Model reference · open weights
roberta-amharic-text-embedding is an open-weight embedding model from rasyosef. AxForge deploys and operates it for you on dedicated EU-owned hardware — with the licence handled where one is required.
Available as managed deployment — configured and operated for you on dedicated EU hardware, quoted per deployment.
What it is
| Released by | rasyosef |
|---|---|
| Type | Embedding models |
| Task | Embeddings |
| Parameters (lead) | 111M |
| Context | 512 tokens |
| Runs with | sentence-transformers |
| Based on | rasyosef/roberta-base-amharic |
| Released | 2025-01-04 |
| Popularity | 598 downloads / month |
| Licence | Open weights |
About
This is a sentence-transformers model finetuned from rasyosef/roberta-base-amharic on the amharic-passage-retrieval-dataset dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.
This model was trained as part of our ACL 2025 Findings paper: Optimized Text Embedding Models and Benchmarks for Amharic Passage Retrieval.
SentenceTransformer(
(0): Transformer({'max_seq_length': 510, 'do_lower_case': False}) with Transformer model: XLMRobertaModel
(1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
(2): Normalize()
)
First install the Sentence Transformers library:
pip install -U sentence-transformers
Then you can load this model and run inference.
from sentence_transformers import SentenceTransformer
# Download from the 🤗 Hub
model = SentenceTransformer("rasyosef/roberta-amharic-text-embedding-base")
# Run inference
sentences = [
"የተደጋገመው የመሬት መንቀጥቀጥና የእሳተ ገሞራ ምልክት በአፋር ክልል",
"በአክሱም ከተማ የሚገኙ ሙስሊም ሴት ተማሪዎች ከሒጃብ መልበስ ጋር በተያያዘ ውዝግብ ከትምህርት ገበታ ውጭ ሆነው እንደሚገኙ የትግራይ እስልምና ጉዳዮች ምክርቤት ስታወቀ። ይህን ለመፍታት ከክልሉ ትምህርት ቢሮ ጋር ንግግር ላይ መሆኑም የክልሉ እስልምና ጉዳዮች ምክርቤት ለዶቼቬለ ገልጿል።",
"በማዕከላዊ ኢትዮጵያ ክልል ሃድያ ዞን ጊቤ ወረዳ በሚገኙ 12 ቀበሌዎች መሠረታዊ የመንግሥት አገልግሎት መስጫ ተቋማት በሙሉና በከፊል በመዘጋታቸው መቸገራቸውን ነዋሪዎች አመለከቱ። ከባለፈው ዓመት ጀምሮ የጤና፣ የትምህርት እና የግብር አሰባሰብ ሥራዎች በየአካባቢያቸው እየተከናወኑ አለመሆናቸውንም ለዶቼ ቬለ ተናግረዋል።",
"የሕዝብ ተወካዮች ምክር ቤት አባል እና የቋሚ ኮሚቴ ሰብሳቢ የነበሩት አቶ ክርስቲያን ታደለ እና የአማራ ክልል ምክር ቤት አባል የሆኑት አቶ ዮሐንስ ቧያለው ከቃሊቲ ወደ ቂሊንጦ ማረሚያ ቤት መዛወራቸውን ጠበቃቸው ተናገሩ።",
"ከተደጋጋሚ መሬት መንቀጥቀጥ በኋላ አፋር ክልል እሳት ከመሬት ውስጥ ሲፈላ ታይቷል፡፡ ከመሬት ውስጥ እሳትና ጭስ የሚተፋው እንፋሎቱ ዛሬ ማለዳውን 11 ሰዓት ግድም ከከባድ ፍንዳታ በኋላየተስተዋለ መሆኑን የአከባቢው ነዋሪዎች እና ባለስልጣናት ለዶቼ ቬለ ተናግረዋል፡፡ አለት የሚያፈናጥር እሳት ነው የተባለው እንፋሎቱ በክልሉ ጋቢረሱ (ዞን 03) ዱለቻ ወረዳ ሰጋንቶ ቀበሌ መከሰቱን የገለጹት የአከባቢው የአይን እማኞች ከዋናው ፍንዳታ በተጨማሪ በዙሪያው ተጨማሪ ፍንዳታዎች መታየት ቀጥሏል ባይ ናቸው፡፡"
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [5, 768]
# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [5, 5]
dim_768{
"truncate_dim": 768
}
| Metric | Value |
|---|---|
| cosine_accuracy@1 | 0.7023 |
| cosine_accuracy@3 | 0.8336 |
| cosine_accuracy@5 | 0.8712 |
| cosine_accuracy@10 | 0.913 |
| cosine_precision@1 | 0.7023 |
| cosine_precision@3 | 0.2779 |
| cosine_precision@5 | 0.1742 |
| cosine_precision@10 | 0.0913 |
| cosine_recall@1 | 0.7023 |
| cosine_recall@3 | 0.8336 |
| cosine_recall@5 | 0.8712 |
| cosine_recall@10 | 0.913 |
| cosine_ndcg@10 | 0.8083 |
| cosine_mrr@10 | 0.7747 |
| cosine_map@100 | 0.7777 |
dim_256{
"truncate_dim": 256
}
| Metric | Value |
|---|---|
| cosine_accuracy@1 | 0.6951 |
| cosine_accuracy@3 | 0.8251 |
| cosine_accuracy@5 | 0.8649 |
| cosine_accuracy@10 | 0.9072 |
| cosine_precision@1 | 0.6951 |
| cosine_precision@3 | 0.275 |
| cosine_precision@5 | 0.173 |
| cosine_precision@10 | 0.0907 |
| cosine_recall@1 | 0.6951 |
| cosine_recall@3 | 0.8251 |
| cosine_recall@5 | 0.8649 |
| cosine_recall@10 | 0.9072 |
| cosine_ndcg@10 | 0.8024 |
| cosine_mrr@10 | 0.7686 |
| cosine_map@100 | 0.7719 |
From the published model card. Full card on the HuggingFace links in the sidebar.
Benchmarks
As published on the model card — the maker's own numbers, not measured by AxForge.
| Task | Dataset | Metric | Score |
|---|---|---|---|
| Information Retrieval | dim 768 | Cosine Accuracy@1 | 0.702 |
| Information Retrieval | dim 768 | Cosine Accuracy@3 | 0.834 |
| Information Retrieval | dim 768 | Cosine Accuracy@5 | 0.871 |
| Information Retrieval | dim 768 | Cosine Accuracy@10 | 0.913 |
| Information Retrieval | dim 768 | Cosine Precision@1 | 0.702 |
| Information Retrieval | dim 768 | Cosine Precision@3 | 0.278 |
| Information Retrieval | dim 768 | Cosine Precision@5 | 0.174 |
| Information Retrieval | dim 768 | Cosine Precision@10 | 0.091 |
| Information Retrieval | dim 768 | Cosine Recall@1 | 0.702 |
| Information Retrieval | dim 768 | Cosine Recall@3 | 0.834 |
| Information Retrieval | dim 768 | Cosine Recall@5 | 0.871 |
| Information Retrieval | dim 768 | Cosine Recall@10 | 0.913 |
| Information Retrieval | dim 768 | Cosine Ndcg@10 | 0.808 |
| Information Retrieval | dim 768 | Cosine Mrr@10 | 0.775 |
| Information Retrieval | dim 768 | Cosine Map@100 | 0.778 |
| Information Retrieval | dim 256 | Cosine Accuracy@1 | 0.695 |
| Information Retrieval | dim 256 | Cosine Accuracy@3 | 0.825 |
| Information Retrieval | dim 256 | Cosine Accuracy@5 | 0.865 |
| Information Retrieval | dim 256 | Cosine Accuracy@10 | 0.907 |
| Information Retrieval | dim 256 | Cosine Precision@1 | 0.695 |
| Information Retrieval | dim 256 | Cosine Precision@3 | 0.275 |
| Information Retrieval | dim 256 | Cosine Precision@5 | 0.173 |
| Information Retrieval | dim 256 | Cosine Precision@10 | 0.091 |
| Information Retrieval | dim 256 | Cosine Recall@1 | 0.695 |
Using it via the API
Once AxForge deploys roberta-amharic-text-embedding for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (roberta-amharic-text-embedding below is illustrative; you get the exact model name on deployment.)
$ curl -sS https://api.axforge.ai/v1/embeddings \
-H "Authorization: Bearer $AXFORGE_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"roberta-amharic-text-embedding","input":"text to embed"}'
Create an account — your API key is available in the console. 3M free tokens every 30 days with every new account.