mteb 2.5.2__py3-none-any.whl → 2.7.9__py3-none-any.whl

mteb/models/model_implementations/gme_v_models.py

@@ -2,19 +2,22 @@ from __future__ import annotations
 
 import logging
 import math
+import warnings
 from typing import TYPE_CHECKING, Any
 
 import torch
-from torch.utils.data import DataLoader
 from tqdm.autonotebook import tqdm
 
-from mteb.abstasks.task_metadata import TaskMetadata
 from mteb.models.abs_encoder import AbsEncoder
 from mteb.models.model_meta import ModelMeta, ScoringFunction
-from mteb.types import Array, BatchedInput, PromptType
+from mteb.types import PromptType
 
 if TYPE_CHECKING:
     from PIL import Image
+    from torch.utils.data import DataLoader
+
+    from mteb.abstasks.task_metadata import TaskMetadata
+    from mteb.types import Array, BatchedInput
 
 logger = logging.getLogger(__name__)
 
@@ -261,9 +264,9 @@ def smart_resize(
         w_bar = ceil_by_factor(width * beta, factor)
 
     if max(h_bar, w_bar) / min(h_bar, w_bar) > MAX_RATIO:
-        logger.warning(
-            f"Absolute aspect ratio must be smaller than {MAX_RATIO}, got {max(h_bar, w_bar) / min(h_bar, w_bar)}"
-        )
+        msg = f"Absolute aspect ratio must be smaller than {MAX_RATIO}, got {max(h_bar, w_bar) / min(h_bar, w_bar)}"
+        logger.warning(msg)
+        warnings.warn(msg)
         if h_bar > w_bar:
             h_bar = w_bar * MAX_RATIO
         else:
@@ -353,13 +356,14 @@ gme_qwen2vl_2b = ModelMeta(
     release_date="2024-12-24",
     modalities=["image", "text"],
     n_parameters=2_210_000_000,
+    n_embedding_parameters=233_373_696,
     memory_usage_mb=8427,
     embed_dim=1536,
     license="apache-2.0",
     max_tokens=32768,
     reference="https://huggingface.co/Alibaba-NLP/gme-Qwen2-VL-2B-Instruct",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["PyTorch"],
+    framework=["PyTorch", "Sentence Transformers", "safetensors", "Transformers"],
     use_instructions=True,
     public_training_code=None,
     public_training_data=None,
@@ -377,13 +381,14 @@ gme_qwen2vl_7b = ModelMeta(
     release_date="2024-12-24",
     modalities=["image", "text"],
     n_parameters=8_290_000_000,
+    n_embedding_parameters=544_997_376,
     memory_usage_mb=31629,
     embed_dim=3584,
     license="apache-2.0",
     max_tokens=32768,
     reference="https://huggingface.co/Alibaba-NLP/gme-Qwen2-VL-7B-Instruct",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["PyTorch"],
+    framework=["PyTorch", "Sentence Transformers", "safetensors", "Transformers"],
     use_instructions=True,
     public_training_code=None,
     public_training_data=None,

mteb/models/model_implementations/google_models.py

@@ -1,17 +1,23 @@
-from typing import Any
+from __future__ import annotations
+
+from typing import TYPE_CHECKING, Any
 
 import numpy as np
 from packaging.version import Version
-from torch.utils.data import DataLoader
 from tqdm.auto import tqdm
 from transformers import __version__ as transformers_version
 
 from mteb._requires_package import requires_package
-from mteb.abstasks.task_metadata import TaskMetadata
 from mteb.models import sentence_transformers_loader
 from mteb.models.abs_encoder import AbsEncoder
 from mteb.models.model_meta import ModelMeta, ScoringFunction
-from mteb.types import Array, BatchedInput, PromptType
+from mteb.types import PromptType
+
+if TYPE_CHECKING:
+    from torch.utils.data import DataLoader
+
+    from mteb.abstasks.task_metadata import TaskMetadata
+    from mteb.types import Array, BatchedInput
 
 MULTILINGUAL_EVALUATED_LANGUAGES = [
     "arb-Arab",
@@ -156,6 +162,7 @@ google_text_emb_004 = ModelMeta(
     revision="1",  # revision is intended for implementation
     release_date="2024-05-14",
     n_parameters=None,
+    n_embedding_parameters=None,
     memory_usage_mb=None,
     max_tokens=2048,
     embed_dim=768,
@@ -181,6 +188,7 @@ google_text_emb_005 = ModelMeta(
     revision="1",  # revision is intended for implementation
     release_date="2024-11-18",
     n_parameters=None,
+    n_embedding_parameters=None,
     memory_usage_mb=None,
     max_tokens=2048,
     embed_dim=768,
@@ -206,6 +214,7 @@ google_text_multilingual_emb_002 = ModelMeta(
     revision="1",
     release_date="2024-05-14",
     n_parameters=None,
+    n_embedding_parameters=None,
     memory_usage_mb=None,
     max_tokens=2048,
     embed_dim=768,
@@ -231,6 +240,7 @@ google_gemini_embedding_001 = ModelMeta(
     revision="1",
     release_date="2025-03-07",
     n_parameters=None,
+    n_embedding_parameters=None,
     memory_usage_mb=None,
     max_tokens=2048,
     embed_dim=3072,
@@ -266,11 +276,12 @@ embedding_gemma_300m = ModelMeta(
     revision="64614b0b8b64f0c6c1e52b07e4e9a4e8fe4d2da2",
     release_date="2025-09-04",
     n_parameters=307_581_696,
+    n_embedding_parameters=201_326_592,
     embed_dim=768,
     max_tokens=2048,
     license="gemma",
     reference="https://ai.google.dev/gemma/docs/embeddinggemma/model_card",
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "safetensors"],
     use_instructions=True,
     public_training_code=None,
     public_training_data=None,

mteb/models/model_implementations/granite_vision_embedding_models.py

@@ -4,20 +4,21 @@ import logging
 from typing import TYPE_CHECKING, Any
 
 import torch
-from torch.utils.data import DataLoader
 from tqdm.auto import tqdm
 
 from mteb._requires_package import (
     requires_image_dependencies,
 )
-from mteb.abstasks.task_metadata import TaskMetadata
 from mteb.models.model_meta import ModelMeta
-from mteb.types import Array, BatchedInput, PromptType
-
-logger = logging.getLogger(__name__)
 
 if TYPE_CHECKING:
     from PIL import Image
+    from torch.utils.data import DataLoader
+
+    from mteb.abstasks.task_metadata import TaskMetadata
+    from mteb.types import Array, BatchedInput, PromptType
+
+logger = logging.getLogger(__name__)
 
 
 class GraniteVisionEmbeddingWrapper:
@@ -172,6 +173,7 @@ granite_vision_embedding = ModelMeta(
     release_date="2025-06-11",
     modalities=["image", "text"],
     n_parameters=2_980_000_000,
+    n_embedding_parameters=None,
     memory_usage_mb=11351,
     max_tokens=128000,
     embed_dim=128,
@@ -179,7 +181,7 @@ granite_vision_embedding = ModelMeta(
     open_weights=True,
     public_training_code=None,
     public_training_data=None,
-    framework=["PyTorch"],
+    framework=["PyTorch", "Transformers", "safetensors"],
     reference="https://huggingface.co/ibm-granite/granite-vision-3.3-2b-embedding",
     similarity_fn_name="MaxSim",
     use_instructions=True,

mteb/models/model_implementations/gritlm_models.py

@@ -44,13 +44,14 @@ gritlm7b = ModelMeta(
     revision="13f00a0e36500c80ce12870ea513846a066004af",
     release_date="2024-02-15",
     n_parameters=7_240_000_000,
+    n_embedding_parameters=131_072_000,
     memory_usage_mb=13813,
     embed_dim=4096,
     license="apache-2.0",
     max_tokens=32768,
     reference="https://huggingface.co/GritLM/GritLM-7B",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["GritLM", "PyTorch"],
+    framework=["GritLM", "PyTorch", "Transformers", "safetensors"],
     use_instructions=True,
     training_datasets=GRIT_LM_TRAINING_DATA,
     # section 3.1 "We finetune our final models from Mistral 7B [68] and Mixtral 8x7B [69] using adaptations of E5 [160] and the Tülu 2 data
@@ -73,13 +74,15 @@ gritlm8x7b = ModelMeta(
     revision="7f089b13e3345510281733ca1e6ff871b5b4bc76",
     release_date="2024-02-15",
     n_parameters=57_920_000_000,
+    n_embedding_parameters=None,
+    n_active_parameters_override=13_000_000_000,
     memory_usage_mb=89079,
     embed_dim=32768,
     license="apache-2.0",
     max_tokens=32768,
     reference="https://huggingface.co/GritLM/GritLM-8x7B",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["GritLM", "PyTorch"],
+    framework=["GritLM", "PyTorch", "Transformers", "safetensors"],
     use_instructions=True,
     training_datasets=GRIT_LM_TRAINING_DATA,
     citation=GRITLM_CITATION,

mteb/models/model_implementations/gte_models.py

@@ -48,12 +48,13 @@ gte_qwen2_7b_instruct = ModelMeta(
     revision="e26182b2122f4435e8b3ebecbf363990f409b45b",
     release_date="2024-06-15",  # initial commit of hf model.
     n_parameters=7_613_000_000,
+    n_embedding_parameters=543_499_264,
     memory_usage_mb=29040,
     embed_dim=3584,
     license="apache-2.0",
     reference="https://huggingface.co/Alibaba-NLP/gte-Qwen2-7B-instruct",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "safetensors", "Transformers"],
     use_instructions=True,
     citation=GTE_CITATION,
     public_training_code=None,
@@ -80,13 +81,14 @@ gte_qwen1_5_7b_instruct = ModelMeta(
     revision="07d27e5226328010336563bc1b564a5e3436a298",
     release_date="2024-04-20",  # initial commit of hf model.
     n_parameters=7_720_000_000,
+    n_embedding_parameters=None,
     memory_usage_mb=29449,
     embed_dim=4096,
     license="apache-2.0",
     max_tokens=32_768,
     reference="https://huggingface.co/Alibaba-NLP/gte-Qwen1.5-7B-instruct",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "safetensors", "Transformers"],
     use_instructions=True,
     public_training_code=None,
     public_training_data=None,
@@ -117,13 +119,14 @@ gte_qwen2_1_5b_instruct = ModelMeta(
     revision="c6c1b92f4a3e1b92b326ad29dd3c8433457df8dd",
     release_date="2024-07-29",  # initial commit of hf model.
     n_parameters=1_780_000_000,
+    n_embedding_parameters=232_928_256,
     memory_usage_mb=6776,
     embed_dim=8960,
     license="apache-2.0",
     max_tokens=32_768,
     reference="https://huggingface.co/Alibaba-NLP/gte-Qwen2-1.5B-instruct",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "safetensors", "Transformers"],
     use_instructions=True,
     public_training_code=None,
     public_training_data=None,
@@ -145,13 +148,14 @@ gte_small_zh = ModelMeta(
     revision="af7bd46fbb00b3a6963c8dd7f1786ddfbfbe973a",
     release_date="2023-11-08",  # initial commit of hf model.
     n_parameters=int(30.3 * 1e6),
+    n_embedding_parameters=10_817_536,
     memory_usage_mb=58,
     embed_dim=1024,
     license="mit",
     max_tokens=512,
     reference="https://huggingface.co/thenlper/gte-small-zh",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "safetensors"],
     use_instructions=False,
     public_training_code=None,
     public_training_data=None,
@@ -173,13 +177,14 @@ gte_base_zh = ModelMeta(
     revision="71ab7947d6fac5b64aa299e6e40e6c2b2e85976c",
     release_date="2023-11-08",  # initial commit of hf model.
     n_parameters=int(102 * 1e6),
+    n_embedding_parameters=16_226_304,
     memory_usage_mb=195,
     embed_dim=1024,
     license="mit",
     max_tokens=512,
     reference="https://huggingface.co/thenlper/gte-base-zh",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "safetensors"],
     use_instructions=False,
     public_training_code=None,
     public_training_data=None,
@@ -201,13 +206,14 @@ gte_large_zh = ModelMeta(
     revision="64c364e579de308104a9b2c170ca009502f4f545",
     release_date="2023-11-08",  # initial commit of hf model.
     n_parameters=int(326 * 1e6),
+    n_embedding_parameters=21_635_072,
     memory_usage_mb=621,
     embed_dim=1024,
     license="mit",
     max_tokens=512,
     reference="https://huggingface.co/thenlper/gte-large-zh",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "safetensors"],
     use_instructions=False,
     public_training_code=None,
     public_training_data=None,
@@ -330,13 +336,14 @@ gte_multilingual_base = ModelMeta(
     revision="ca1791e0bcc104f6db161f27de1340241b13c5a4",
     release_date="2024-07-20",  # initial commit of hf model.
     n_parameters=int(305 * 1e6),
+    n_embedding_parameters=192_036_864,
     memory_usage_mb=582,
     embed_dim=768,
     license="apache-2.0",
     max_tokens=8192,
     reference="https://huggingface.co/Alibaba-NLP/gte-multilingual-base",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "safetensors", "Transformers"],
     use_instructions=False,
     public_training_code=None,
     public_training_data=None,  # couldn't find
@@ -359,13 +366,20 @@ gte_modernbert_base = ModelMeta(
     revision="7ca8b4ca700621b67618669f5378fe5f5820b8e4",
     release_date="2025-01-21",  # initial commit of hf model.
     n_parameters=int(149 * 1e6),
+    n_embedding_parameters=None,
     memory_usage_mb=284,
     embed_dim=768,
     license="apache-2.0",
     max_tokens=8192,
     reference="https://huggingface.co/Alibaba-NLP/gte-modernbert-base",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=[
+        "Sentence Transformers",
+        "PyTorch",
+        "Transformers",
+        "ONNX",
+        "safetensors",
+    ],
     use_instructions=False,
     public_training_code=None,  # couldn't find
     public_training_data=None,
@@ -396,13 +410,20 @@ gte_base_en_v15 = ModelMeta(
     revision="a829fd0e060bb84554da0dfd354d0de0f7712b7f",  # can be any
     release_date="2024-06-20",  # initial commit of hf model
     n_parameters=137_000_000,
+    n_embedding_parameters=23_445_504,
     memory_usage_mb=None,
     embed_dim=768,
     license="apache-2.0",
     max_tokens=8192,
     reference="https://huggingface.co/Alibaba-NLP/gte-base-en-v1.5",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=[
+        "Sentence Transformers",
+        "PyTorch",
+        "Transformers",
+        "ONNX",
+        "safetensors",
+    ],
     use_instructions=False,
     superseded_by=None,
     adapted_from=None,
@@ -410,21 +431,21 @@ gte_base_en_v15 = ModelMeta(
     public_training_data=None,
     training_datasets=None,
     citation="""@misc{zhang2024mgte,
-  title={mGTE: Generalized Long-Context Text Representation and Reranking Models for Multilingual Text Retrieval}, 
+  title={mGTE: Generalized Long-Context Text Representation and Reranking Models for Multilingual Text Retrieval},
   author={Xin Zhang and Yanzhao Zhang and Dingkun Long and Wen Xie and Ziqi Dai and Jialong Tang and Huan Lin and Baosong Yang and Pengjun Xie and Fei Huang and Meishan Zhang and Wenjie Li and Min Zhang},
   year={2024},
   eprint={2407.19669},
   archivePrefix={arXiv},
   primaryClass={cs.CL},
-  url={https://arxiv.org/abs/2407.19669}, 
+  url={https://arxiv.org/abs/2407.19669},
 }
 @misc{li2023gte,
-  title={Towards General Text Embeddings with Multi-stage Contrastive Learning}, 
+  title={Towards General Text Embeddings with Multi-stage Contrastive Learning},
   author={Zehan Li and Xin Zhang and Yanzhao Zhang and Dingkun Long and Pengjun Xie and Meishan Zhang},
   year={2023},
   eprint={2308.03281},
   archivePrefix={arXiv},
   primaryClass={cs.CL},
-  url={https://arxiv.org/abs/2308.03281}, 
+  url={https://arxiv.org/abs/2308.03281},
 }""",
 )

mteb/models/model_implementations/hinvec_models.py

@@ -1,9 +1,13 @@
+from __future__ import annotations
+
 import logging
+from typing import TYPE_CHECKING
 
 from mteb.models.model_meta import ModelMeta
 from mteb.models.sentence_transformer_wrapper import sentence_transformers_loader
-from mteb.types import PromptType
 
+if TYPE_CHECKING:
+    from mteb.types import PromptType
 logger = logging.getLogger(__name__)
 
 
@@ -43,13 +47,14 @@ Hinvec_bidir = ModelMeta(
     revision="d4fc678720cc1b8c5d18599ce2d9a4d6090c8b6b",
     release_date="2025-06-19",
     n_parameters=939_591_680,
+    n_embedding_parameters=None,
     memory_usage_mb=3715,
     embed_dim=2048,
     license="cc-by-nc-4.0",
     max_tokens=2048,
     reference="https://huggingface.co/Sailesh97/Hinvec",
     similarity_fn_name="cosine",
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "Transformers", "safetensors"],
     use_instructions=True,
     training_datasets=hinvec_training_datasets,
     public_training_code=None,

mteb/models/model_implementations/human.py

@@ -9,6 +9,7 @@ human = ModelMeta(
     revision="2025_09_25",
     release_date=None,
     n_parameters=None,
+    n_embedding_parameters=None,
     memory_usage_mb=None,
     embed_dim=None,
     license=None,

mteb/models/model_implementations/ibm_granite_models.py

@@ -100,13 +100,20 @@ granite_107m_multilingual = ModelMeta(
     revision="47db56afe692f731540413c67dd818ff492277e7",
     release_date="2024-12-18",
     n_parameters=107_000_000,
+    n_embedding_parameters=96_000_768,
     memory_usage_mb=204,
     embed_dim=384,
     license="apache-2.0",
     max_tokens=512,
     reference="https://huggingface.co/ibm-granite/granite-embedding-107m-multilingual",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=[
+        "Sentence Transformers",
+        "PyTorch",
+        "Transformers",
+        "ONNX",
+        "safetensors",
+    ],
     adapted_from=None,
     superseded_by=None,
     public_training_code=None,
@@ -125,13 +132,20 @@ granite_278m_multilingual = ModelMeta(
     revision="84e3546b88b0cb69f8078608a1df558020bcbf1f",
     release_date="2024-12-18",
     n_parameters=278_000_000,
+    n_embedding_parameters=192_001_536,
     memory_usage_mb=530,
     embed_dim=768,
     license="apache-2.0",
     max_tokens=512,
     reference="https://huggingface.co/ibm-granite/granite-embedding-278m-multilingual",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=[
+        "Sentence Transformers",
+        "PyTorch",
+        "Transformers",
+        "ONNX",
+        "safetensors",
+    ],
     adapted_from=None,
     superseded_by=None,
     public_training_code=None,
@@ -150,13 +164,20 @@ granite_30m_english = ModelMeta(
     revision="eddbb57470f896b5f8e2bfcb823d8f0e2d2024a5",
     release_date="2024-12-18",
     n_parameters=30_000_000,
+    n_embedding_parameters=19_301_760,
     memory_usage_mb=58,
     embed_dim=384,
     license="apache-2.0",
     max_tokens=512,
     reference="https://huggingface.co/ibm-granite/granite-embedding-30m-english",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=[
+        "Sentence Transformers",
+        "PyTorch",
+        "ONNX",
+        "safetensors",
+        "Transformers",
+    ],
     adapted_from=None,
     superseded_by=None,
     public_training_code=None,
@@ -175,13 +196,20 @@ granite_125m_english = ModelMeta(
     revision="e48d3a5b47eaa18e3fe07d4676e187fd80f32730",
     release_date="2024-12-18",
     n_parameters=125_000_000,
+    n_embedding_parameters=38_603_520,
     memory_usage_mb=238,
     embed_dim=768,
     license="apache-2.0",
     max_tokens=512,
     reference="https://huggingface.co/ibm-granite/granite-embedding-125m-english",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=[
+        "Sentence Transformers",
+        "PyTorch",
+        "ONNX",
+        "safetensors",
+        "Transformers",
+    ],
     adapted_from=None,
     superseded_by=None,
     public_training_code=None,
@@ -201,13 +229,14 @@ granite_english_r2 = ModelMeta(
     revision="6e7b8ce0e76270394ac4669ba4bbd7133b60b7f9",
     release_date="2025-08-15",
     n_parameters=149_000_000,
+    n_embedding_parameters=None,
     memory_usage_mb=284,
     embed_dim=768,
     license="apache-2.0",
     max_tokens=8192,
     reference="https://huggingface.co/ibm-granite/granite-embedding-english-r2",
     similarity_fn_name="cosine",
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "safetensors", "Transformers"],
     adapted_from=None,
     superseded_by=None,
     public_training_code=None,
@@ -226,13 +255,14 @@ granite_small_english_r2 = ModelMeta(
     revision="54a8d2616a0844355a5164432d3f6dafb37b17a3",
     release_date="2025-08-15",
     n_parameters=47_000_000,
+    n_embedding_parameters=None,
     memory_usage_mb=91,
     embed_dim=384,
     license="apache-2.0",
     max_tokens=8192,
     reference="https://huggingface.co/ibm-granite/granite-embedding-small-english-r2",
     similarity_fn_name="cosine",
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "safetensors", "Transformers"],
     adapted_from=None,
     superseded_by=None,
     public_training_code=None,

mteb/models/model_implementations/inf_models.py

@@ -56,13 +56,14 @@ inf_retriever_v1 = ModelMeta(
     revision="cb70ca7c31dfa866b2eff2dad229c144d8ddfd91",
     release_date="2024-12-24",  # initial commit of hf model.
     n_parameters=7_069_121_024,
+    n_embedding_parameters=None,
     memory_usage_mb=13483,
     embed_dim=3584,
     license="apache-2.0",
     max_tokens=32768,
     reference="https://huggingface.co/infly/inf-retriever-v1",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "safetensors", "Transformers"],
     use_instructions=True,
     adapted_from="Alibaba-NLP/gte-Qwen2-7B-instruct",
     public_training_code=None,
@@ -83,13 +84,14 @@ inf_retriever_v1_1_5b = ModelMeta(
     revision="c9c05c2dd50707a486966ba81703021ae2094a06",
     release_date="2025-02-08",  # initial commit of hf model.
     n_parameters=1_543_268_864,
+    n_embedding_parameters=232_928_256,
     memory_usage_mb=2944,
     embed_dim=1536,
     license="apache-2.0",
     max_tokens=32768,
     reference="https://huggingface.co/infly/inf-retriever-v1-1.5b",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "safetensors", "Transformers"],
     use_instructions=True,
     adapted_from="Alibaba-NLP/gte-Qwen2-1.5B-instruct",
     public_training_code=None,

mteb/models/model_implementations/jasper_models.py

@@ -1,11 +1,10 @@
+from __future__ import annotations
+
 import logging
-from collections.abc import Callable
-from typing import Any
+from typing import TYPE_CHECKING, Any
 
 import torch
-from torch.utils.data import DataLoader
 
-from mteb.abstasks.task_metadata import TaskMetadata
 from mteb.models.abs_encoder import AbsEncoder
 from mteb.models.instruct_wrapper import InstructSentenceTransformerModel
 from mteb.models.model_implementations.bge_models import (
@@ -17,7 +16,15 @@ from mteb.models.model_implementations.e5_instruct import E5_MISTRAL_TRAINING_DA
 from mteb.models.model_implementations.nvidia_models import nvidia_training_datasets
 from mteb.models.model_implementations.qzhou_models import qzhou_training_data
 from mteb.models.model_meta import ModelMeta, ScoringFunction
-from mteb.types import Array, BatchedInput, PromptType
+from mteb.types import PromptType
+
+if TYPE_CHECKING:
+    from collections.abc import Callable
+
+    from torch.utils.data import DataLoader
+
+    from mteb.abstasks.task_metadata import TaskMetadata
+    from mteb.types import Array, BatchedInput
 
 logger = logging.getLogger(__name__)
 
@@ -292,13 +299,14 @@ jasper_en_v1 = ModelMeta(
     revision="d6330ce98f8a0d741e781df845904c9484f00efa",
     release_date="2024-12-11",  # first commit
     n_parameters=1_999_000_000,
+    n_embedding_parameters=232_932_864,
     memory_usage_mb=3802,
     max_tokens=131072,
     embed_dim=8960,
     license="apache-2.0",
     reference="https://huggingface.co/infgrad/jasper_en_vision_language_v1",
     similarity_fn_name=ScoringFunction.COSINE,
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "safetensors"],
     use_instructions=True,
     adapted_from=None,
     superseded_by=None,
@@ -339,13 +347,14 @@ Jasper_Token_Compression_600M = ModelMeta(
     revision="06a100f753a5a96d9e583b3af79c6fcdfacc4719",
     release_date="2025-11-14",
     n_parameters=595776512,
+    n_embedding_parameters=None,
     memory_usage_mb=2272,
     embed_dim=2048,
     license="mit",
     max_tokens=32768,
     reference="https://huggingface.co/infgrad/Jasper-Token-Compression-600M",
     similarity_fn_name="cosine",
-    framework=["Sentence Transformers", "PyTorch"],
+    framework=["Sentence Transformers", "PyTorch", "safetensors"],
     use_instructions=True,
     public_training_code="https://github.com/DunZhang/Jasper-Token-Compression-Training",
     # public_training_data: unsupervised data for distillation