EmbeddingGemma-300M
A small open-source multilingual embedding model released by Google with approximately 300 million parameters. It maps text into dense vectors for tasks such as semantic retrieval. Its training data consists primarily of high-resource languages.
Timeline (last 90 days)
EmbeddingGemma-300M is a small open-source embedding model released by Google with about 300 million parameters, supporting multiple languages but with training data dominated by high-resource languages.
Without fine-tuning, EmbeddingGemma-300M achieves only 51.7% to 60.8% recall@1 on the Belebele dataset for Tajik, Kazakh, Uzbek and Kyrgyz.
Saidzoda-CA-EmbeddingGemma-300M is a single multilingual embedding model fine-tuned from EmbeddingGemma-300M, serving four languages simultaneously: Tajik, Kazakh, Uzbek and Kyrgyz.
In testing on the full BEIR SciFact dataset (5,183 documents, 300 queries), Qwen3-Embedding-8B ranked first with 0.7953 nDCG@10, with EmbeddingGemma-300M close behind at 0.7861.
Parameter count does not reliably predict indexing speed: EmbeddingGemma-300M indexes more slowly than the larger BGE-M3 (315.0 vs 433.0 chunks per second).
All Facts (5)
EmbeddingGemma-300M is a small open-source embedding model released by Google with about 300 million parameters, supporting multiple languages but with training data dominated by high-resource languages.
50%UnverifiedWithout fine-tuning, EmbeddingGemma-300M achieves only 51.7% to 60.8% recall@1 on the Belebele dataset for Tajik, Kazakh, Uzbek and Kyrgyz.
50%UnverifiedSaidzoda-CA-EmbeddingGemma-300M is a single multilingual embedding model fine-tuned from EmbeddingGemma-300M, serving four languages simultaneously: Tajik, Kazakh, Uzbek and Kyrgyz.
50%UnverifiedIn testing on the full BEIR SciFact dataset (5,183 documents, 300 queries), Qwen3-Embedding-8B ranked first with 0.7953 nDCG@10, with EmbeddingGemma-300M close behind at 0.7861.
50%UnverifiedParameter count does not reliably predict indexing speed: EmbeddingGemma-300M indexes more slowly than the larger BGE-M3 (315.0 vs 433.0 chunks per second).
50%