elephant-rerank-v1-text-small is the text reranker model in the Agentic Intelligence Lab Elephant Rerank V1 family.
This ModelScope release is maintained by agentic-intelligence-lab to make Elephant rerank models easier to download and deploy in mainland China. It mirrors and renames the upstream HuggingFace model llm-semantic-router/mmbert-rerank-32k-2d-matryoshka under a consistent Elephant model namespace.
Positioning
This model is a multilingual long-context cross-encoder reranker for retrieval pipelines, agent memory systems, and RAG applications.
Embedding models are usually used for fast candidate generation. A reranker is used after that stage to score query-document pairs with higher precision. elephant-rerank-v1-text-small is designed for the second stage: take a query and a set of candidate passages, then assign relevance scores for final ordering.
The model is especially useful when passages are longer than the 512-token window used by many rerankers, or when relevant information may appear late in a document.
Agentic systems often retrieve many candidate memories, documents, tools, or execution traces before deciding what to use. The first retrieval stage needs to be fast; the final ordering stage needs to be precise. This reranker is designed for that final ordering stage.
Key advantages:
Long-context pair scoring: score query-passage pairs with up to 32K tokens of context.
Useful after vector retrieval: rerank candidates from Elephant embeddings or any other first-stage retriever.
2D Matryoshka flexibility: use different layer and dimension heads to trade quality for cost.
Multilingual coverage: suitable for mixed-language retrieval and international corpora.
Agent-friendly use cases: memory selection, tool ranking, evidence ordering, and long-document RAG.
Recommended use cases
Scenario
Recommendation
Long-document RAG
Rerank retrieved chunks or longer passages before generation
Agent memory recall
Reorder memory candidates by query relevance
Tool and skill ranking
Rank candidate tools after broad semantic retrieval
Evidence selection
Pick the strongest supporting records for answer synthesis
Multilingual search
Rerank candidates from mixed-language corpora
Quality-speed tuning
Use 2D Matryoshka layer/dimension heads for runtime budgets
Quick start on ModelScope
pip install modelscope transformers torch
This package contains the ModernBERT encoder weights plus the 2D Matryoshka classification heads. Loading the full reranker requires the custom reranker wrapper used by the upstream training/export code.
python
1import torch
2from modelscope import snapshot_download
3from transformers import AutoTokenizer
45# Use the reranker wrapper from the upstream training package.6# The wrapper is expected to load `model.safetensors`, `classification_heads.pt`,7# and `matryoshka_config.json` from the local model directory.8from train_rerank import Matryoshka2DReranker
910repo_id ="agentic-intelligence-lab/elephant-rerank-v1-text-small"11local_dir = snapshot_download(repo_id)1213model = Matryoshka2DReranker.from_pretrained(local_dir)14tokenizer = AutoTokenizer.from_pretrained(local_dir)1516model.eval()17device = torch.device("cuda"if torch.cuda.is_available()else"cpu")18model = model.to(device)1920pairs =[21(22"What is machine learning?",23"Machine learning is a subset of AI that enables systems to learn from data.",24),25(26"What is machine learning?",27"The weather is sunny today.",28),29]3031scores = model.compute_score(pairs, tokenizer, normalize=True)32print(scores)
2D Matryoshka scoring
The model provides multiple layer and dimension heads. This allows one checkpoint to serve several quality/cost profiles.
1defrerank(query:str, passages:list[str], top_k:int=10)->list[tuple[str,float]]:2 pairs =[(query, passage)for passage in passages]3 scores = model.compute_score(pairs, tokenizer, normalize=True)4 ranked =sorted(zip(passages, scores), key=lambda item: item[1], reverse=True)5return ranked[:top_k]67query ="How does photosynthesis work?"8passages =[9"Photosynthesis is the process by which plants convert sunlight into energy.",10"The stock market closed higher today.",11"Plants use chlorophyll to absorb light during photosynthesis.",12"Python is a popular programming language.",13]1415results = rerank(query, passages, top_k=2)16print(results)
Evaluation snapshot
Evaluation
Metric
Score
Long document, answer at start
Accuracy
100%
Long document, answer at end
Accuracy
100%
High-resource multilingual validation
Accuracy
100%
Low-resource multilingual validation
Accuracy
100%
BEIR SciFact
MRR
94.9
BEIR NFCorpus
MRR
87.2
BEIR HotpotQA
MRR
100.0
BEIR FiQA
MRR
93.9
The long-document validation checks whether the reranker can still find relevant information when it appears late in a long passage. This is the main reason to use this model over short-window rerankers in long-context RAG and memory workflows.
Files
File
Description
model.safetensors
ModernBERT encoder weights
classification_heads.pt
2D Matryoshka reranking heads
matryoshka_config.json
Layer/dimension head configuration
config.json
ModernBERT configuration
tokenizer.json / tokenizer_config.json
Tokenizer assets
training_args.json
Training/export configuration snapshot
README.md
This model card
Lineage
This ModelScope package is published by agentic-intelligence-lab as part of the Elephant model release line. It mirrors the upstream HuggingFace model llm-semantic-router/mmbert-rerank-32k-2d-matryoshka and keeps the model artifacts unchanged except for the repository naming and model card presentation.
The model is built from llm-semantic-router/mmbert-32k-yarn, a ModernBERT-based multilingual encoder extended to 32K context with YaRN position interpolation.
Limitations
This is a custom reranker export; the complete scoring path requires the upstream Matryoshka2DReranker wrapper or an equivalent implementation.
Training data is primarily based on BGE-M3 style query-passage pairs, so specialized domains may benefit from fine-tuning.
Although the model supports 32K tokens, very long query-passage pairs still increase compute and memory cost.
Layer and dimension reduction trade quality for efficiency and should be validated for each production workload.
For very short passages where latency is the only priority, a smaller short-window reranker may be faster.
