SentenceTransformer based on google/embeddinggemma-2

This is a sentence-transformers model finetuned from google/embeddinggemma-2 on the sentence-transformers/all-nli dataset. It maps inputs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, classification, clustering, and more.

Model Details

Model Description

Model Type: Sentence Transformer
Base model: google/embeddinggemma-2
Maximum Sequence Length: 1000000000000000019884624838656 tokens
Output Dimensionality: 768 dimensions
Similarity Function: Cosine Similarity
Supported Modalities: Text, Image, Audio, Video, Message
Training Dataset:
- sentence-transformers/all-nli
Language: en
License: apache-2.0

Model Sources

Documentation: Sentence Transformers Documentation
Repository: Sentence Transformers on GitHub
Hugging Face: Sentence Transformers on Hugging Face

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'transformer_task': 'feature-extraction', 'modality_config': {'text': {'method': 'forward', 'method_output_name': 'last_hidden_state'}, 'image': {'method': 'forward', 'method_output_name': 'last_hidden_state'}, 'audio': {'method': 'forward', 'method_output_name': 'last_hidden_state'}, 'video': {'method': 'forward', 'method_output_name': 'last_hidden_state'}, 'message': {'method': 'forward', 'method_output_name': 'last_hidden_state', 'format': 'structured'}}, 'module_output_name': 'token_embeddings', 'architecture': 'EmbeddingGemma2Model'})
  (1): Pooling({'embedding_dimension': 768, 'pooling_mode': 'mean', 'include_prompt': True})
  (2): Normalize({'module_input_name': 'sentence_embedding', 'module_output_name': 'sentence_embedding'})
)

Usage

Direct Usage (Sentence Transformers)

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("ML-Intern-lab/embeddinggemma-2-allnli-triplet")
# Run inference
queries = [
    'Two women are embracing while holding to go packages.',
]
documents = [
    'Two woman are holding packages.',
    'Two kids in numbered jerseys wash their hands.',
    'A man selling donuts to a customer.',
]
query_embeddings = model.encode_query(queries)
document_embeddings = model.encode_document(documents)
print(query_embeddings.shape, document_embeddings.shape)
# [1, 768] [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(query_embeddings, document_embeddings)
print(similarities)
# tensor([[0.8584, 0.3772, 0.3133]])

Evaluation

Metrics

Triplet

Dataset: all-nli-dev
Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.9459

Training Details

Training Dataset

sentence-transformers/all-nli

Dataset: sentence-transformers/all-nli
Size: 557,850 training samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 100 samples:

	anchor	positive	negative
type	string	string	string
modality	text	text	text
details	min: 8 tokens mean: 14.08 tokens max: 28 tokens	min: 6 tokens mean: 13.85 tokens max: 28 tokens	min: 5 tokens mean: 14.85 tokens max: 28 tokens

Samples:

anchor	positive	negative
`A person on a horse jumps over a broken down airplane.`	`A person is outdoors, on a horse.`	`A person is at a diner, ordering an omelette.`
`Children smiling and waving at camera`	`There are children present`	`The kids are frowning`
`A boy is jumping on skateboard in the middle of a red bridge.`	`The boy does a skateboarding trick.`	`The boy skates down the sidewalk.`

Loss: CachedMultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim",
    "mini_batch_size": 16,
    "mini_batch_num_tokens": null,
    "gather_across_devices": false,
    "directions": [
        "query_to_doc"
    ],
    "partition_mode": "joint",
    "hardness_mode": null,
    "hardness_strength": 0.0
}

Evaluation Dataset

sentence-transformers/all-nli

Dataset: sentence-transformers/all-nli
Size: 6,584 evaluation samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 100 samples:

	anchor	positive	negative
type	string	string	string
modality	text	text	text
details	min: 7 tokens mean: 17.35 tokens max: 38 tokens	min: 5 tokens mean: 9.86 tokens max: 28 tokens	min: 5 tokens mean: 10.28 tokens max: 18 tokens

Samples:

anchor	positive	negative
`Two women are embracing while holding to go packages.`	`Two woman are holding packages.`	`The men are fighting outside a deli.`
`Two young children in blue jerseys, one with the number 9 and one with the number 2 are standing on wooden steps in a bathroom and washing their hands in a sink.`	`Two kids in numbered jerseys wash their hands.`	`Two kids in jackets walk to school.`
`A man selling donuts to a customer during a world exhibition event held in the city of Angeles`	`A man selling donuts to a customer.`	`A woman drinks her coffee in a small cafe.`

Loss: CachedMultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim",
    "mini_batch_size": 16,
    "mini_batch_num_tokens": null,
    "gather_across_devices": false,
    "directions": [
        "query_to_doc"
    ],
    "partition_mode": "joint",
    "hardness_mode": null,
    "hardness_strength": 0.0
}

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 128
num_train_epochs: 1
max_steps: 1500
learning_rate: 2e-05
warmup_steps: 0.1
bf16: True
per_device_eval_batch_size: 128
batch_sampler: no_duplicates

All Hyperparameters

Click to expand

per_device_train_batch_size: 128
num_train_epochs: 1
max_steps: 1500
learning_rate: 2e-05
lr_scheduler_type: linear
lr_scheduler_kwargs: None
warmup_steps: 0.1
optim: adamw_torch_fused
optim_args: None
weight_decay: 0.0
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
optim_target_modules: None
gradient_accumulation_steps: 1
average_tokens_across_devices: True
max_grad_norm: 1.0
label_smoothing_factor: 0.0
bf16: True
fp16: False
bf16_full_eval: False
fp16_full_eval: False
tf32: None
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
torch_compile: False
torch_compile_backend: None
torch_compile_mode: None
use_liger_kernel: False
liger_kernel_config: None
use_cache: False
neftune_noise_alpha: None
torch_empty_cache_steps: None
auto_find_batch_size: False
log_on_each_node: True
logging_nan_inf_filter: True
include_num_input_tokens_seen: no
log_level: passive
log_level_replica: warning
disable_tqdm: False
project: huggingface
trackio_space_id: None
trackio_bucket_id: None
trackio_static_space_id: None
per_device_eval_batch_size: 128
prediction_loss_only: True
eval_on_start: False
eval_do_concat_batches: True
eval_use_gather_object: False
eval_accumulation_steps: None
include_for_metrics: []
batch_eval_metrics: False
save_only_model: False
save_on_each_node: False
enable_jit_checkpoint: False
push_to_hub: False
hub_private_repo: None
hub_model_id: None
hub_strategy: every_save
hub_always_push: False
hub_revision: None
load_best_model_at_end: False
ignore_data_skip: False
restore_callback_states_from_checkpoint: False
full_determinism: False
seed: 42
data_seed: None
use_cpu: False
accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
parallelism_config: None
dataloader_drop_last: False
dataloader_num_workers: 0
dataloader_pin_memory: True
dataloader_persistent_workers: False
dataloader_prefetch_factor: None
dataloader_multiprocessing_context: None
dataloader_in_order: True
remove_unused_columns: True
label_names: None
train_sampling_strategy: random
length_column_name: length
ddp_find_unused_parameters: None
ddp_bucket_cap_mb: None
ddp_broadcast_buffers: False
ddp_static_graph: None
ddp_backend: None
ddp_timeout: 1800
fsdp: None
fsdp_config: None
deepspeed: None
debug: []
skip_memory_metrics: True
do_predict: False
resume_from_checkpoint: None
local_rank: -1
prompts: None
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional
router_mapping: {}
learning_rate_mapping: {}
warmup_ratio: None

Training Logs

Epoch	Step	Training Loss	Validation Loss	all-nli-dev_cosine_accuracy
-1	-1	-	-	0.9036
0.0046	20	1.6320	-	-
0.0092	40	1.5520	-	-
0.0138	60	1.4255	-	-
0.0184	80	1.2147	-	-
0.0229	100	1.0517	-	-
0.0275	120	0.8570	-	-
0.0321	140	0.8000	-	-
0.0367	160	0.7947	-	-
0.0413	180	0.7525	-	-
0.0459	200	0.6815	-	-
0.0505	220	0.6914	-	-
0.0551	240	0.6813	-	-
0.0574	250	-	0.3524	0.9406
0.0596	260	0.6740	-	-
0.0642	280	0.6508	-	-
0.0688	300	0.6537	-	-
0.0734	320	0.6400	-	-
0.0780	340	0.6726	-	-
0.0826	360	0.6459	-	-
0.0872	380	0.6518	-	-
0.0918	400	0.6561	-	-
0.0964	420	0.6366	-	-
0.1009	440	0.6344	-	-
0.1055	460	0.6060	-	-
0.1101	480	0.6290	-	-
0.1147	500	0.7045	0.3279	0.9443
0.1193	520	0.6654	-	-
0.1239	540	0.6329	-	-
0.1285	560	0.6248	-	-
0.1331	580	0.6289	-	-
0.1376	600	0.6293	-	-
0.1422	620	0.6200	-	-
0.1468	640	0.6097	-	-
0.1514	660	0.6558	-	-
0.1560	680	0.6488	-	-
0.1606	700	0.6108	-	-
0.1652	720	0.6311	-	-
0.1698	740	0.6408	-	-
0.1721	750	-	0.3174	0.9449
0.1744	760	0.6349	-	-
0.1789	780	0.6083	-	-
0.1835	800	0.6153	-	-
0.1881	820	0.6040	-	-
0.1927	840	0.6180	-	-
0.1973	860	0.5697	-	-
0.2019	880	0.6652	-	-
0.2065	900	0.6085	-	-
0.2111	920	0.6082	-	-
0.2156	940	0.6263	-	-
0.2202	960	0.5992	-	-
0.2248	980	0.5950	-	-
0.2294	1000	0.5778	0.3149	0.9453
0.2340	1020	0.5890	-	-
0.2386	1040	0.6056	-	-
0.2432	1060	0.6154	-	-
0.2478	1080	0.6102	-	-
0.2524	1100	0.5502	-	-
0.2569	1120	0.6006	-	-
0.2615	1140	0.6149	-	-
0.2661	1160	0.5941	-	-
0.2707	1180	0.6153	-	-
0.2753	1200	0.5927	-	-
0.2799	1220	0.5759	-	-
0.2845	1240	0.5894	-	-
0.2868	1250	-	0.3132	0.9456
0.2891	1260	0.5852	-	-
0.2936	1280	0.5578	-	-
0.2982	1300	0.5786	-	-
0.3028	1320	0.5501	-	-
0.3074	1340	0.6244	-	-
0.3120	1360	0.6009	-	-
0.3166	1380	0.5955	-	-
0.3212	1400	0.5901	-	-
0.3258	1420	0.6017	-	-
0.3304	1440	0.6022	-	-
0.3349	1460	0.5904	-	-
0.3395	1480	0.6046	-	-
0.3441	1500	0.5862	0.3131	0.9459

Training Time

Training: 2.1 hours
Evaluation: 12.4 minutes
Total: 2.3 hours

Framework Versions

Python: 3.12.12
Sentence Transformers: 6.1.0
Transformers: 5.19.0
PyTorch: 2.14.1+cu130
Accelerate: 1.15.0
Datasets: 5.1.0
Tokenizers: 0.23.3

Additional Resources

Training and Finetuning Embedding Models with Sentence Transformers: the end-to-end guide for training or finetuning Sentence Transformer models.
Introduction to Matryoshka Embedding Models: variable-size embeddings that can be truncated with minimal quality loss.
Binary and Scalar Embedding Quantization for Significantly Faster & Cheaper Retrieval: post-training compression of embedding vectors.
Multimodal Embedding & Reranker Models with Sentence Transformers: use text, image, audio, and video models through the same API.
Training and Finetuning Multimodal Embedding & Reranker Models with Sentence Transformers: train multimodal embedding models, with a Visual Document Retrieval walkthrough.

Citation

BibTeX

Sentence Transformers

@inproceedings{reimers-2019-sentence-bert,
    title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/1908.10084",
}

CachedMultipleNegativesRankingLoss

@misc{gao2021scaling,
    title={Scaling Deep Contrastive Learning Batch Size under Memory Limited Setup},
    author={Luyu Gao and Yunyi Zhang and Jiawei Han and Jamie Callan},
    year={2021},
    eprint={2101.06983},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

MultipleNegativesRankingLoss

@misc{oord2019representationlearningcontrastivepredictive,
      title={Representation Learning with Contrastive Predictive Coding},
      author={Aaron van den Oord and Yazhe Li and Oriol Vinyals},
      year={2019},
      eprint={1807.03748},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/1807.03748},
}

Downloads last month: 26

Safetensors

Model size

0.3B params

Tensor type

BF16

Model tree for ML-Intern-lab/embeddinggemma-2-allnli-triplet

Base model

google/embeddinggemma-2

Finetuned

(123)

this model

Papers for ML-Intern-lab/embeddinggemma-2-allnli-triplet

Evaluation results

Cosine Accuracy on all nli dev
self-reported

0.946