Language Specific Token Based Target Language Routing

via “language-specific token-based target language routing”

translation model by undefined. 13,09,929 downloads.

Unique: Uses learned language-specific tokens as a control mechanism rather than separate model heads or adapters, enabling zero-shot translation to unseen language pairs by leveraging the shared M2M-100 embedding space. This approach requires no architectural changes or additional parameters per language.

vs others: More flexible than single-language-pair models (no model switching overhead) but less robust than explicit language-specific fine-tuning, which would require separate model checkpoints per target language.

via “language-pair-routing-with-shared-vocabulary”

translation model by undefined. 4,72,848 downloads.

Unique: Uses a single shared vocabulary with explicit language tag tokens (e.g., '<2en>', '<2fr>') prepended to source text to condition the encoder on target language, rather than using separate decoder heads or routing logic; enables zero-shot translation through learned language representations in the shared embedding space

vs others: Simpler and more efficient than maintaining separate models per language pair or using pivot-language routing; more flexible than fixed language pair models while maintaining single-model deployment simplicity