mteb 2.7.0__py3-none-any.whl → 2.7.1__py3-none-any.whl

mteb/_evaluators/retrieval_metrics.py

@@ -140,7 +140,7 @@ def calculate_pmrr(original_run, new_run, changed_qrels):
     changes = []
     for qid in changed_qrels.keys():
         if qid + "-og" not in original_run or qid + "-changed" not in new_run:
-            logging.warning(f"Query {qid} not found in the runs for calculating p-MRR")
+            logger.warning(f"Query {qid} not found in the runs for calculating p-MRR")
             continue
         original_qid_run = original_run[qid + "-og"]
         new_qid_run = new_run[qid + "-changed"]

mteb/abstasks/retrieval_dataset_loaders.py

@@ -136,7 +136,7 @@ class RetrievalDatasetLoader:
                 "_id", "id"
             )
         logger.info("Loaded %d %s Documents.", len(corpus_ds), self.split.upper())
-        logger.info("Doc Example: %s", corpus_ds[0])
+        logger.debug("Doc Example: %s", corpus_ds[0])
         return corpus_ds
 
     def _load_queries(self) -> QueryDatasetType:
@@ -152,7 +152,7 @@ class RetrievalDatasetLoader:
             )
 
         logger.info("Loaded %d %s queries.", len(queries_ds), self.split.upper())
-        logger.info("Query Example: %s", queries_ds[0])
+        logger.debug("Query Example: %s", queries_ds[0])
 
         return queries_ds
 

mteb/abstasks/text/reranking.py

@@ -100,7 +100,7 @@ class AbsTaskReranking(AbsTaskRetrieval):
         if self.metadata.name not in OLD_FORMAT_RERANKING_TASKS:
             return
 
-        logging.info(
+        logger.info(
             f"Transforming old format to standard format for {self.metadata.name}"
         )
 

mteb/benchmarks/benchmarks/__init__.py

@@ -6,6 +6,7 @@ from mteb.benchmarks.benchmarks.benchmarks import (
     BUILT_MTEB,
     C_MTEB,
     CHEMTEB,
+    CHEMTEB_V1_1,
     CODE_RAG,
     ENCODECHKA,
     FA_MTEB,
@@ -70,6 +71,7 @@ __all__ = [
     "BRIGHT_LONG",
     "BUILT_MTEB",
     "CHEMTEB",
+    "CHEMTEB_V1_1",
     "CODE_RAG",
     "C_MTEB",
     "ENCODECHKA",

mteb/benchmarks/benchmarks/benchmarks.py

@@ -1656,6 +1656,7 @@ FA_MTEB_2 = Benchmark(
 
 CHEMTEB = Benchmark(
     name="ChemTEB",
+    aliases=["ChemTEB(v1)"],
     display_name="Chemical",
     icon="https://github.com/DennisSuitters/LibreICONS/raw/2d2172d15e3c6ca03c018629d60050e4b99e5c55/svg-color/libre-gui-purge.svg",
     tasks=get_tasks(
@@ -1701,6 +1702,62 @@ CHEMTEB = Benchmark(
 """,
 )
 
+CHEMTEB_V1_1 = Benchmark(
+    name="ChemTEB(v1.1)",
+    aliases=["ChemTEB(latest)"],
+    display_name="Chemical",
+    icon="https://github.com/DennisSuitters/LibreICONS/raw/2d2172d15e3c6ca03c018629d60050e4b99e5c55/svg-color/libre-gui-purge.svg",
+    tasks=get_tasks(
+        tasks=[
+            "PubChemSMILESBitextMining",
+            "SDSEyeProtectionClassification",
+            "SDSGlovesClassification",
+            "WikipediaBioMetChemClassification",
+            "WikipediaGreenhouseEnantiopureClassification",
+            "WikipediaSolidStateColloidalClassification",
+            "WikipediaOrganicInorganicClassification",
+            "WikipediaCryobiologySeparationClassification",
+            "WikipediaChemistryTopicsClassification",
+            "WikipediaTheoreticalAppliedClassification",
+            "WikipediaChemFieldsClassification",
+            "WikipediaLuminescenceClassification",
+            "WikipediaIsotopesFissionClassification",
+            "WikipediaSaltsSemiconductorsClassification",
+            "WikipediaBiolumNeurochemClassification",
+            "WikipediaCrystallographyAnalyticalClassification",
+            "WikipediaCompChemSpectroscopyClassification",
+            "WikipediaChemEngSpecialtiesClassification",
+            "WikipediaChemistryTopicsClustering",
+            "WikipediaSpecialtiesInChemistryClustering",
+            "PubChemAISentenceParaphrasePC",
+            "PubChemSMILESPC",
+            "PubChemSynonymPC",
+            "PubChemWikiParagraphsPC",
+            "PubChemWikiPairClassification",
+            "ChemNQRetrieval",
+            "ChemHotpotQARetrieval",
+            "ChemRxivRetrieval",
+        ],
+    ),
+    description="ChemTEB evaluates the performance of text embedding models on chemical domain data. This version adds the ChemRxivRetrieval task.",
+    reference="https://arxiv.org/abs/2412.00532",
+    citation=r"""
+@article{kasmaee2024chemteb,
+  author = {Kasmaee, Ali Shiraee and Khodadad, Mohammad and Saloot, Mohammad Arshi and Sherck, Nick and Dokas, Stephen and Mahyar, Hamidreza and Samiee, Soheila},
+  journal = {arXiv preprint arXiv:2412.00532},
+  title = {ChemTEB: Chemical Text Embedding Benchmark, an Overview of Embedding Models Performance \\& Efficiency on a Specific Domain},
+  year = {2024},
+}
+
+@article{kasmaee2025chembed,
+  author = {Kasmaee, Ali Shiraee and Khodadad, Mohammad and Astaraki, Mahdi and Saloot, Mohammad Arshi and Sherck, Nicholas and Mahyar, Hamidreza and Samiee, Soheila},
+  journal = {arXiv preprint arXiv:2508.01643},
+  title = {Chembed: Enhancing chemical literature search through domain-specific text embeddings},
+  year = {2025},
+}
+""",
+)
+
 BEIR_NL = Benchmark(
     name="BEIR-NL",
     display_name="BEIR-NL",
@@ -2350,17 +2407,16 @@ VIDORE_V3 = VidoreBenchmark(
         ]
     ),
     description="ViDoRe V3 sets a new industry gold standard for multi-modal, enterprise document visual retrieval evaluation. It addresses a critical challenge in production RAG systems: retrieving accurate information from complex, visually-rich documents. The benchmark includes both open and closed datasets: to submit results on private tasks, please [open an issue](https://github.com/embeddings-benchmark/mteb/issues?template=eval_request.yaml).",
-    reference="https://huggingface.co/blog/QuentinJG/introducing-vidore-v3",
+    reference="https://arxiv.org/abs/2601.08620",
     citation=r"""
-@misc{mace2025vidorev3,
-  author = {Macé, Quentin and Loison, Antonio and EDY, Antoine and Xing, Victor and Viaud, Gautier},
-  day = {5},
-  howpublished = {\url{https://huggingface.co/blog/QuentinJG/introducing-vidore-v3}},
-  journal = {Hugging Face Blog},
-  month = {November},
-  publisher = {Hugging Face},
-  title = {ViDoRe V3: a comprehensive evaluation of retrieval for enterprise use-cases},
-  year = {2025},
+@article{loison2026vidorev3comprehensiveevaluation,
+  archiveprefix = {arXiv},
+  author = {António Loison and Quentin Macé and Antoine Edy and Victor Xing and Tom Balough and Gabriel Moreira and Bo Liu and Manuel Faysse and Céline Hudelot and Gautier Viaud},
+  eprint = {2601.08620},
+  primaryclass = {cs.AI},
+  title = {ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios},
+  url = {https://arxiv.org/abs/2601.08620},
+  year = {2026},
 }
 """,
 )

mteb/descriptive_stats/Retrieval/ChemRxivRetrieval.json

@@ -0,0 +1,30 @@
+{
+    "test": {
+        "num_samples": 74457,
+        "number_of_characters": 76109543,
+        "documents_text_statistics": {
+            "total_text_length": 75549698,
+            "min_text_length": 121,
+            "average_text_length": 1087.7189916063176,
+            "max_text_length": 25438,
+            "unique_texts": 69150
+        },
+        "documents_image_statistics": null,
+        "queries_text_statistics": {
+            "total_text_length": 559845,
+            "min_text_length": 57,
+            "average_text_length": 111.969,
+            "max_text_length": 224,
+            "unique_texts": 5000
+        },
+        "queries_image_statistics": null,
+        "relevant_docs_statistics": {
+            "num_relevant_docs": 5000,
+            "min_relevant_docs_per_query": 1,
+            "average_relevant_docs_per_query": 1.0,
+            "max_relevant_docs_per_query": 1,
+            "unique_relevant_docs": 5000
+        },
+        "top_ranked_statistics": null
+    }
+}

mteb/descriptive_stats/Retrieval/EuroPIRQRetrieval.json

@@ -0,0 +1,116 @@
+{
+    "test": {
+        "num_samples": 30300,
+        "number_of_characters": 17320243,
+        "documents_text_statistics": {
+            "total_text_length": 17276572,
+            "min_text_length": 316,
+            "average_text_length": 575.8857333333333,
+            "max_text_length": 1008,
+            "unique_texts": 28361
+        },
+        "documents_image_statistics": null,
+        "queries_text_statistics": {
+            "total_text_length": 43671,
+            "min_text_length": 67,
+            "average_text_length": 145.57,
+            "max_text_length": 345,
+            "unique_texts": 300
+        },
+        "queries_image_statistics": null,
+        "relevant_docs_statistics": {
+            "num_relevant_docs": 300,
+            "min_relevant_docs_per_query": 1,
+            "average_relevant_docs_per_query": 1.0,
+            "max_relevant_docs_per_query": 1,
+            "unique_relevant_docs": 300
+        },
+        "top_ranked_statistics": null,
+        "hf_subset_descriptive_stats": {
+            "en": {
+                "num_samples": 10100,
+                "number_of_characters": 5517678,
+                "documents_text_statistics": {
+                    "total_text_length": 5503635,
+                    "min_text_length": 316,
+                    "average_text_length": 550.3635,
+                    "max_text_length": 726,
+                    "unique_texts": 9422
+                },
+                "documents_image_statistics": null,
+                "queries_text_statistics": {
+                    "total_text_length": 14043,
+                    "min_text_length": 68,
+                    "average_text_length": 140.43,
+                    "max_text_length": 305,
+                    "unique_texts": 100
+                },
+                "queries_image_statistics": null,
+                "relevant_docs_statistics": {
+                    "num_relevant_docs": 100,
+                    "min_relevant_docs_per_query": 1,
+                    "average_relevant_docs_per_query": 1.0,
+                    "max_relevant_docs_per_query": 1,
+                    "unique_relevant_docs": 100
+                },
+                "top_ranked_statistics": null
+            },
+            "fi": {
+                "num_samples": 10100,
+                "number_of_characters": 5953462,
+                "documents_text_statistics": {
+                    "total_text_length": 5938809,
+                    "min_text_length": 326,
+                    "average_text_length": 593.8809,
+                    "max_text_length": 1008,
+                    "unique_texts": 9422
+                },
+                "documents_image_statistics": null,
+                "queries_text_statistics": {
+                    "total_text_length": 14653,
+                    "min_text_length": 67,
+                    "average_text_length": 146.53,
+                    "max_text_length": 345,
+                    "unique_texts": 100
+                },
+                "queries_image_statistics": null,
+                "relevant_docs_statistics": {
+                    "num_relevant_docs": 100,
+                    "min_relevant_docs_per_query": 1,
+                    "average_relevant_docs_per_query": 1.0,
+                    "max_relevant_docs_per_query": 1,
+                    "unique_relevant_docs": 100
+                },
+                "top_ranked_statistics": null
+            },
+            "pt": {
+                "num_samples": 10100,
+                "number_of_characters": 5849103,
+                "documents_text_statistics": {
+                    "total_text_length": 5834128,
+                    "min_text_length": 325,
+                    "average_text_length": 583.4128,
+                    "max_text_length": 774,
+                    "unique_texts": 9517
+                },
+                "documents_image_statistics": null,
+                "queries_text_statistics": {
+                    "total_text_length": 14975,
+                    "min_text_length": 69,
+                    "average_text_length": 149.75,
+                    "max_text_length": 320,
+                    "unique_texts": 100
+                },
+                "queries_image_statistics": null,
+                "relevant_docs_statistics": {
+                    "num_relevant_docs": 100,
+                    "min_relevant_docs_per_query": 1,
+                    "average_relevant_docs_per_query": 1.0,
+                    "max_relevant_docs_per_query": 1,
+                    "unique_relevant_docs": 100
+                },
+                "top_ranked_statistics": null
+            }
+        }
+    }
+}

mteb/models/model_implementations/nvidia_llama_nemoretriever_colemb.py

@@ -1,17 +1,15 @@
-from typing import TYPE_CHECKING, Any
+from typing import Any
 
 import torch
+from packaging.version import Version
 from torch.utils.data import DataLoader
+from transformers import __version__ as transformers_version
 
 from mteb.abstasks.task_metadata import TaskMetadata
 from mteb.models.abs_encoder import AbsEncoder
 from mteb.models.model_meta import ModelMeta
 from mteb.types import Array, BatchedInput, PromptType
 
-if TYPE_CHECKING:
-    pass
-
-
 LLAMA_NEMORETRIEVER_CITATION = """@misc{xu2025llamanemoretrievercolembedtopperforming,
       title={Llama Nemoretriever Colembed: Top-Performing Text-Image Retrieval Model},
       author={Mengyao Xu and Gabriel Moreira and Ronay Ak and Radek Osmulski and Yauhen Babakhin and Zhiding Yu and Benedikt Schifferer and Even Oldridge},
@@ -34,6 +32,14 @@ class LlamaNemoretrieverColembed(AbsEncoder):
         attn_implementation="flash_attention_2",
         **kwargs,
     ):
+        required_transformers_version = "4.49.0"
+
+        if Version(transformers_version) != Version(required_transformers_version):
+            raise RuntimeError(
+                f"transformers version {transformers_version} is not match with required "
+                f"install version {required_transformers_version} to run `nvidia/llama-nemoretriever-colembed`"
+            )
+
         from transformers import AutoModel
 
         self.model = AutoModel.from_pretrained(

mteb/models/model_implementations/nvidia_models.py

@@ -10,8 +10,9 @@ from tqdm import tqdm
 from transformers import AutoModel, AutoTokenizer
 from transformers import __version__ as transformers_version
 
-from mteb import TaskMetadata
 from mteb._requires_package import requires_package
+from mteb.abstasks.task_metadata import TaskMetadata
+from mteb.models import CrossEncoderWrapper
 from mteb.models.abs_encoder import AbsEncoder
 from mteb.models.instruct_wrapper import InstructSentenceTransformerModel
 from mteb.models.model_meta import ModelMeta, ScoringFunction
@@ -20,23 +21,23 @@ from mteb.types import Array, BatchedInput, PromptType
 logger = logging.getLogger(__name__)
 
 NV_RETRIEVER_CITATION = """@misc{lee2025nvembedimprovedtechniquestraining,
-      title={NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models}, 
+      title={NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models},
       author={Chankyu Lee and Rajarshi Roy and Mengyao Xu and Jonathan Raiman and Mohammad Shoeybi and Bryan Catanzaro and Wei Ping},
       year={2025},
       eprint={2405.17428},
       archivePrefix={arXiv},
       primaryClass={cs.CL},
-      url={https://arxiv.org/abs/2405.17428}, 
+      url={https://arxiv.org/abs/2405.17428},
 }"""
 
 LlamaEmbedNemotron_CITATION = """@misc{babakhin2025llamaembednemotron8buniversaltextembedding,
-      title={Llama-Embed-Nemotron-8B: A Universal Text Embedding Model for Multilingual and Cross-Lingual Tasks}, 
+      title={Llama-Embed-Nemotron-8B: A Universal Text Embedding Model for Multilingual and Cross-Lingual Tasks},
       author={Yauhen Babakhin and Radek Osmulski and Ronay Ak and Gabriel Moreira and Mengyao Xu and Benedikt Schifferer and Bo Liu and Even Oldridge},
       year={2025},
       eprint={2511.07025},
       archivePrefix={arXiv},
       primaryClass={cs.CL},
-      url={https://arxiv.org/abs/2511.07025}, 
+      url={https://arxiv.org/abs/2511.07025},
 }"""
 
 
@@ -629,3 +630,55 @@ llama_embed_nemotron_8b = ModelMeta(
     contacts=["ybabakhin"],
     citation=LlamaEmbedNemotron_CITATION,
 )
+
+
+def _nemotron_rerank_model(model: str, revision: str, **kwargs) -> CrossEncoderWrapper:
+    required_transformers_version = "4.47.1"
+
+    if Version(transformers_version) != Version(required_transformers_version):
+        raise RuntimeError(
+            f"transformers version {transformers_version} is not match with required "
+            f"install version {required_transformers_version} to run `nvidia/llama-nemotron-rerank-1b-v2`"
+        )
+
+    return CrossEncoderWrapper(
+        model=model,
+        revision=revision,
+        **kwargs,
+    )
+
+
+nemotron_rerank_1b_v2 = ModelMeta(
+    loader=_nemotron_rerank_model,
+    loader_kwargs=dict(
+        trust_remote_code=True,
+        query_prefix="question:",
+        passage_prefix=" \n \n passage:",
+        model_kwargs={"torch_dtype": torch.float32},
+    ),
+    name="nvidia/llama-nemotron-rerank-1b-v2",
+    revision="78efcfdc23b53a753f6c73f2d78b18132a34ac4d",
+    release_date="2025-10-16",
+    languages=["eng-Latn"],
+    n_parameters=1235816448,
+    memory_usage_mb=2357.0,
+    max_tokens=4096,
+    embed_dim=2048,
+    license="https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/",
+    open_weights=True,
+    public_training_code=None,
+    public_training_data=None,
+    framework=["PyTorch", "Sentence Transformers"],
+    reference="https://huggingface.co/nvidia/llama-nemotron-rerank-1b-v2",
+    similarity_fn_name=ScoringFunction.COSINE,
+    use_instructions=None,
+    training_datasets=set(
+        # private
+    ),
+    adapted_from="meta-llama/Llama-3.2-1B",
+    superseded_by=None,
+    modalities=["text"],
+    model_type=["cross-encoder"],
+    citation=None,
+    contacts=None,
+)

mteb/models/model_implementations/voyage_models.py

@@ -25,6 +25,9 @@ VOYAGE_DTYPE_TRANSLATION = {
 
 # Total token limits per model based on VoyageAI documentation
 VOYAGE_TOTAL_TOKEN_LIMITS = {
+    "voyage-4-large": 120_000,
+    "voyage-4": 320_000,
+    "voyage-4-lite": 1_000_000,
     "voyage-3.5-lite": 1_000_000,
     "voyage-3.5": 320_000,
     "voyage-2": 320_000,
@@ -206,6 +209,32 @@ model_prompts = {
     PromptType.document.value: "document",
 }
 
+voyage_4 = ModelMeta(
+    name="voyageai/voyage-4",
+    model_type=["dense"],
+    revision="1",
+    release_date="2026-01-15",
+    languages=None,  # supported languages not specified
+    loader=VoyageModel,
+    loader_kwargs=dict(
+        max_tokens=32000,
+        model_prompts=model_prompts,
+    ),
+    max_tokens=32000,
+    embed_dim=1024,
+    open_weights=False,
+    n_parameters=None,
+    memory_usage_mb=None,
+    license=None,
+    reference="https://blog.voyageai.com/2026/01/15/voyage-4/",
+    similarity_fn_name="cosine",
+    framework=["API"],
+    use_instructions=True,
+    training_datasets=VOYAGE_TRAINING_DATA,
+    public_training_code=None,
+    public_training_data=None,
+)
+
 voyage_4_lite = ModelMeta(
     name="voyageai/voyage-4-lite",
     model_type=["dense"],
@@ -310,6 +339,7 @@ voyage_3_5 = ModelMeta(
     training_datasets=VOYAGE_TRAINING_DATA,
     public_training_code=None,
     public_training_data=None,
+    superseded_by="voyageai/voyage-4",
 )
 
 voyage_3_5_int8 = ModelMeta(

mteb/models/model_implementations/voyage_v.py

@@ -16,6 +16,8 @@ from mteb.types import Array, BatchedInput, PromptType
 if TYPE_CHECKING:
     from PIL import Image
 
+logger = logging.getLogger(__name__)
+
 
 def _downsample_image(
     image: Image.Image, max_pixels: int = 16000000, target_longest_side: int = 4000
@@ -37,17 +39,17 @@ def _downsample_image(
             new_width = int(width * (target_longest_side / height))
 
         new_size = (new_width, new_height)
-        logging.info(
+        logger.info(
             f"Downsampling image from {width}x{height} to {new_width}x{new_height}"
         )
         return image.resize(new_size, Image.LANCZOS)
     if width > height:
         if width > 10000:
-            logging.error("Processing extremely wide images.")
+            logger.error("Processing extremely wide images.")
             return image.resize((10000, height), Image.LANCZOS)
     else:
         if height > 10000:
-            logging.error("Processing extremely high images.")
+            logger.error("Processing extremely high images.")
             return image.resize((width, 10000), Image.LANCZOS)
     return image
 

mteb/models/model_meta.py

@@ -331,7 +331,7 @@ class ModelMeta(BaseModel):
                 revision = revisions[0].commit_id if revisions else None
 
             release_date = cls.fetch_release_date(model_name)
-            model_license = card_data.license
+            model_license = card_data.license if card_data.license != "other" else None
             n_parameters = cls._calculate_num_parameters_from_hub(model_name)
             memory_usage_mb = cls._calculate_memory_usage_mb(model_name, n_parameters)
             if model_config and hasattr(model_config, "hidden_size"):

mteb/models/sentence_transformer_wrapper.py

@@ -266,13 +266,24 @@ class SentenceTransformerMultimodalEncoderWrapper(SentenceTransformerEncoderWrap
 
 
 class CrossEncoderWrapper:
-    """Wrapper for CrossEncoder models."""
+    """Wrapper for CrossEncoder models.
+
+    Args:
+        model: The CrossEncoder model to use. Can be a string (model name) or a CrossEncoder model.
+        revision: The revision of the model to use.
+        device: The device used to load the model.
+        query_prefix: A prefix to add to all queries.
+        passage_prefix: A prefix to add to all passages.
+        **kwargs: Additional arguments to pass to the CrossEncoder model.
+    """
 
     def __init__(
         self,
         model: CrossEncoder | str,
         revision: str | None = None,
         device: str | None = None,
+        query_prefix: str = "",
+        passage_prefix: str = "",
         **kwargs,
     ) -> None:
         from sentence_transformers import CrossEncoder
@@ -283,6 +294,8 @@ class CrossEncoderWrapper:
             self.model = CrossEncoder(model, revision=revision, device=device, **kwargs)
 
         self.mteb_model_meta = ModelMeta.from_cross_encoder(self.model)
+        self.query_prefix = query_prefix
+        self.passage_prefix = passage_prefix
 
     def predict(
         self,
@@ -311,10 +324,10 @@ class CrossEncoderWrapper:
             The predicted relevance scores for each inputs pair.
         """
         all_queries_with_instructions = [
-            text for batch in inputs1 for text in batch["text"]
+            self.query_prefix + text for batch in inputs1 for text in batch["text"]
         ]
         all_corpus_with_instructions = [
-            text for batch in inputs2 for text in batch["text"]
+            self.passage_prefix + text for batch in inputs2 for text in batch["text"]
         ]
 
         return self.model.predict(

mteb/tasks/retrieval/eng/__init__.py

@@ -18,6 +18,7 @@ from .built_bench_retrieval import BuiltBenchRetrieval
 from .chat_doctor_retrieval import ChatDoctorRetrieval
 from .chem_hotpot_qa_retrieval import ChemHotpotQARetrieval
 from .chem_nq_retrieval import ChemNQRetrieval
+from .chemrxiv import ChemRxivRetrieval
 from .cirr_it2i_retrieval import CIRRIT2IRetrieval
 from .climate_fever_retrieval import (
     ClimateFEVER,
@@ -254,6 +255,7 @@ __all__ = [
     "ChatDoctorRetrieval",
     "ChemHotpotQARetrieval",
     "ChemNQRetrieval",
+    "ChemRxivRetrieval",
     "ClimateFEVER",
     "ClimateFEVERHardNegatives",
     "ClimateFEVERHardNegativesV2",

mteb/tasks/retrieval/eng/chemrxiv.py

@@ -0,0 +1,33 @@
+from mteb.abstasks.retrieval import AbsTaskRetrieval
+from mteb.abstasks.task_metadata import TaskMetadata
+
+
+class ChemRxivRetrieval(AbsTaskRetrieval):
+    metadata = TaskMetadata(
+        name="ChemRxivRetrieval",
+        dataset={
+            "path": "BASF-AI/ChemRxivRetrieval",
+            "revision": "5377aa18f309ec440ff6325a4c2cd3362c2cb8d7",
+        },
+        description="A retrieval task based on ChemRxiv papers where queries are LLM-synthesized to match specific paragraphs.",
+        reference="https://arxiv.org/abs/2508.01643",
+        type="Retrieval",
+        category="t2t",
+        modalities=["text"],
+        eval_splits=["test"],
+        eval_langs=["eng-Latn"],
+        main_score="ndcg_at_10",
+        date=("2025-01-01", "2025-05-01"),
+        domains=["Chemistry"],
+        task_subtypes=["Question answering", "Article retrieval"],
+        license="cc-by-nc-sa-4.0",
+        annotations_creators="LM-generated and reviewed",
+        dialect=[],
+        sample_creation="found",
+        bibtex_citation="""@article{kasmaee2025chembed,
+  author = {Kasmaee, Ali Shiraee and Khodadad, Mohammad and Astaraki, Mahdi and Saloot, Mohammad Arshi and Sherck, Nicholas and Mahyar, Hamidreza and Samiee, Soheila},
+  journal = {arXiv preprint arXiv:2508.01643},
+  title = {Chembed: Enhancing chemical literature search through domain-specific text embeddings},
+  year = {2025},
+}""",
+    )

mteb/tasks/retrieval/multilingual/__init__.py

@@ -6,6 +6,7 @@ from .cross_lingual_semantic_discrimination_wmt21 import (
     CrossLingualSemanticDiscriminationWMT21,
 )
 from .cur_ev1_retrieval import CUREv1Retrieval
+from .euro_pirq_retrieval import EuroPIRQRetrieval
 from .indic_qa_retrieval import IndicQARetrieval
 from .jina_vdr_bench_retrieval import (
     JinaVDRAirbnbSyntheticRetrieval,
@@ -107,6 +108,7 @@ __all__ = [
     "CUREv1Retrieval",
     "CrossLingualSemanticDiscriminationWMT19",
     "CrossLingualSemanticDiscriminationWMT21",
+    "EuroPIRQRetrieval",
     "IndicQARetrieval",
     "JinaVDRAirbnbSyntheticRetrieval",
     "JinaVDRArabicChartQARetrieval",

mteb/tasks/retrieval/multilingual/euro_pirq_retrieval.py

@@ -0,0 +1,43 @@
+from mteb.abstasks.retrieval import AbsTaskRetrieval
+from mteb.abstasks.task_metadata import TaskMetadata
+
+_LANGUAGES = {
+    "en": ["eng-Latn"],
+    "fi": ["fin-Latn"],
+    "pt": ["por-Latn"],
+}
+
+
+class EuroPIRQRetrieval(AbsTaskRetrieval):
+    metadata = TaskMetadata(
+        name="EuroPIRQRetrieval",
+        description="The EuroPIRQ retrieval dataset is a multilingual collection designed for evaluating retrieval and cross-lingual retrieval tasks. Dataset contains 10,000 parallel passages & 100 parallel queries (synthetic) in three languages: English, Portuguese, and Finnish, constructed from the European Union's DGT-Acquis corpus.",
+        reference="https://huggingface.co/datasets/eherra/EuroPIRQ-retrieval",
+        dataset={
+            "path": "eherra/EuroPIRQ-retrieval",
+            "revision": "59225ed25fbcea2185e1acbc8c3c80f1a8cd8341",
+        },
+        type="Retrieval",
+        category="t2t",
+        modalities=["text"],
+        eval_splits=["test"],
+        eval_langs=_LANGUAGES,
+        main_score="ndcg_at_10",
+        date=("2025-12-01", "2025-12-31"),
+        domains=["Legal"],
+        task_subtypes=[],
+        license="not specified",
+        annotations_creators="LM-generated and reviewed",
+        dialect=[],
+        sample_creation="found",
+        is_public=True,
+        bibtex_citation=r"""
+@misc{eherra_2025_europirq,
+  author = { {Elias Herranen} },
+  publisher = { Hugging Face },
+  title = { EuroPIRQ: European Parallel Information Retrieval Queries },
+  url = { https://huggingface.co/datasets/eherra/EuroPIRQ-retrieval },
+  year = {2025},
+}
+""",
+    )

mteb/tasks/retrieval/multilingual/vidore3_bench_retrieval.py

@@ -15,7 +15,7 @@ class Vidore3FinanceEnRetrieval(AbsTaskRetrieval):
     metadata = TaskMetadata(
         name="Vidore3FinanceEnRetrieval",
         description="Retrieve associated pages according to questions. This task, Finance - EN, is a corpus of reports from american banking companies, intended for long-document understanding tasks. Original queries were created in english, then translated to french, german, italian, portuguese and spanish.",
-        reference="https://huggingface.co/blog/QuentinJG/introducing-vidore-v3",
+        reference="https://arxiv.org/abs/2601.08620",
         dataset={
             "path": "vidore/vidore_v3_finance_en_mteb_format",
             "revision": "fa78cb14152b3dde8c5defdc4e3ddf50de69dfeb",
@@ -34,15 +34,14 @@ class Vidore3FinanceEnRetrieval(AbsTaskRetrieval):
         modalities=["text", "image"],
         sample_creation="created and machine-translated",
         bibtex_citation=r"""
-@misc{mace2025vidorev3,
-  author = {Macé, Quentin and Loison, Antonio and EDY, Antoine and Xing, Victor and Viaud, Gautier},
-  day = {5},
-  howpublished = {\url{https://huggingface.co/blog/QuentinJG/introducing-vidore-v3}},
-  journal = {Hugging Face Blog},
-  month = {November},
-  publisher = {Hugging Face},
-  title = {ViDoRe V3: a comprehensive evaluation of retrieval for enterprise use-cases},
-  year = {2025},
+@article{loison2026vidorev3comprehensiveevaluation,
+  archiveprefix = {arXiv},
+  author = {António Loison and Quentin Macé and Antoine Edy and Victor Xing and Tom Balough and Gabriel Moreira and Bo Liu and Manuel Faysse and Céline Hudelot and Gautier Viaud},
+  eprint = {2601.08620},
+  primaryclass = {cs.AI},
+  title = {ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios},
+  url = {https://arxiv.org/abs/2601.08620},
+  year = {2026},
 }
 """,
         prompt={"query": "Find a screenshot that is relevant to the user's question."},
@@ -53,7 +52,7 @@ class Vidore3FinanceFrRetrieval(AbsTaskRetrieval):
     metadata = TaskMetadata(
         name="Vidore3FinanceFrRetrieval",
         description="Retrieve associated pages according to questions. This task, Finance - FR, is a corpus of reports from french companies in the luxury domain, intended for long-document understanding tasks. Original queries were created in french, then translated to english, german, italian, portuguese and spanish.",
-        reference="https://huggingface.co/blog/QuentinJG/introducing-vidore-v3",
+        reference="https://arxiv.org/abs/2601.08620",
         dataset={
             "path": "vidore/vidore_v3_finance_fr_mteb_format",
             "revision": "8a2adfda85a7967c7252129703d9b3c7c9f038a9",
@@ -71,15 +70,14 @@ class Vidore3FinanceFrRetrieval(AbsTaskRetrieval):
         dialect=[],
         sample_creation="created and machine-translated",
         bibtex_citation=r"""
-@misc{mace2025vidorev3,
-  author = {Macé, Quentin and Loison, Antonio and EDY, Antoine and Xing, Victor and Viaud, Gautier},
-  day = {5},
-  howpublished = {\url{https://huggingface.co/blog/QuentinJG/introducing-vidore-v3}},
-  journal = {Hugging Face Blog},
-  month = {November},
-  publisher = {Hugging Face},
-  title = {ViDoRe V3: a comprehensive evaluation of retrieval for enterprise use-cases},
-  year = {2025},
+@article{loison2026vidorev3comprehensiveevaluation,
+  archiveprefix = {arXiv},
+  author = {António Loison and Quentin Macé and Antoine Edy and Victor Xing and Tom Balough and Gabriel Moreira and Bo Liu and Manuel Faysse and Céline Hudelot and Gautier Viaud},
+  eprint = {2601.08620},
+  primaryclass = {cs.AI},
+  title = {ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios},
+  url = {https://arxiv.org/abs/2601.08620},
+  year = {2026},
 }
 """,
         prompt={"query": "Find a screenshot that is relevant to the user's question."},
@@ -91,7 +89,7 @@ class Vidore3IndustrialRetrieval(AbsTaskRetrieval):
     metadata = TaskMetadata(
         name="Vidore3IndustrialRetrieval",
         description="Retrieve associated pages according to questions. This dataset, Industrial reports, is a corpus of technical documents on military aircraft (fueling, mechanics...), intended for complex-document understanding tasks. Original queries were created in english, then translated to french, german, italian, portuguese and spanish.",
-        reference="https://huggingface.co/blog/QuentinJG/introducing-vidore-v3",
+        reference="https://arxiv.org/abs/2601.08620",
         dataset={
             "path": "vidore/vidore_v3_industrial_mteb_format",
             "revision": "f732b725cf4a70803210edfe265a04f8bd5328f6",
@@ -110,15 +108,14 @@ class Vidore3IndustrialRetrieval(AbsTaskRetrieval):
         modalities=["text", "image"],
         sample_creation="created and machine-translated",
         bibtex_citation=r"""
-@misc{mace2025vidorev3,
-  author = {Macé, Quentin and Loison, Antonio and EDY, Antoine and Xing, Victor and Viaud, Gautier},
-  day = {5},
-  howpublished = {\url{https://huggingface.co/blog/QuentinJG/introducing-vidore-v3}},
-  journal = {Hugging Face Blog},
-  month = {November},
-  publisher = {Hugging Face},
-  title = {ViDoRe V3: a comprehensive evaluation of retrieval for enterprise use-cases},
-  year = {2025},
+@article{loison2026vidorev3comprehensiveevaluation,
+  archiveprefix = {arXiv},
+  author = {António Loison and Quentin Macé and Antoine Edy and Victor Xing and Tom Balough and Gabriel Moreira and Bo Liu and Manuel Faysse and Céline Hudelot and Gautier Viaud},
+  eprint = {2601.08620},
+  primaryclass = {cs.AI},
+  title = {ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios},
+  url = {https://arxiv.org/abs/2601.08620},
+  year = {2026},
 }
 """,
         prompt={"query": "Find a screenshot that is relevant to the user's question."},
@@ -130,7 +127,7 @@ class Vidore3PharmaceuticalsRetrieval(AbsTaskRetrieval):
     metadata = TaskMetadata(
         name="Vidore3PharmaceuticalsRetrieval",
         description="Retrieve associated pages according to questions. This dataset, Pharmaceutical, is a corpus of slides from the FDA, intended for long-document understanding tasks. Original queries were created in english, then translated to french, german, italian, portuguese and spanish.",
-        reference="https://huggingface.co/blog/QuentinJG/introducing-vidore-v3",
+        reference="https://arxiv.org/abs/2601.08620",
         dataset={
             "path": "vidore/vidore_v3_pharmaceuticals_mteb_format",
             "revision": "237ed4f43c7fb3c4df07ec4e9dd0a4366be555b0",
@@ -149,15 +146,14 @@ class Vidore3PharmaceuticalsRetrieval(AbsTaskRetrieval):
         modalities=["text", "image"],
         sample_creation="created and machine-translated",
         bibtex_citation=r"""
-@misc{mace2025vidorev3,
-  author = {Macé, Quentin and Loison, Antonio and EDY, Antoine and Xing, Victor and Viaud, Gautier},
-  day = {5},
-  howpublished = {\url{https://huggingface.co/blog/QuentinJG/introducing-vidore-v3}},
-  journal = {Hugging Face Blog},
-  month = {November},
-  publisher = {Hugging Face},
-  title = {ViDoRe V3: a comprehensive evaluation of retrieval for enterprise use-cases},
-  year = {2025},
+@article{loison2026vidorev3comprehensiveevaluation,
+  archiveprefix = {arXiv},
+  author = {António Loison and Quentin Macé and Antoine Edy and Victor Xing and Tom Balough and Gabriel Moreira and Bo Liu and Manuel Faysse and Céline Hudelot and Gautier Viaud},
+  eprint = {2601.08620},
+  primaryclass = {cs.AI},
+  title = {ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios},
+  url = {https://arxiv.org/abs/2601.08620},
+  year = {2026},
 }
 """,
         prompt={"query": "Find a screenshot that is relevant to the user's question."},
@@ -169,7 +165,7 @@ class Vidore3ComputerScienceRetrieval(AbsTaskRetrieval):
     metadata = TaskMetadata(
         name="Vidore3ComputerScienceRetrieval",
         description="Retrieve associated pages according to questions. This dataset, Computer Science, is a corpus of textbooks from the openstacks website, intended for long-document understanding tasks. Original queries were created in english, then translated to french, german, italian, portuguese and spanish.",
-        reference="https://huggingface.co/blog/QuentinJG/introducing-vidore-v3",
+        reference="https://arxiv.org/abs/2601.08620",
         dataset={
             "path": "vidore/vidore_v3_computer_science_mteb_format",
             "revision": "fb7fb69f81f7db62790f40494124b8ad22b424ab",
@@ -188,15 +184,14 @@ class Vidore3ComputerScienceRetrieval(AbsTaskRetrieval):
         modalities=["text", "image"],
         sample_creation="created and machine-translated",
         bibtex_citation=r"""
-@misc{mace2025vidorev3,
-  author = {Macé, Quentin and Loison, Antonio and EDY, Antoine and Xing, Victor and Viaud, Gautier},
-  day = {5},
-  howpublished = {\url{https://huggingface.co/blog/QuentinJG/introducing-vidore-v3}},
-  journal = {Hugging Face Blog},
-  month = {November},
-  publisher = {Hugging Face},
-  title = {ViDoRe V3: a comprehensive evaluation of retrieval for enterprise use-cases},
-  year = {2025},
+@article{loison2026vidorev3comprehensiveevaluation,
+  archiveprefix = {arXiv},
+  author = {António Loison and Quentin Macé and Antoine Edy and Victor Xing and Tom Balough and Gabriel Moreira and Bo Liu and Manuel Faysse and Céline Hudelot and Gautier Viaud},
+  eprint = {2601.08620},
+  primaryclass = {cs.AI},
+  title = {ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios},
+  url = {https://arxiv.org/abs/2601.08620},
+  year = {2026},
 }
 """,
         prompt={"query": "Find a screenshot that is relevant to the user's question."},
@@ -208,7 +203,7 @@ class Vidore3HrRetrieval(AbsTaskRetrieval):
     metadata = TaskMetadata(
         name="Vidore3HrRetrieval",
         description="Retrieve associated pages according to questions. This dataset, HR, is a corpus of reports released by the european union, intended for complex-document understanding tasks. Original queries were created in english, then translated to french, german, italian, portuguese and spanish.",
-        reference="https://huggingface.co/blog/QuentinJG/introducing-vidore-v3",
+        reference="https://arxiv.org/abs/2601.08620",
         dataset={
             "path": "vidore/vidore_v3_hr_mteb_format",
             "revision": "bc7d43d64815ed30f664168c8052106484aba7fd",
@@ -227,15 +222,14 @@ class Vidore3HrRetrieval(AbsTaskRetrieval):
         modalities=["text", "image"],
         sample_creation="created and machine-translated",
         bibtex_citation=r"""
-@misc{mace2025vidorev3,
-  author = {Macé, Quentin and Loison, Antonio and EDY, Antoine and Xing, Victor and Viaud, Gautier},
-  day = {5},
-  howpublished = {\url{https://huggingface.co/blog/QuentinJG/introducing-vidore-v3}},
-  journal = {Hugging Face Blog},
-  month = {November},
-  publisher = {Hugging Face},
-  title = {ViDoRe V3: a comprehensive evaluation of retrieval for enterprise use-cases},
-  year = {2025},
+@article{loison2026vidorev3comprehensiveevaluation,
+  archiveprefix = {arXiv},
+  author = {António Loison and Quentin Macé and Antoine Edy and Victor Xing and Tom Balough and Gabriel Moreira and Bo Liu and Manuel Faysse and Céline Hudelot and Gautier Viaud},
+  eprint = {2601.08620},
+  primaryclass = {cs.AI},
+  title = {ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios},
+  url = {https://arxiv.org/abs/2601.08620},
+  year = {2026},
 }
 """,
         prompt={"query": "Find a screenshot that is relevant to the user's question."},
@@ -247,7 +241,7 @@ class Vidore3EnergyRetrieval(AbsTaskRetrieval):
     metadata = TaskMetadata(
         name="Vidore3EnergyRetrieval",
         description="Retrieve associated pages according to questions. This dataset, Energy Fr, is a corpus of reports on energy supply in europe, intended for complex-document understanding tasks. Original queries were created in french, then translated to english, german, italian, portuguese and spanish.",
-        reference="https://huggingface.co/blog/QuentinJG/introducing-vidore-v3",
+        reference="https://arxiv.org/abs/2601.08620",
         dataset={
             "path": "vidore/vidore_v3_energy_mteb_format",
             "revision": "84fca99e5978604bae30f2436eacb6dbaa0532e9",
@@ -266,15 +260,14 @@ class Vidore3EnergyRetrieval(AbsTaskRetrieval):
         modalities=["text", "image"],
         sample_creation="created and machine-translated",
         bibtex_citation=r"""
-@misc{mace2025vidorev3,
-  author = {Macé, Quentin and Loison, Antonio and EDY, Antoine and Xing, Victor and Viaud, Gautier},
-  day = {5},
-  howpublished = {\url{https://huggingface.co/blog/QuentinJG/introducing-vidore-v3}},
-  journal = {Hugging Face Blog},
-  month = {November},
-  publisher = {Hugging Face},
-  title = {ViDoRe V3: a comprehensive evaluation of retrieval for enterprise use-cases},
-  year = {2025},
+@article{loison2026vidorev3comprehensiveevaluation,
+  archiveprefix = {arXiv},
+  author = {António Loison and Quentin Macé and Antoine Edy and Victor Xing and Tom Balough and Gabriel Moreira and Bo Liu and Manuel Faysse and Céline Hudelot and Gautier Viaud},
+  eprint = {2601.08620},
+  primaryclass = {cs.AI},
+  title = {ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios},
+  url = {https://arxiv.org/abs/2601.08620},
+  year = {2026},
 }
 """,
         prompt={"query": "Find a screenshot that is relevant to the user's question."},
@@ -286,7 +279,7 @@ class Vidore3PhysicsRetrieval(AbsTaskRetrieval):
     metadata = TaskMetadata(
         name="Vidore3PhysicsRetrieval",
         description="Retrieve associated pages according to questions. This dataset, Physics, is a corpus of course slides on french bachelor level physics lectures, intended for complex visual understanding tasks. Original queries were created in french, then translated to english, german, italian, portuguese and spanish.",
-        reference="https://huggingface.co/blog/QuentinJG/introducing-vidore-v3",
+        reference="https://arxiv.org/abs/2601.08620",
         dataset={
             "path": "vidore/vidore_v3_physics_mteb_format",
             "revision": "2c18ef90ab3ef93a9d86ecc6521cdae2a29f8300",
@@ -305,15 +298,14 @@ class Vidore3PhysicsRetrieval(AbsTaskRetrieval):
         modalities=["text", "image"],
         sample_creation="created and machine-translated",
         bibtex_citation=r"""
-@misc{mace2025vidorev3,
-  author = {Macé, Quentin and Loison, Antonio and EDY, Antoine and Xing, Victor and Viaud, Gautier},
-  day = {5},
-  howpublished = {\url{https://huggingface.co/blog/QuentinJG/introducing-vidore-v3}},
-  journal = {Hugging Face Blog},
-  month = {November},
-  publisher = {Hugging Face},
-  title = {ViDoRe V3: a comprehensive evaluation of retrieval for enterprise use-cases},
-  year = {2025},
+@article{loison2026vidorev3comprehensiveevaluation,
+  archiveprefix = {arXiv},
+  author = {António Loison and Quentin Macé and Antoine Edy and Victor Xing and Tom Balough and Gabriel Moreira and Bo Liu and Manuel Faysse and Céline Hudelot and Gautier Viaud},
+  eprint = {2601.08620},
+  primaryclass = {cs.AI},
+  title = {ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios},
+  url = {https://arxiv.org/abs/2601.08620},
+  year = {2026},
 }
 """,
         prompt={"query": "Find a screenshot that is relevant to the user's question."},
@@ -325,7 +317,7 @@ class Vidore3NuclearRetrieval(AbsTaskRetrieval):
     metadata = TaskMetadata(
         name="Vidore3NuclearRetrieval",
         description="Retrieve associated pages according to questions.",
-        reference="https://huggingface.co/blog/QuentinJG/introducing-vidore-v3",
+        reference="https://arxiv.org/abs/2601.08620",
         dataset={
             "path": "mteb-private/Vidore3NuclearRetrieval",
             "revision": "a463fc67fefc01152153101e88a32d5f9515e3e3",
@@ -344,15 +336,14 @@ class Vidore3NuclearRetrieval(AbsTaskRetrieval):
         modalities=["text", "image"],
         sample_creation="created and machine-translated",
         bibtex_citation=r"""
-@misc{mace2025vidorev3,
-  author = {Macé, Quentin and Loison, Antonio and EDY, Antoine and Xing, Victor and Viaud, Gautier},
-  day = {5},
-  howpublished = {\url{https://huggingface.co/blog/QuentinJG/introducing-vidore-v3}},
-  journal = {Hugging Face Blog},
-  month = {November},
-  publisher = {Hugging Face},
-  title = {ViDoRe V3: a comprehensive evaluation of retrieval for enterprise use-cases},
-  year = {2025},
+@article{loison2026vidorev3comprehensiveevaluation,
+  archiveprefix = {arXiv},
+  author = {António Loison and Quentin Macé and Antoine Edy and Victor Xing and Tom Balough and Gabriel Moreira and Bo Liu and Manuel Faysse and Céline Hudelot and Gautier Viaud},
+  eprint = {2601.08620},
+  primaryclass = {cs.AI},
+  title = {ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios},
+  url = {https://arxiv.org/abs/2601.08620},
+  year = {2026},
 }
 """,
         prompt={"query": "Find a screenshot that is relevant to the user's question."},
@@ -364,7 +355,7 @@ class Vidore3TelecomRetrieval(AbsTaskRetrieval):
     metadata = TaskMetadata(
         name="Vidore3TelecomRetrieval",
         description="Retrieve associated pages according to questions.",
-        reference="https://huggingface.co/blog/QuentinJG/introducing-vidore-v3",
+        reference="https://arxiv.org/abs/2601.08620",
         dataset={
             "path": "mteb-private/Vidore3TelecomRetrieval",
             "revision": "a54635a274ef2835721b7cbe3eb27483b9ec964b",
@@ -383,15 +374,14 @@ class Vidore3TelecomRetrieval(AbsTaskRetrieval):
         modalities=["text", "image"],
         sample_creation="created and machine-translated",
         bibtex_citation=r"""
-@misc{mace2025vidorev3,
-  author = {Macé, Quentin and Loison, Antonio and EDY, Antoine and Xing, Victor and Viaud, Gautier},
-  day = {5},
-  howpublished = {\url{https://huggingface.co/blog/QuentinJG/introducing-vidore-v3}},
-  journal = {Hugging Face Blog},
-  month = {November},
-  publisher = {Hugging Face},
-  title = {ViDoRe V3: a comprehensive evaluation of retrieval for enterprise use-cases},
-  year = {2025},
+@article{loison2026vidorev3comprehensiveevaluation,
+  archiveprefix = {arXiv},
+  author = {António Loison and Quentin Macé and Antoine Edy and Victor Xing and Tom Balough and Gabriel Moreira and Bo Liu and Manuel Faysse and Céline Hudelot and Gautier Viaud},
+  eprint = {2601.08620},
+  primaryclass = {cs.AI},
+  title = {ViDoRe V3: A Comprehensive Evaluation of Retrieval Augmented Generation in Complex Real-World Scenarios},
+  url = {https://arxiv.org/abs/2601.08620},
+  year = {2026},
 }
 """,
         prompt={"query": "Find a screenshot that is relevant to the user's question."},

{mteb-2.7.0.dist-info → mteb-2.7.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mteb
-Version: 2.7.0
+Version: 2.7.1
 Summary: Massive Text Embedding Benchmark
 Author-email: MTEB Contributors <niklas@huggingface.co>, Kenneth Enevoldsen <kenneth.enevoldsen@cas.au.dk>, Nouamane Tazi <nouamane@huggingface.co>, Nils Reimers <info@nils-reimers.de>
 Maintainer-email: Kenneth Enevoldsen <kenneth.enevoldsen@cas.au.dk>, Roman Solomatin <risolomatin@gmail.com>, Isaac Chung <chungisaac1217@gmail.com>

{mteb-2.7.0.dist-info → mteb-2.7.1.dist-info}/RECORD

@@ -21,7 +21,7 @@ mteb/_evaluators/clustering_evaluator.py,sha256=YfjwpZL03Tiu0jfd6HBSavrOs_WALwIp
 mteb/_evaluators/evaluator.py,sha256=CZIFM84KJcbygBqHrd6Reg0FA1ZT5dfVjFTuBYgGyMw,906
 mteb/_evaluators/pair_classification_evaluator.py,sha256=kgxvnzsZPFF7c6GNRWtyAP3AJ3v_heRQUYwmYvNATzI,6440
 mteb/_evaluators/retrieval_evaluator.py,sha256=UVrGtjOQdbDD2ZeH1oVm9ouIlU7Uup2a09ylzm0-QaM,3024
-mteb/_evaluators/retrieval_metrics.py,sha256=jd5BEinfGPdHPK3kXf6kfBs0gvDuBXEWyx-RS7G4c8g,23756
+mteb/_evaluators/retrieval_metrics.py,sha256=fwsPemssWqQRvo0t4PZTGrrv5KXK8N6U35kOFSkRbEU,23755
 mteb/_evaluators/sklearn_evaluator.py,sha256=lipgxkGXWeKGD2jhaNL9WQqLBS1_2L_WZpUobZR7iBI,3838
 mteb/_evaluators/zeroshot_classification_evaluator.py,sha256=Y5ZFUsBAWXMpKnZ6Iw1K0lm0_L4chLMxC4LAdZCBbPs,2296
 mteb/_evaluators/image/__init__.py,sha256=CsQd7OMkeV2Phun7paPWjayZ5qRnvj8H0TYBFeqMxag,148
@@ -43,7 +43,7 @@ mteb/abstasks/multilabel_classification.py,sha256=D_bdK3yyZFgaH1pg9PpO5LHQMVaJP3
 mteb/abstasks/pair_classification.py,sha256=-T8q7JQRjHZ3ihLqnuR2yuuci9xljvUqCldsHGnGLGc,13767
 mteb/abstasks/regression.py,sha256=sROjvfasLS89KRPUTsc1ONFsBTzfoqlpxLQfIkKBQXs,8763
 mteb/abstasks/retrieval.py,sha256=keuf7GeYyM0ODPlVFGZIFBmBoSil0c1glmcu6C0YzhA,26921
-mteb/abstasks/retrieval_dataset_loaders.py,sha256=WukcFAn54rUpXULCG43eysHozXHAxo2CaPhQyL_2Yg8,9401
+mteb/abstasks/retrieval_dataset_loaders.py,sha256=Jcr2hHe5sqWc8eL3WDhR9J7bJokD7tyhz_z3eD14o0M,9403
 mteb/abstasks/sts.py,sha256=b_19QvclU0_q0aEJdfvB3weUQ96bUS7uDuuRrjpz8Bc,9245
 mteb/abstasks/task_metadata.py,sha256=7TM_ls5bzYA1dHFq3VQgeioiyLrvMQz4i3hmWIsnD4M,27029
 mteb/abstasks/zeroshot_classification.py,sha256=O8jxoBgnrRx-BzOnr1aJVK3wIEBwkA4xuMxdVxlzJqI,6053
@@ -54,14 +54,14 @@ mteb/abstasks/image/__init__.py,sha256=NgvMJnp1g2mUv27RL-TvzA7s1BOdMG-EB1CrZfdbW
 mteb/abstasks/image/image_text_pair_classification.py,sha256=4RfPdAlb4ZlAE-9DiFQUvm7YpcQg9kTY7HRjRVenLmk,7935
 mteb/abstasks/text/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 mteb/abstasks/text/bitext_mining.py,sha256=0Ff7t1jMEonajHzB5CzZZoSMVkC5KTa0muoytcNzOjM,11152
-mteb/abstasks/text/reranking.py,sha256=QMgAAndGYRzvQdlhjLRMxrh_yrJZ0VQH40I-7mXo1O0,7872
+mteb/abstasks/text/reranking.py,sha256=vD5YUwst--zJ01bU40Z7juqQyuv8jrareRTtoCwt8E4,7871
 mteb/abstasks/text/summarization.py,sha256=qCUWfxaVAW0EYTFClUh9puBMGttyV6C7qKpCcHPAZN8,7148
 mteb/benchmarks/__init__.py,sha256=MQEVeli-zLaJ7Xg0z7RhXQwsdmm7Ht_W2Ln0rZo1Szc,225
 mteb/benchmarks/_create_table.py,sha256=b2RqGqi0ZonKbHecEcZiF4pkfE96smFRIzxOI82ETA8,22304
 mteb/benchmarks/benchmark.py,sha256=RheQOo0iQbu_ylN7oFLr2r-z_ahrMCTvKscVuwUx6yo,5694
 mteb/benchmarks/get_benchmark.py,sha256=nzR6cu5yXu1kIJKhd4A2R62xp43Z62bluPbOpNXHMWQ,2545
-mteb/benchmarks/benchmarks/__init__.py,sha256=NeFt0CGdY25-UWJ2aGCjDnpzr5-geih_1jjp1GQXJz4,2256
-mteb/benchmarks/benchmarks/benchmarks.py,sha256=w62HyautMAuRrwxUHjGpDqxTrXkb2zTD5cqXJV29Fyo,99644
+mteb/benchmarks/benchmarks/__init__.py,sha256=767VzDgayFSB3KJ05o0gFdOeo2f1EarbEfnCsy_karw,2294
+mteb/benchmarks/benchmarks/benchmarks.py,sha256=zAm1NTggaWb5Am6O30FUf-8zIbbF39eiuhSyLHzlpxQ,102370
 mteb/benchmarks/benchmarks/rteb_benchmarks.py,sha256=QnCSrTTaBfcRlAQp2Nu81tgv1idMXqiM16Fp2zKJ5Ys,10607
 mteb/cli/__init__.py,sha256=v-csUr3eUZElIvrGB6QGtaIdndDfNWEe9oZchsGsJpg,64
 mteb/cli/_display_tasks.py,sha256=pWKupzak8uxEIwJZbYpZpteeVprOgVT9Wr0HYeypitQ,2206
@@ -1082,6 +1082,7 @@ mteb/descriptive_stats/Retrieval/CUREv1.json,sha256=lfz-JMLuE5KlWqM08CUF80Nex1Oh
 mteb/descriptive_stats/Retrieval/ChatDoctorRetrieval.json,sha256=VyyOcmc9GQ1zqXrvAv0pFR5OAK-pE0fQYGJOsl4Wvok,991
 mteb/descriptive_stats/Retrieval/ChemHotpotQARetrieval.json,sha256=V2oT8C6Wh5VR6LeBlxrYH0VUJeEisjJELOZHpN2NP1o,2950
 mteb/descriptive_stats/Retrieval/ChemNQRetrieval.json,sha256=6E8SefjKJdqVC1R0JvGOPrJdpogDT-1H9X19GryVfI8,985
+mteb/descriptive_stats/Retrieval/ChemRxivRetrieval.json,sha256=kIfyVNgoLtO1QpnKguIovQVUio8RVdc1bBY_kwogT-Q,986
 mteb/descriptive_stats/Retrieval/ClimateFEVER-Fa.json,sha256=B1bp66dEPtY6p1zgHTK3fyBN_eFbmprpjhwUovVs4mI,1016
 mteb/descriptive_stats/Retrieval/ClimateFEVER-NL.json,sha256=RaHI8NhlXCD28FGFd1-Pofm4lvD-_Kcszde2OgPxQk8,1017
 mteb/descriptive_stats/Retrieval/ClimateFEVER-VN.json,sha256=D2xgItr4sBQiW3x8qr12firW0cXsOHygKZcDrtjhQ10,1016
@@ -1140,6 +1141,7 @@ mteb/descriptive_stats/Retrieval/EnglishFinance3Retrieval.json,sha256=IpCB7e6cwO
 mteb/descriptive_stats/Retrieval/EnglishFinance4Retrieval.json,sha256=tdSyer5_q9rPZSD0LEjjXluCbWY_W_FewhOTQycU1U0,995
 mteb/descriptive_stats/Retrieval/EnglishHealthcare1Retrieval.json,sha256=c_QduudRLaQVFARJxSc-YTFwHeNyhiu7IMv6JGlYC5c,1005
 mteb/descriptive_stats/Retrieval/EstQA.json,sha256=qqmmX7ExWg4152S6yiUAyHzaLS0I-QUaOn9KVBZ3UZ8,981
+mteb/descriptive_stats/Retrieval/EuroPIRQRetrieval.json,sha256=FIpF0h7hhF7O4jn-rEa9a-_8tPD1MAY-klTkSX3_SI8,4586
 mteb/descriptive_stats/Retrieval/FEVER-FaHardNegatives.json,sha256=KMDX6lg671e76rAboz30KgpvJAPxVSa6Thgefj1tgM4,987
 mteb/descriptive_stats/Retrieval/FEVER-NL.json,sha256=K810hO-zNJWg3-i30oma72wsjORGu8lLLd4PN_w0VVs,1020
 mteb/descriptive_stats/Retrieval/FEVER-VN.json,sha256=MnadQ0FGkXCFCDCI8eGwFYbouo8ta_1W_WlrqNx1clI,1018
@@ -1459,10 +1461,10 @@ mteb/models/__init__.py,sha256=ABTuoqiBjBtBWW3LYY7ItBHdylR6jWoy06HH0g6j6fU,910
 mteb/models/abs_encoder.py,sha256=6e9UAk7ckYsJ6hItIPMfSaPdlX_FxfJ-OdJ87oqT2OM,16625
 mteb/models/get_model_meta.py,sha256=wVh2FaWevJ10hJlbm-FQtTQazLMfnkEV3IK7PUyBPOQ,6082
 mteb/models/instruct_wrapper.py,sha256=GLHg9KcgYu2rF15LEMKesRpPudGfKE2y-aLXVG_CLj0,9670
-mteb/models/model_meta.py,sha256=x8EuA8Zpc4DqhK_50v5TAZ7n2J2yhHqf5U0ldCpPnw0,31101
+mteb/models/model_meta.py,sha256=BDT4Q_aYPm3RUgtXq2poe1VyUUHob_fLrDKA4BPgYrg,31143
 mteb/models/models_protocols.py,sha256=5WYOZw3-T-wK7ux0YZVCfbcMTkAisqAqbu44ZNoir4A,9250
 mteb/models/search_wrappers.py,sha256=9lXLXUyL6atMCwXp-HBUong6msT3UAUY9QI7zKXbSVU,20945
-mteb/models/sentence_transformer_wrapper.py,sha256=3sAev15a07ULA_ikFBBsMta0toy9AGPBbBPi37j_K-A,12932
+mteb/models/sentence_transformer_wrapper.py,sha256=liMzSqr0bC-yI_oEL50ckEc4ti9nibDLIiukP2crhOA,13543
 mteb/models/vllm_wrapper.py,sha256=ebX4JIKPoxW4PIlr3BnaoLGuMGRpHzBe_ZwvFscx1D0,12286
 mteb/models/cache_wrappers/__init__.py,sha256=1w1TnMwulWJSzNkLXjbh5MY3sqgHWc6vUntYn49i9X8,169
 mteb/models/cache_wrappers/cache_backend_protocol.py,sha256=iGWdqDEoaCxUVEnwsXhy-m9d2QX8KTaQ9m2ZyawrMes,1634
@@ -1545,8 +1547,8 @@ mteb/models/model_implementations/nbailab.py,sha256=LM00HJIr4yrA45qh2O21BIDXku9K
 mteb/models/model_implementations/no_instruct_sentence_models.py,sha256=qLiMok_OxKvIYXWnP0KNWqH1monZx-OdSZrSx3QEhtI,4049
 mteb/models/model_implementations/nomic_models.py,sha256=dmQC_cWg6hAmiBHK7fXoXEiGBJnJvrq0RsnCcJ2qe1Q,15137
 mteb/models/model_implementations/nomic_models_vision.py,sha256=usCKfZCR7aEi_DnNmVAYjH-lXx_ipQkBVtUAmhJ90QI,6870
-mteb/models/model_implementations/nvidia_llama_nemoretriever_colemb.py,sha256=DR66nmrIw1dgq7I1AcdgSC-ZqE29dsszVnHsrMxyCT4,6450
-mteb/models/model_implementations/nvidia_models.py,sha256=XnNiyByz6EFrISz1Msb-cXLVQfKnYP5HMRzAXC1KPDo,24937
+mteb/models/model_implementations/nvidia_llama_nemoretriever_colemb.py,sha256=pJqkKBNns6jUYlpI3aGtpmrhXOSYgEvCje1ir_yXdpc,6889
+mteb/models/model_implementations/nvidia_models.py,sha256=KcnH7wGhz4LQ2F_y9Q9cJTr89DNKeHjVkVgHUfftzIY,26685
 mteb/models/model_implementations/octen_models.py,sha256=FFR1-yG2euN-6kgL4qJNHYB6cPsds4NGYFPmc5tHhoE,8514
 mteb/models/model_implementations/openai_models.py,sha256=905BajYi_XyOZgqU3AeKpwIttLoUitaAyc48sTWI6Jg,9482
 mteb/models/model_implementations/openclip_models.py,sha256=MyosgeYSrgBXGuGFtI2Tyxksxpb7bADFJVSYFCLweVA,11622
@@ -1592,8 +1594,8 @@ mteb/models/model_implementations/vdr_models.py,sha256=8jlfABvO7Z9ebzAPFHqln3B2I
 mteb/models/model_implementations/vi_vn_models.py,sha256=Ep2zj4Xvjyu0a_YiLsYvolKdMGSOtzm-N-yNyXmfNwA,6328
 mteb/models/model_implementations/vista_models.py,sha256=GkQFHIwwjxwM0wDuo-dWJBo4dLExlHtHfXwhcdKA5uQ,10884
 mteb/models/model_implementations/vlm2vec_models.py,sha256=EeWl3kpS_1VDJs4t1QmpaWSuglLPL2GyZu27fVY1VT8,11802
-mteb/models/model_implementations/voyage_models.py,sha256=BdAHT7tpLVu9CMDdX9LzJKAJ6CncZKYIfMk2XdNKxV8,21707
-mteb/models/model_implementations/voyage_v.py,sha256=eFdSOKka5VoLjViZk5umlgTw_ETjyXv4yhZ9SoCR-p0,8124
+mteb/models/model_implementations/voyage_models.py,sha256=jwVjgx9E6-rZxd8wmuNbbQxKTobuWBNWrs0ezqp4Oik,22525
+mteb/models/model_implementations/voyage_v.py,sha256=JqtXnICeaODRZHBj_Xsaf3PfIG-XPKopblNxAXHqYNo,8159
 mteb/models/model_implementations/xyz_models.py,sha256=gjwCx3U4AxMcJDTSWVoYV6xeyXLw7lUZI5D6Q7JjWho,1322
 mteb/models/model_implementations/youtu_models.py,sha256=THwWRabutW-qC-JZOVhxXWjKHVyMElzt_xm81ixzN50,5995
 mteb/models/model_implementations/yuan_models.py,sha256=j-QIKECPg4TiBW_3Bp6g5yr2UOdFziFSeoGE4uKepSM,980
@@ -2208,7 +2210,7 @@ mteb/tasks/retrieval/deu/german_qu_ad_retrieval.py,sha256=RYZCPgKaNPi14ncxnvz8_7
 mteb/tasks/retrieval/deu/legal_qu_ad_retrieval.py,sha256=l_UNVkOQxqlKp6wef2BM-GKtyYnmYLYThGZZnUsR_-c,1594
 mteb/tasks/retrieval/ell/__init__.py,sha256=46naXAZtJzyezyqOQGRIlr4zQVkqGQJdj7ztjMrez9Y,72
 mteb/tasks/retrieval/ell/greek_civics_qa.py,sha256=uKmiWHHZvL1o5m17XCrzi16kHWFNhlEUHrtlk-LR1as,2514
-mteb/tasks/retrieval/eng/__init__.py,sha256=loq6PjQTX-ZpCUgy6wFZ74LCpZVmzbiRQKtJQlJbvQM,16093
+mteb/tasks/retrieval/eng/__init__.py,sha256=dVW7pHRu8SXp2_PJwn6InbfWVQ2VVy0PJvr_-bLBrc4,16158
 mteb/tasks/retrieval/eng/aila_casedocs_retrieval.py,sha256=UKoN9oE8C412REf8MV16aUDgE5NwkHxnXsh4dcLztpk,1398
 mteb/tasks/retrieval/eng/aila_statutes_retrieval.py,sha256=GugjZwaWmTlNyYzK8ACKZHIiUw2YBvLaVyTngN_qQyM,1366
 mteb/tasks/retrieval/eng/alpha_nli_retrieval.py,sha256=GpOkizyeOs_ZMbRu9g1yAXdKkQr8PObUepP1OulbJio,1714
@@ -2229,6 +2231,7 @@ mteb/tasks/retrieval/eng/built_bench_retrieval.py,sha256=pqsIaVMl0ugGl5wx1oHPooq
 mteb/tasks/retrieval/eng/chat_doctor_retrieval.py,sha256=_PQYn3jXgeLKjnb_uKxzHtM9SCICV3bsmgEcEoD2OXo,3582
 mteb/tasks/retrieval/eng/chem_hotpot_qa_retrieval.py,sha256=sLTfqBf967htSU7Ego7zkEC8QYVWFI12YoHxa-urWEw,2114
 mteb/tasks/retrieval/eng/chem_nq_retrieval.py,sha256=8bl4PRKJwYgNF0sZPZQINgn81-r3c_2gDoMQJYdpb8I,1886
+mteb/tasks/retrieval/eng/chemrxiv.py,sha256=-HS_axsMPaEKr8T0d9WvgfJ_UmAr00InHFjFpt4VDVo,1404
 mteb/tasks/retrieval/eng/cirr_it2i_retrieval.py,sha256=o_4fVGosZjYsfMQy7tzQGaq8aijhUwXKcV7MYIYY4SY,1583
 mteb/tasks/retrieval/eng/climate_fever_retrieval.py,sha256=obIROibY5nQ8oDGbkjG5Z3jrJ-rAXVmF_1w_XN6Nqm0,4615
 mteb/tasks/retrieval/eng/cqa_dupstack_android_retrieval.py,sha256=-Uj8BOIPyjl8egm34qZGRKULGgEoaoac4wOhdWXAPgE,1674
@@ -2393,11 +2396,12 @@ mteb/tasks/retrieval/kor/auto_rag_retrieval.py,sha256=tgffW8zMpDSv1FCOdS4_4SL5zK
 mteb/tasks/retrieval/kor/ko_strategy_qa.py,sha256=jk13ORetYtF0q36h8ljD6TeTHUwvK5F5ZbDoMCP3eWk,1156
 mteb/tasks/retrieval/kor/kovidore2_bench_retrieval.py,sha256=AjOLe2l9drBWOCeGzQqxfee3gwwU6ElAJ7-5pbFr6C8,6208
 mteb/tasks/retrieval/kor/squad_kor_v1_retrieval.py,sha256=M7T5FkN1efK7euRslx-LZN7hS_QdIwqtUuVlWO-dico,1631
-mteb/tasks/retrieval/multilingual/__init__.py,sha256=mfVGkoB4DO5ktlg8ia-4nImFVmZcqXh1XkgCkIff0tY,6765
+mteb/tasks/retrieval/multilingual/__init__.py,sha256=rbeuLmNYooHPjgROuEOH84Q6QmGhuXnedej0d6xAgqc,6841
 mteb/tasks/retrieval/multilingual/belebele_retrieval.py,sha256=gaVLEwuLEwMutMi9V-obpiYKbpllX2QNm2j3MVeebfE,7027
 mteb/tasks/retrieval/multilingual/cross_lingual_semantic_discrimination_wmt19.py,sha256=_6r34ZvRiLVENYcrd87NjilybGaetBwKFEbO29zYmBU,4676
 mteb/tasks/retrieval/multilingual/cross_lingual_semantic_discrimination_wmt21.py,sha256=Puy0PjpRr4M_Bbxdl7oWfa7pQGM04zaRaTNlnhyKejM,4677
 mteb/tasks/retrieval/multilingual/cur_ev1_retrieval.py,sha256=dwzo2sqjamM_xkSiC-jbapyhDFezSJpM4S8KfBsuLPk,4562
+mteb/tasks/retrieval/multilingual/euro_pirq_retrieval.py,sha256=rlbgWMRdQgDD8z4ZBPgGU1fRAqjmoFDzh0uD_P6qR-4,1602
 mteb/tasks/retrieval/multilingual/indic_qa_retrieval.py,sha256=K7iWZ-yTftZFQiXBOlkTJXGpQXs-ZFt6OQj_L6HjEwk,1872
 mteb/tasks/retrieval/multilingual/jina_vdr_bench_retrieval.py,sha256=dw2YuFrA5OWFTL5zREiix93oAj7WcpcnAhCRh8YRoHI,44579
 mteb/tasks/retrieval/multilingual/mintaka_retrieval.py,sha256=SwOliONITZM679LIBSMrvx_VymqE-zRN6YiYahhzfzw,2229
@@ -2414,7 +2418,7 @@ mteb/tasks/retrieval/multilingual/ru_sci_bench_retrieval.py,sha256=Mmcvrt_1cIxPf
 mteb/tasks/retrieval/multilingual/statcan_dialogue_dataset_retrieval.py,sha256=iFUQUlO_ogBdQBVYBQW3o-AJDQ792yg1pJtRxA5I3Qo,3796
 mteb/tasks/retrieval/multilingual/vdr_multilingual_retrieval.py,sha256=UduWKefwP7bPYxiDlztPEvSWXmTdw0xElglMbPY6XhA,4449
 mteb/tasks/retrieval/multilingual/vidore2_bench_retrieval.py,sha256=vOfiruHywYkP8pccdAuGLyYyFTw1zK0qcXDnUFA8Z5A,9091
-mteb/tasks/retrieval/multilingual/vidore3_bench_retrieval.py,sha256=wOoC--IVTz0dR6RMVICbz6OWxfCyVahGDSfX_TScCgA,16934
+mteb/tasks/retrieval/multilingual/vidore3_bench_retrieval.py,sha256=V3jtSlWhoKR1PCvHsH0HrONy-oFghomwqihBonQs_50,17414
 mteb/tasks/retrieval/multilingual/web_faq_retrieval.py,sha256=TM-Q98yXZny_PKHAFNEvw9o9ET_L6VM3aNis1NJ9DgM,2686
 mteb/tasks/retrieval/multilingual/wikipedia_retrieval_multilingual.py,sha256=zyqAt63bHXNU_I37jb891pwWUyGzZUGkXCyhWlRbed8,1569
 mteb/tasks/retrieval/multilingual/wit_t2i_retrieval.py,sha256=_swZhhMRs5OhzBdJVqQF1i9ZrTvAxaVrG0TpkPWkoHo,4359
@@ -2618,9 +2622,9 @@ mteb/types/_metadata.py,sha256=NN-W0S6a5TDV7UkpRx1pyWtGF4TyyCyoPUfHOwdeci8,2290
 mteb/types/_result.py,sha256=UKNokV9pu3G74MGebocU512aU_fFU9I9nPKnrG9Q0iE,1035
 mteb/types/_string_validators.py,sha256=PY-dYq4E8O50VS3bLYdldPWp400fl_WzUjfVSkNWe8U,523
 mteb/types/statistics.py,sha256=GwkBPmAr18Onu-vHtzHs0PFrhCozdOMiT13HwnWL4ZM,3961
-mteb-2.7.0.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-mteb-2.7.0.dist-info/METADATA,sha256=ww41PdZGaQnKWIX3vetD7jRnSf7O36TDKY7OSFGa1aE,14457
-mteb-2.7.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-mteb-2.7.0.dist-info/entry_points.txt,sha256=8IJoEJFKoDHmVnNev-qJ9pp4Ln7_1-ma9QsXnzVCzGU,39
-mteb-2.7.0.dist-info/top_level.txt,sha256=OLVIjcQAlWBz0bdmutKlWHLF42FF0hp4uVAg3ZyiG4U,5
-mteb-2.7.0.dist-info/RECORD,,
+mteb-2.7.1.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+mteb-2.7.1.dist-info/METADATA,sha256=olR8WXq0vPmp5FUUNww2hAMIuhr5irC4wnymgFN0xhY,14457
+mteb-2.7.1.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+mteb-2.7.1.dist-info/entry_points.txt,sha256=8IJoEJFKoDHmVnNev-qJ9pp4Ln7_1-ma9QsXnzVCzGU,39
+mteb-2.7.1.dist-info/top_level.txt,sha256=OLVIjcQAlWBz0bdmutKlWHLF42FF0hp4uVAg3ZyiG4U,5
+mteb-2.7.1.dist-info/RECORD,,