mix, and the community signals feeding the next Tuvaluan-English iteration.
TVL Stage B local snapshotFriezaForce/tvl-en-llm-translation-stage-b
Showing a packaged local snapshot because the live training backend is not reachable in this environment.
Train NLL1.3800down 7.4%
Step 8,400
Validation NLL1.6300
Step 8,000
chrF++41.8
Step 8,000
BLEU18.4
26.0% exact
Last updatedMay 15, 12:00 AM
CheckpointBest eval snapshot at 8,000
Sampler step8400
Loss curve
Training is moving in the right direction
Dataset
What the model is learning from
342,000 examplesEvaluations
Generation quality
| Step | chrF++ | BLEU | Exact |
|---|
| 8,000 | 41.8 | 18.4 | 26.0% |
| 6,000 | 37.8 | 15.6 | 21.0% |
| 4,000 | 31.2 | 10.9 | 14.0% |
| 2,000 | 24.6 | 6.8 | 8.0% |
134Ktranslation
72Kchat
54Kqa
36Ksummarization
26Kgrammar
20Ksafety