ktcapraz/embeddinggemma-300m-german-search-categories

embeddinggemma-300m fine-tuned on German Search Categories

This is a sentence-transformers model finetuned from google/embeddinggemma-300m on the search-categories-german-triplets dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: google/embeddinggemma-300m
• Maximum Sequence Length: 2048 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • search-categories-german-triplets
• Language: de
• 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({'max_seq_length': 2048, 'do_lower_case': False, 'architecture': 'Gemma3TextModel'})
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
  (2): Dense({'in_features': 768, 'out_features': 3072, 'bias': False, 'activation_function': 'torch.nn.modules.linear.Identity'})
  (3): Dense({'in_features': 3072, 'out_features': 768, 'bias': False, 'activation_function': 'torch.nn.modules.linear.Identity'})
  (4): Normalize()
)

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("ktcapraz/embeddinggemma-300m-german-search-categories")
# Run inference
queries = [
    "Wann endet die Verhandlungsfrist f\u00fcr Spielerlisten 2024?",
]
documents = [
    'A broad, general question, related to a specific time.',
    'A specific question about a person or entity, with no time constraint.',
    'A keyword-based search for a broad topic, with no time constraint.',
]
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([[0.7721, 0.3525, 0.1172]])

Evaluation

Metrics

Information Retrieval

• Datasets: german-search-cats-eval and german-search-cats-test
• Evaluated with InformationRetrievalEvaluator

Metric	german-search-cats-eval	german-search-cats-test
cosine_accuracy@1	0.1989	0.7507
cosine_accuracy@3	0.5265	0.9884
cosine_accuracy@5	0.7575	0.9996
cosine_accuracy@10	1.0	1.0
cosine_precision@1	0.1989	0.7507
cosine_precision@3	0.1755	0.3295
cosine_precision@5	0.1515	0.1999
cosine_precision@10	0.1	0.1
cosine_recall@1	0.1989	0.7507
cosine_recall@3	0.5265	0.9884
cosine_recall@5	0.7575	0.9996
cosine_recall@10	1.0	1.0
cosine_ndcg@10	0.5624	0.9023
cosine_mrr@10	0.4255	0.8681
cosine_map@100	0.4255	0.8681

Information Retrieval

• Dataset: german-search-cats-eval
• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.7544
cosine_accuracy@3	0.988
cosine_accuracy@5	0.9997
cosine_accuracy@10	1.0
cosine_precision@1	0.7544
cosine_precision@3	0.3293
cosine_precision@5	0.1999
cosine_precision@10	0.1
cosine_recall@1	0.7544
cosine_recall@3	0.988
cosine_recall@5	0.9997
cosine_recall@10	1.0
cosine_ndcg@10	0.9035
cosine_mrr@10	0.8698
cosine_map@100	0.8698

Training Details

Training Dataset

search-categories-german-triplets

• Dataset: search-categories-german-triplets at bbf15f6
• Size: 73,141 training samples
• Columns: anchor, positive, and negatives
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negatives
  type	string	string	list
  details	min: 4 tokens mean: 12.97 tokens max: 25 tokens	min: 13 tokens mean: 16.08 tokens max: 18 tokens	size: 3 elements

• Samples:

  anchor	positive	negatives
  Fleischerzeugung Umweltkosten Deutschland	A keyword-based search for a broad topic, with no time constraint.	['A keyword-based search for a broad topic, related to a specific time.', 'A broad, general question, with no time constraint.', 'A specific keyword search for a person or entity, with no time constraint.']
  Schule Rösrath Mpox Schließung	A specific keyword search for a person or entity, with no time constraint.	['A specific keyword search for a person or entity, related to a specific time.', 'A specific question about a person or entity, with no time constraint.', 'A keyword-based search for a broad topic, with no time constraint.']
  Was ändert sich bei der Passbeantragung ab Mai 2024?	A broad, general question, related to a specific time.	['A keyword-based search for a broad topic, related to a specific time.', 'A broad, general question, with no time constraint.', 'A specific question about a person or entity, related to a specific time.']

• Loss: CachedMultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim",
      "mini_batch_size": 8,
      "gather_across_devices": false
  }
  

Evaluation Dataset

search-categories-german-triplets

• Dataset: search-categories-german-triplets at bbf15f6
• Size: 15,674 evaluation samples
• Columns: anchor, positive, and negatives
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negatives
  type	string	string	list
  details	min: 4 tokens mean: 12.59 tokens max: 26 tokens	min: 13 tokens mean: 16.21 tokens max: 18 tokens	size: 3 elements

• Samples:

  anchor	positive	negatives
  Weihnachtsgeschäft Einzelhandel November Dezember 2024	A specific keyword search for a person or entity, related to a specific time.	['A keyword-based search for a broad topic, related to a specific time.', 'A specific keyword search for a person or entity, with no time constraint.', 'A specific question about a person or entity, related to a specific time.']
  Wie reagiert die Union auf Mützenichs Vorstoß?	A specific question about a person or entity, with no time constraint.	['A specific keyword search for a person or entity, with no time constraint.', 'A specific question about a person or entity, related to a specific time.', 'A broad, general question, with no time constraint.']
  Al-Manar TV Sperrverfügung Telekom	A specific keyword search for a person or entity, with no time constraint.	['A specific keyword search for a person or entity, related to a specific time.', 'A specific question about a person or entity, with no time constraint.', 'A keyword-based search for a broad topic, with no time constraint.']

• Loss: CachedMultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim",
      "mini_batch_size": 8,
      "gather_across_devices": false
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 128
• per_device_eval_batch_size: 128
• learning_rate: 1e-05
• num_train_epochs: 1
• warmup_ratio: 0.1
• fp16: True
• prompts: {'anchor': 'task: search result | query: ', 'positive': 'task: classification | query: '}
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 128
• per_device_eval_batch_size: 128
• per_gpu_train_batch_size: None
• per_gpu_eval_batch_size: None
• gradient_accumulation_steps: 1
• eval_accumulation_steps: None
• torch_empty_cache_steps: None
• learning_rate: 1e-05
• weight_decay: 0.0
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1.0
• num_train_epochs: 1
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.1
• warmup_steps: 0
• log_level: passive
• log_level_replica: warning
• log_on_each_node: True
• logging_nan_inf_filter: True
• save_safetensors: True
• save_on_each_node: False
• save_only_model: False
• restore_callback_states_from_checkpoint: False
• no_cuda: False
• use_cpu: False
• use_mps_device: False
• seed: 42
• data_seed: None
• jit_mode_eval: False
• use_ipex: False
• bf16: False
• fp16: True
• fp16_opt_level: O1
• half_precision_backend: auto
• bf16_full_eval: False
• fp16_full_eval: False
• tf32: None
• local_rank: 0
• ddp_backend: None
• tpu_num_cores: None
• tpu_metrics_debug: False
• debug: []
• dataloader_drop_last: False
• dataloader_num_workers: 0
• dataloader_prefetch_factor: None
• past_index: -1
• disable_tqdm: False
• remove_unused_columns: True
• label_names: None
• load_best_model_at_end: False
• ignore_data_skip: False
• fsdp: []
• fsdp_min_num_params: 0
• fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
• fsdp_transformer_layer_cls_to_wrap: None
• 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
• deepspeed: None
• label_smoothing_factor: 0.0
• optim: adamw_torch_fused
• optim_args: None
• adafactor: False
• group_by_length: False
• length_column_name: length
• ddp_find_unused_parameters: None
• ddp_bucket_cap_mb: None
• ddp_broadcast_buffers: False
• dataloader_pin_memory: True
• dataloader_persistent_workers: False
• skip_memory_metrics: True
• use_legacy_prediction_loop: False
• push_to_hub: False
• resume_from_checkpoint: None
• hub_model_id: None
• hub_strategy: every_save
• hub_private_repo: None
• hub_always_push: False
• hub_revision: None
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• include_inputs_for_metrics: False
• include_for_metrics: []
• eval_do_concat_batches: True
• fp16_backend: auto
• push_to_hub_model_id: None
• push_to_hub_organization: None
• mp_parameters:
• auto_find_batch_size: False
• full_determinism: False
• torchdynamo: None
• ray_scope: last
• ddp_timeout: 1800
• torch_compile: False
• torch_compile_backend: None
• torch_compile_mode: None
• include_tokens_per_second: False
• include_num_input_tokens_seen: False
• neftune_noise_alpha: None
• optim_target_modules: None
• batch_eval_metrics: False
• eval_on_start: False
• use_liger_kernel: False
• liger_kernel_config: None
• eval_use_gather_object: False
• average_tokens_across_devices: False
• prompts: {'anchor': 'task: search result | query: ', 'positive': 'task: classification | query: '}
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional
• router_mapping: {}
• learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss	Validation Loss	german-search-cats-eval_cosine_ndcg@10	german-search-cats-test_cosine_ndcg@10
-1	-1	-	-	0.5624	-
0.0350	20	1.8489	-	-	-
0.0699	40	0.9438	-	-	-
-1	-1	-	-	0.8633	-
0.0350	20	0.6601	-	-	-
0.0699	40	0.4921	-	-	-
0.1049	60	0.6323	-	-	-
0.1399	80	0.8749	-	-	-
0.1748	100	0.755	0.6962	0.8705	-
0.2098	120	0.6572	-	-	-
0.2448	140	0.7488	-	-	-
0.2797	160	0.7288	-	-	-
0.3147	180	0.9539	-	-	-
0.3497	200	0.7782	0.6118	0.8888	-
0.3846	220	0.5965	-	-	-
0.4196	240	0.6789	-	-	-
0.4545	260	0.6537	-	-	-
0.4895	280	0.6572	-	-	-
0.5245	300	0.6221	0.5587	0.8966	-
0.5594	320	0.5079	-	-	-
0.5944	340	0.5949	-	-	-
0.6294	360	0.6828	-	-	-
0.6643	380	0.6628	-	-	-
0.6993	400	0.6654	0.5398	0.8986	-
0.7343	420	0.5939	-	-	-
0.7692	440	0.6976	-	-	-
0.8042	460	0.6453	-	-	-
0.8392	480	0.5204	-	-	-
0.8741	500	0.5709	0.5383	0.9035	-
0.9091	520	0.5452	-	-	-
0.9441	540	0.635	-	-	-
0.9790	560	0.8884	-	-	-
-1	-1	-	-	-	0.9023

Framework Versions

• Python: 3.10.18
• Sentence Transformers: 5.1.0
• Transformers: 4.56.1
• PyTorch: 2.8.0+cu128
• Accelerate: 1.10.1
• Datasets: 4.1.0
• Tokenizers: 0.22.0

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}
}