HelaLM-4B: LoRA Adapter for Sinhala Language on Gemma-3-4B

A LoRA adapter fine-tunes the Gemma-3-4B-Instruct model for the Sinhala language. The adapter is reported to be 92.3%-preferred in an open evaluation. The author, organization, and specific training data details are unknown.

Use Cases

• Generate Sinhala text based on the fine-tuned Gemma-3-4B model.
• Benchmark parameter-efficient fine-tuning techniques for low-resource languages.
• Serve as a component in a larger NLP pipeline for Sinhala content creation or translation.

Strengths

• Reported to achieve a 92.3% preference score in an open evaluation.
• Uses the Gemma-3-4B-Instruct model as a base, which is a known, capable foundation.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and license are unknown, which may limit suitability assessment.

Provenance

Source

kaggle

Collection Method

Likely fine-tuned via Low-Rank Adaptation (LoRA) on a Sinhala corpus.