--- language: - tr license: apache-2.0 tags: - sentence-transformers - sentence-similarity - feature-extraction - generated_from_trainer - dataset_size:70941 - loss:MatryoshkaLoss - loss:CachedMultipleNegativesRankingLoss base_model: Qwen/Qwen3-Embedding-0.6B widget: - source_sentence: Bağımsız akıllı cihaz kampanyalarının detayları nelerdir? sentences: - Vodafone'un kampanyalarına katılan aboneler, seçtikleri tarifeye göre belirli indirimlerden yararlanabilirler. Örneğin, Cep Avantaj tarifeleri üzerinden 10 TL ile 20 TL arasında indirim sağlanmaktadır. - Kampanyalar, farklı cihaz modelleri için aylık ödeme planları sunmaktadır. - Vodafone'un kampanyaları, sadece internet paketleri ile ilgilidir. - source_sentence: İnternet hattımı nasıl iptal ettirebilirim? sentences: - Vodafone'da, müşterinin taşımak istediği numara yerine yanlışlıkla başka bir numaranın taşındığı durumlar, hatalı taşıma sürecini kapsamaktadır. - İnternet hattınızı iptal etmek için sadece online form doldurmanız yeterlidir. - İptal işlemi için müşteri hizmetlerini arayarak talepte bulunmanız ve iptal dilekçesini göndermeniz gerekmektedir. - source_sentence: Vodafone kampanyalarında veri kullanımı ve cezai şartlar sentences: - Yurtdışında geçerli olan tarifeler, yalnızca kurumsal müşterilere yöneliktir. - Vodafone kampanyaları, kullanıcıların istedikleri kadar veri kullanmalarına izin verir ve cezai şartlar uygulanmaz. - Vodafone'un kampanyalarında, kullanıcıların veri paketleri kullanımı belirli limitler dahilinde gerçekleşir ve kampanyadan yararlanma koşulları vardır. - source_sentence: Alcatel One Touch POP 7 Tablet'in işletim sistemi nedir? sentences: - Yabancılar için sunulan Limitsiz Fiber Kampanyası, belirli hızlarda internet paketleri sunmaktadır ve katılım için yabancı uyruklu olma şartı aranmaktadır. - Alcatel One Touch POP 7 Tablet, iOS işletim sistemi ile çalışan bir cihazdır. - Alcatel One Touch POP 7 Tablet, Android 4.2 işletim sistemi ile çalışmaktadır. - source_sentence: Vodafone Net'in internet hız garantisi var mı? sentences: - Ek data paketlerinin geçerlilik süreleri genellikle 30 gün olarak belirlenmiştir, ancak bazı paketler 7 gün geçerlilik süresine sahiptir. - Vodafone Net, tüm abonelerine en az 100 Mbps hız garantisi vermektedir. - Vodafone Net, internet hızını garanti etmemekte, bu hız abonenin hattının uygunluğuna ve santrale olan mesafeye bağlı olarak değişiklik göstermektedir. datasets: - seroe/vodex-turkish-triplets pipeline_tag: sentence-similarity library_name: sentence-transformers metrics: - cosine_accuracy model-index: - name: Qwen3-Embedding-0.6B Türkçe Triplet Matryoshka results: - task: type: triplet name: Triplet dataset: name: tr triplet dev 1024d type: tr-triplet-dev-1024d metrics: - type: cosine_accuracy value: 0.9672672152519226 name: Cosine Accuracy - type: cosine_accuracy value: 0.9776706695556641 name: Cosine Accuracy - task: type: triplet name: Triplet dataset: name: tr triplet dev 768d type: tr-triplet-dev-768d metrics: - type: cosine_accuracy value: 0.9690433740615845 name: Cosine Accuracy - type: cosine_accuracy value: 0.9776706695556641 name: Cosine Accuracy - task: type: triplet name: Triplet dataset: name: tr triplet dev 512d type: tr-triplet-dev-512d metrics: - type: cosine_accuracy value: 0.9718345403671265 name: Cosine Accuracy - type: cosine_accuracy value: 0.9781781435012817 name: Cosine Accuracy - task: type: triplet name: Triplet dataset: name: tr triplet dev 256d type: tr-triplet-dev-256d metrics: - type: cosine_accuracy value: 0.9687896370887756 name: Cosine Accuracy - type: cosine_accuracy value: 0.9771631360054016 name: Cosine Accuracy - task: type: triplet name: Triplet dataset: name: all nli test 1024d type: all-nli-test-1024d metrics: - type: cosine_accuracy value: 0.9764078855514526 name: Cosine Accuracy - task: type: triplet name: Triplet dataset: name: all nli test 768d type: all-nli-test-768d metrics: - type: cosine_accuracy value: 0.9759005308151245 name: Cosine Accuracy - task: type: triplet name: Triplet dataset: name: all nli test 512d type: all-nli-test-512d metrics: - type: cosine_accuracy value: 0.9748858213424683 name: Cosine Accuracy - task: type: triplet name: Triplet dataset: name: all nli test 256d type: all-nli-test-256d metrics: - type: cosine_accuracy value: 0.9756468534469604 name: Cosine Accuracy --- # Qwen3-Embedding-0.6B Türkçe Triplet Matryoshka This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [Qwen/Qwen3-Embedding-0.6B](https://huggingface.co/Qwen/Qwen3-Embedding-0.6B) on the [vodex-turkish-triplets](https://huggingface.co/datasets/seroe/vodex-turkish-triplets) dataset. It maps sentences & paragraphs to a 1024-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more. ## Model Details ## ⚠️ Domain-Specific Warning This model was fine-tuned on Turkish data specifically sourced from the **telecommunications domain**. While it performs well on telecom-related tasks such as mobile services, billing, campaigns, and subscription details, it may not generalize well to other domains. Please assess its performance carefully before applying it outside of telecommunications use cases. ### Model Description - **Model Type:** Sentence Transformer - **Base model:** [Qwen/Qwen3-Embedding-0.6B](https://huggingface.co/Qwen/Qwen3-Embedding-0.6B) - **Maximum Sequence Length:** 32768 tokens - **Output Dimensionality:** 1024 dimensions - **Similarity Function:** Cosine Similarity - **Training Dataset:** - [vodex-turkish-triplets](https://huggingface.co/datasets/seroe/vodex-turkish-triplets) - **Language:** tr - **License:** apache-2.0 ### Model Sources - **Documentation:** [Sentence Transformers Documentation](https://sbert.net) - **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) - **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers) ### Full Model Architecture ``` SentenceTransformer( (0): Transformer({'max_seq_length': 32768, 'do_lower_case': False}) with Transformer model: Qwen3Model (1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': True, 'include_prompt': True}) (2): Normalize() ) ``` ## Usage ### Direct Usage (Sentence Transformers) First install the Sentence Transformers library: ```bash pip install -U sentence-transformers ``` Then you can load this model and run inference. ```python from sentence_transformers import SentenceTransformer # Download from the 🤗 Hub model = SentenceTransformer("seroe/Qwen3-Embedding-0.6B-turkish-triplet-matryoshka") # Run inference sentences = [ "Vodafone Net'in internet hız garantisi var mı?", 'Vodafone Net, internet hızını garanti etmemekte, bu hız abonenin hattının uygunluğuna ve santrale olan mesafeye bağlı olarak değişiklik göstermektedir.', 'Vodafone Net, tüm abonelerine en az 100 Mbps hız garantisi vermektedir.', ] embeddings = model.encode(sentences) print(embeddings.shape) # [3, 1024] # Get the similarity scores for the embeddings similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [3, 3] ``` ## Evaluation ### Metrics #### Triplet * Datasets: `tr-triplet-dev-1024d` and `all-nli-test-1024d` * Evaluated with [TripletEvaluator](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) with these parameters: ```json { "truncate_dim": 1024 } ``` | Metric | tr-triplet-dev-1024d | all-nli-test-1024d | |:--------------------|:---------------------|:-------------------| | **cosine_accuracy** | **0.9673** | **0.9764** | #### Triplet * Datasets: `tr-triplet-dev-768d` and `all-nli-test-768d` * Evaluated with [TripletEvaluator](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) with these parameters: ```json { "truncate_dim": 768 } ``` | Metric | tr-triplet-dev-768d | all-nli-test-768d | |:--------------------|:--------------------|:------------------| | **cosine_accuracy** | **0.969** | **0.9759** | #### Triplet * Datasets: `tr-triplet-dev-512d` and `all-nli-test-512d` * Evaluated with [TripletEvaluator](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) with these parameters: ```json { "truncate_dim": 512 } ``` | Metric | tr-triplet-dev-512d | all-nli-test-512d | |:--------------------|:--------------------|:------------------| | **cosine_accuracy** | **0.9718** | **0.9749** | #### Triplet * Datasets: `tr-triplet-dev-256d` and `all-nli-test-256d` * Evaluated with [TripletEvaluator](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) with these parameters: ```json { "truncate_dim": 256 } ``` | Metric | tr-triplet-dev-256d | all-nli-test-256d | |:--------------------|:--------------------|:------------------| | **cosine_accuracy** | **0.9688** | **0.9756** | #### Triplet * Dataset: `tr-triplet-dev-1024d` * Evaluated with [TripletEvaluator](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) with these parameters: ```json { "truncate_dim": 1024 } ``` | Metric | Value | |:--------------------|:-----------| | **cosine_accuracy** | **0.9777** | #### Triplet * Dataset: `tr-triplet-dev-768d` * Evaluated with [TripletEvaluator](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) with these parameters: ```json { "truncate_dim": 768 } ``` | Metric | Value | |:--------------------|:-----------| | **cosine_accuracy** | **0.9777** | #### Triplet * Dataset: `tr-triplet-dev-512d` * Evaluated with [TripletEvaluator](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) with these parameters: ```json { "truncate_dim": 512 } ``` | Metric | Value | |:--------------------|:-----------| | **cosine_accuracy** | **0.9782** | #### Triplet * Dataset: `tr-triplet-dev-256d` * Evaluated with [TripletEvaluator](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.TripletEvaluator) with these parameters: ```json { "truncate_dim": 256 } ``` | Metric | Value | |:--------------------|:-----------| | **cosine_accuracy** | **0.9772** | ## Training Details ### Training Dataset #### vodex-turkish-triplets * Dataset: [vodex-turkish-triplets](https://huggingface.co/datasets/seroe/vodex-turkish-triplets) at [0c9fab0](https://huggingface.co/datasets/seroe/vodex-turkish-triplets/tree/0c9fab08a042b11b30064b5adc205f626c8a6add) * Size: 70,941 training samples * Columns: query, positive, and negative * Approximate statistics based on the first 1000 samples: | | query | positive | negative | |:--------|:----------------------------------------------------------------------------------|:------------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------| | type | string | string | string | | details | | | | * Samples: | query | positive | negative | |:-----------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:--------------------------------------------------------------------------------------------------------------------------------------------------------| | Kampanya tarihleri ve katılım şartları | Kampanya, 11 Ekim 2018'de başlayıp 29 Ekim 2018'de sona erecek. Katılımcıların belirli bilgileri doldurması ve Vodafone Müzik pass veya Video pass sahibi olmaları gerekiyor. | Kampanya, sadece İstanbul'daki kullanıcılar için geçerli olup, diğer şehirlerden katılım mümkün değildir. | | Taahhüt süresi dolmadan başka bir kampanyaya geçiş yapılırsa ne olur? | Eğer abone taahhüt süresi dolmadan başka bir kampanyaya geçerse, bu durumda önceki kampanya süresince sağlanan indirimler ve diğer faydalar, iptal tarihinden sonraki fatura ile tahsil edilecektir. | Aboneler, taahhüt süresi dolmadan başka bir kampanyaya geçtiklerinde, yeni kampanyadan faydalanmak için ek bir ücret ödemek zorundadırlar. | | FreeZone üyeliğimi nasıl sorgulayabilirim? | Üyeliğinizi sorgulamak için FREEZONESORGU yazarak 1525'e SMS gönderebilirsiniz. | Üyeliğinizi sorgulamak için Vodafone mağazasına gitmeniz gerekmektedir. | * Loss: [MatryoshkaLoss](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#matryoshkaloss) with these parameters: ```json { "loss": "CachedMultipleNegativesRankingLoss", "matryoshka_dims": [ 1024, 768, 512, 256 ], "matryoshka_weights": [ 1, 1, 1, 1 ], "n_dims_per_step": -1 } ``` ### Evaluation Dataset #### vodex-turkish-triplets * Dataset: [vodex-turkish-triplets](https://huggingface.co/datasets/seroe/vodex-turkish-triplets) at [0c9fab0](https://huggingface.co/datasets/seroe/vodex-turkish-triplets/tree/0c9fab08a042b11b30064b5adc205f626c8a6add) * Size: 3,941 evaluation samples * Columns: query, positive, and negative * Approximate statistics based on the first 1000 samples: | | query | positive | negative | |:--------|:----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------| | type | string | string | string | | details | | | | * Samples: | query | positive | negative | |:-----------------------------------------------------------------------|:----------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------| | Vodafone Net'e geçiş yaparken bağlantı ücreti var mı? | Vodafone Net'e geçişte 264 TL bağlantı ücreti bulunmaktadır ve bu ücret 24 ay boyunca aylık 11 TL olarak faturalandırılmaktadır. | Vodafone Net'e geçişte bağlantı ücreti yoktur ve tüm işlemler ücretsizdir. | | Bağımsız akıllı cihaz kampanyalarının detayları nelerdir? | Kampanyalar, farklı cihaz modelleri için aylık ödeme planları sunmaktadır. | Vodafone'un kampanyaları, sadece internet paketleri ile ilgilidir. | | Fibermax hizmeti iptal edilirse ne gibi sonuçlar doğar? | İptal işlemi taahhüt süresi bitmeden yapılırsa, indirimler ve ücretsiz hizmet bedelleri ödenmelidir. | Fibermax hizmeti iptal edildiğinde, kullanıcıdan hiçbir ücret talep edilmez. | * Loss: [MatryoshkaLoss](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#matryoshkaloss) with these parameters: ```json { "loss": "CachedMultipleNegativesRankingLoss", "matryoshka_dims": [ 1024, 768, 512, 256 ], "matryoshka_weights": [ 1, 1, 1, 1 ], "n_dims_per_step": -1 } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: steps - `per_device_train_batch_size`: 2048 - `per_device_eval_batch_size`: 256 - `weight_decay`: 0.01 - `num_train_epochs`: 2 - `lr_scheduler_type`: cosine - `warmup_ratio`: 0.05 - `save_only_model`: True - `bf16`: True - `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`: 2048 - `per_device_eval_batch_size`: 256 - `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`: 5e-05 - `weight_decay`: 0.01 - `adam_beta1`: 0.9 - `adam_beta2`: 0.999 - `adam_epsilon`: 1e-08 - `max_grad_norm`: 1.0 - `num_train_epochs`: 2 - `max_steps`: -1 - `lr_scheduler_type`: cosine - `lr_scheduler_kwargs`: {} - `warmup_ratio`: 0.05 - `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`: True - `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`: True - `fp16`: False - `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} - `deepspeed`: None - `label_smoothing_factor`: 0.0 - `optim`: adamw_torch - `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 - `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 - `eval_use_gather_object`: False - `average_tokens_across_devices`: False - `prompts`: None - `batch_sampler`: no_duplicates - `multi_dataset_batch_sampler`: proportional
### Training Logs | Epoch | Step | Training Loss | Validation Loss | tr-triplet-dev-1024d_cosine_accuracy | tr-triplet-dev-768d_cosine_accuracy | tr-triplet-dev-512d_cosine_accuracy | tr-triplet-dev-256d_cosine_accuracy | all-nli-test-1024d_cosine_accuracy | all-nli-test-768d_cosine_accuracy | all-nli-test-512d_cosine_accuracy | all-nli-test-256d_cosine_accuracy | |:------:|:----:|:-------------:|:---------------:|:------------------------------------:|:-----------------------------------:|:-----------------------------------:|:-----------------------------------:|:----------------------------------:|:---------------------------------:|:---------------------------------:|:---------------------------------:| | 0.3429 | 12 | 10.2876 | 3.2218 | 0.9145 | 0.9211 | 0.9262 | 0.9229 | - | - | - | - | | 0.6857 | 24 | 6.1342 | 2.5250 | 0.9531 | 0.9561 | 0.9571 | 0.9553 | - | - | - | - | | 0.3429 | 12 | 4.8969 | 2.3174 | 0.9597 | 0.9632 | 0.9617 | 0.9632 | - | - | - | - | | 0.6857 | 24 | 4.2031 | 2.0383 | 0.9673 | 0.9690 | 0.9718 | 0.9688 | - | - | - | - | | 0.3429 | 12 | 3.3893 | 2.3286 | 0.9650 | 0.9655 | 0.9652 | 0.9652 | - | - | - | - | | 0.6857 | 24 | 3.0878 | 2.1443 | 0.9728 | 0.9739 | 0.9749 | 0.9736 | - | - | - | - | | 1.0286 | 36 | 3.504 | 1.8128 | 0.9708 | 0.9716 | 0.9716 | 0.9723 | - | - | - | - | | 1.3714 | 48 | 2.4279 | 1.8915 | 0.9779 | 0.9774 | 0.9782 | 0.9777 | - | - | - | - | | 1.7143 | 60 | 2.2489 | 1.8638 | 0.9777 | 0.9777 | 0.9782 | 0.9772 | - | - | - | - | | -1 | -1 | - | - | - | - | - | - | 0.9764 | 0.9759 | 0.9749 | 0.9756 | ### Framework Versions - Python: 3.10.12 - Sentence Transformers: 4.2.0.dev0 - Transformers: 4.52.3 - PyTorch: 2.7.0+cu126 - Accelerate: 1.7.0 - Datasets: 3.6.0 - Tokenizers: 0.21.1 ## Citation ### BibTeX #### Sentence Transformers ```bibtex @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", } ``` #### MatryoshkaLoss ```bibtex @misc{kusupati2024matryoshka, title={Matryoshka Representation Learning}, author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi}, year={2024}, eprint={2205.13147}, archivePrefix={arXiv}, primaryClass={cs.LG} } ``` #### CachedMultipleNegativesRankingLoss ```bibtex @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} } ```