model: mixedbread-ai/mxbai-edge-colbert-v0-32m and mixedbread-ai/mxbai-edge-colbert-v0-17m

mteb/models/model_implementations/mdbr_models.py

@@ -1,5 +1,7 @@
 from mteb.models.model_implementations.arctic_models import arctic_v1_training_datasets
-from mteb.models.model_implementations.mxbai_models import mixedbread_training_data
+from mteb.models.model_implementations.mixedbread_ai_models import (
+    mixedbread_training_data,
+)
 from mteb.models.model_meta import ModelMeta
 from mteb.models.sentence_transformer_wrapper import sentence_transformers_loader
 

mteb/models/model_implementations/mxbai_models.py → mteb/models/model_implementations/mixedbread_ai_models.py

@@ -1,3 +1,4 @@
+from mteb.models.model_implementations.pylate_models import MultiVectorModel
 from mteb.models.model_meta import (
     ModelMeta,
     ScoringFunction,
@@ -239,3 +240,93 @@
 }""",
     contacts=None,
 )
+
+mxbai_edge_colbert_v0_17m = ModelMeta(
+    loader=MultiVectorModel,
+    name="mixedbread-ai/mxbai-edge-colbert-v0-17m",
+    model_type=["late-interaction"],
+    languages=["eng-Latn"],
+    open_weights=True,
+    revision="23ae07f5bf028bc0d1f80c82e6e2dd2311f13a46",
+    public_training_code=None,
+    public_training_data=None,
+    release_date="2025-10-16",
+    n_parameters=int(17 * 1e6),
+    memory_usage_mb=64,
+    max_tokens=7999,
+    embed_dim=None,
+    license="apache-2.0",
+    similarity_fn_name=ScoringFunction.MAX_SIM,
+    framework=["PyLate", "ColBERT", "Transformers", "safetensors"],
+    reference="https://huggingface.co/mixedbread-ai/mxbai-edge-colbert-v0-17m",
+    use_instructions=False,
+    adapted_from="https://huggingface.co/jhu-clsp/ettin-encoder-17m",
+    superseded_by=None,
+    training_datasets={
+        "CornStack",
+        "MSMARCO",
+        "NQ",
+        "HotpotQA",
+        "AmazonQA",
+        "LoTTE",
+        "MultiLongDocRetrieval",
+        # "FineWeb",
+        # "PubMedQA",
+        # "TriviaQA",
+    },
+    citation="""@misc{takehi2025fantasticsmallretrieverstrain,
+      title={Fantastic (small) Retrievers and How to Train Them: mxbai-edge-colbert-v0 Tech Report},
+      author={Rikiya Takehi and Benjamin Clavié and Sean Lee and Aamir Shakir},
+      year={2025},
+      eprint={2510.14880},
+      archivePrefix={arXiv},
+      primaryClass={cs.IR},
+      url={https://arxiv.org/abs/2510.14880},
+}""",
+    contacts=None,
+)
+
+mxbai_edge_colbert_v0_32m = ModelMeta(
+    loader=MultiVectorModel,
+    name="mixedbread-ai/mxbai-edge-colbert-v0-32m",
+    model_type=["late-interaction"],
+    languages=["eng-Latn"],
+    open_weights=True,
+    revision="2f12870a85dae80680b9babc59992c9a2bc59e4a",
+    public_training_code=None,
+    public_training_data=None,
+    release_date="2025-10-16",
+    n_parameters=int(32 * 1e6),
+    memory_usage_mb=122,
+    max_tokens=511,
+    embed_dim=None,
+    license="apache-2.0",
+    similarity_fn_name=ScoringFunction.MAX_SIM,
+    framework=["PyLate", "ColBERT", "Transformers", "safetensors"],
+    reference="https://huggingface.co/mixedbread-ai/mxbai-edge-colbert-v0-32m",
+    use_instructions=False,
+    adapted_from="https://huggingface.co/jhu-clsp/ettin-encoder-32m",
+    superseded_by=None,
+    training_datasets={
+        "CornStack",
+        "MSMARCO",
+        "NQ",
+        "HotpotQA",
+        "AmazonQA",
+        "LoTTE",
+        "MultiLongDocRetrieval",
+        # "FineWeb",
+        # "PubMedQA",
+        # "TriviaQA",
+    },
+    citation="""@misc{takehi2025fantasticsmallretrieverstrain,
+      title={Fantastic (small) Retrievers and How to Train Them: mxbai-edge-colbert-v0 Tech Report},
+      author={Rikiya Takehi and Benjamin Clavié and Sean Lee and Aamir Shakir},
+      year={2025},
+      eprint={2510.14880},
+      archivePrefix={arXiv},
+      primaryClass={cs.IR},
+      url={https://arxiv.org/abs/2510.14880},
+}""",
+    contacts=None,
+)