DIMI-embedding-v2 / README.md

Initial commit of fine-tuned GTE model on Arabic triplets

d4a6e4f verified 14 days ago

18.6 kB

	---
	tags:
	- sentence-transformers
	- sentence-similarity
	- feature-extraction
	- generated_from_trainer
	- dataset_size:498670
	- loss:MatryoshkaLoss
	- loss:MultipleNegativesRankingLoss
	base_model: Alibaba-NLP/gte-multilingual-base
	widget:
	- source_sentence: كم يبغ عدد السكان في المملكة المتحدة؟
	sentences:
	- هناك العديد من الناس الحاضرين.
	- كم عدد سكان أوكرانيا؟
	- لماذا باراك أوباما غير مؤهل للترشح في انتخابات الرئاسة لعام 2016؟
	- source_sentence: ماذا يجب أن أعرف عن ممارسة الأعمال التجارية في بلدك كرائد أعمال؟
	sentences:
	- إذا كان بإمكانك العيش في أي مكان في العالم لمدة عام، أين سيكون ذلك ولماذا؟
	- ماذا يجب أن أعطي صديقي في عيد الميلاد؟
	- ماذا يجب أن أعرف عن ممارسة الأعمال التجارية في بلدك؟
	- source_sentence: الرجل يرسم
	sentences:
	- رجل يستخدم الطلاء الرذاذ لرسم صورة على الحائط.
	- العرض مقرّر غداً.
	- مساء من الترفيه تحت النجوم هو أساسا جنوب كاليفورنيا.
	- source_sentence: لماذا لا يزال دونالد ترامب "يتجنب" قضية إقرار ضريبة الدخل؟
	sentences:
	- الحديقة لديها بوابة
	- لماذا لا يبدأ ترامب في قول "الحقيقة" عن طريق الإفصاح عن إقراراته الضريبية؟
	- كيف يمكنني التحقق من حسابي على إنستغرام مع علامة زرقاء؟
	- source_sentence: لا أعتقد ذلك
	sentences:
	- رجل واحد في قميص برتقالي يرتدي خوذة بيضاء يركب دراجة.
	- هناك أشخاص يأكلون في مطعم.
	- أخشى لا يا سيدي
	pipeline_tag: sentence-similarity
	library_name: sentence-transformers
	metrics:
	- pearson_cosine
	- spearman_cosine
	model-index:
	- name: SentenceTransformer based on Alibaba-NLP/gte-multilingual-base
	results:
	- task:
	type: semantic-similarity
	name: Semantic Similarity
	dataset:
	name: arabic sts17
	type: arabic-sts17
	metrics:
	- type: pearson_cosine
	value: 0.8112776989727821
	name: Pearson Cosine
	- type: spearman_cosine
	value: 0.8156442694344616
	name: Spearman Cosine
	---

	# SentenceTransformer based on Alibaba-NLP/gte-multilingual-base

	This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [Alibaba-NLP/gte-multilingual-base](https://huggingface.co/Alibaba-NLP/gte-multilingual-base). 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: [Alibaba-NLP/gte-multilingual-base](https://huggingface.co/Alibaba-NLP/gte-multilingual-base) <!-- at revision 9fdd4ee8bba0e2808a34e0e739576f6740d2b225 -->
	- Maximum Sequence Length: 8192 tokens
	- Output Dimensionality: 768 dimensions
	- Similarity Function: Cosine Similarity
	<!-- - Training Dataset: Unknown -->
	<!-- - Language: Unknown -->
	<!-- - License: Unknown -->

	### 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': 8192, 'do_lower_case': False}) with Transformer model: NewModel
	(1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, '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': False, '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("sentence_transformers_model_id")
	# Run inference
	sentences = [
	'لا أعتقد ذلك',
	'أخشى لا يا سيدي',
	'رجل واحد في قميص برتقالي يرتدي خوذة بيضاء يركب دراجة.',
	]
	embeddings = model.encode(sentences)
	print(embeddings.shape)
	# [3, 768]

	# Get the similarity scores for the embeddings
	similarities = model.similarity(embeddings, embeddings)
	print(similarities.shape)
	# [3, 3]
	```

	<!--
	### Direct Usage (Transformers)

	<details><summary>Click to see the direct usage in Transformers</summary>

	</details>
	-->

	<!--
	### Downstream Usage (Sentence Transformers)

	You can finetune this model on your own dataset.

	<details><summary>Click to expand</summary>

	</details>
	-->

	<!--
	### Out-of-Scope Use

	List how the model may foreseeably be misused and address what users ought not to do with the model.
	-->

	## Evaluation

	### Metrics

	#### Semantic Similarity

	* Dataset: `arabic-sts17`
	* Evaluated with [<code>EmbeddingSimilarityEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.EmbeddingSimilarityEvaluator)

	\| Metric \| Value \|
	\|:--------------------\|:-----------\|
	\| pearson_cosine \| 0.8113 \|
	\| spearman_cosine \| 0.8156 \|

	<!--
	## Bias, Risks and Limitations

	What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.
	-->

	<!--
	### Recommendations

	What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.
	-->

	## Training Details

	### Training Dataset

	#### Unnamed Dataset


	* Size: 498,670 training samples
	* Columns: <code>sentence_0</code> and <code>sentence_1</code>
	* Approximate statistics based on the first 1000 samples:
	\| \| sentence_0 \| sentence_1 \|
	\|:--------\|:----------------------------------------------------------------------------------\|:----------------------------------------------------------------------------------\|
	\| type \| string \| string \|
	\| details \| <ul><li>min: 4 tokens</li><li>mean: 19.59 tokens</li><li>max: 82 tokens</li></ul> \| <ul><li>min: 4 tokens</li><li>mean: 13.98 tokens</li><li>max: 69 tokens</li></ul> \|
	* Samples:
	\| sentence_0 \| sentence_1 \|
	\|:---------------------------------------------------\|:------------------------------------------------------------------------------------------------------------------------------\|
	\| <code>ولد صغير يرتدي ملابس زرقاء يرتدي حذاء</code> \| <code>الصبي الصغير يرتدي ملابسه</code> \|
	\| <code>كيف يتم بناء كاميرات المراقبة؟</code> \| <code>ما هي كاميرا المراقبة؟</code> \|
	\| <code>لماذا الطاقة الإجمالية للكون صفر؟</code> \| <code>إذا كان إجمالي الطاقة في الكون صفر، فهل يعني ذلك أن هناك طريقة لـ "صنع" المادة/الطاقة من خلال صنع نوع من النظير؟</code> \|
	* Loss: [<code>MatryoshkaLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#matryoshkaloss) with these parameters:
	```json
	{
	"loss": "MultipleNegativesRankingLoss",
	"matryoshka_dims": [
	768,
	384,
	128
	],
	"matryoshka_weights": [
	1,
	1,
	1
	],
	"n_dims_per_step": -1
	}
	```

	### Training Hyperparameters
	#### Non-Default Hyperparameters

	- `eval_strategy`: steps
	- `per_device_train_batch_size`: 24
	- `per_device_eval_batch_size`: 24
	- `fp16`: True
	- `multi_dataset_batch_sampler`: round_robin

	#### All Hyperparameters
	<details><summary>Click to expand</summary>

	- `overwrite_output_dir`: False
	- `do_predict`: False
	- `eval_strategy`: steps
	- `prediction_loss_only`: True
	- `per_device_train_batch_size`: 24
	- `per_device_eval_batch_size`: 24
	- `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`: 0.0
	- `adam_beta1`: 0.9
	- `adam_beta2`: 0.999
	- `adam_epsilon`: 1e-08
	- `max_grad_norm`: 1
	- `num_train_epochs`: 3
	- `max_steps`: -1
	- `lr_scheduler_type`: linear
	- `lr_scheduler_kwargs`: {}
	- `warmup_ratio`: 0.0
	- `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}
	- `tp_size`: 0
	- `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`: batch_sampler
	- `multi_dataset_batch_sampler`: round_robin

	</details>

	### Training Logs
	\| Epoch \| Step \| Training Loss \| arabic-sts17_spearman_cosine \|
	\|:------:\|:-----:\|:-------------:\|:----------------------------:\|
	\| 0.0481 \| 500 \| 1.6592 \| - \|
	\| 0.0963 \| 1000 \| 1.177 \| - \|
	\| 0.1444 \| 1500 \| 1.0053 \| - \|
	\| 0.1925 \| 2000 \| 0.9125 \| 0.8135 \|
	\| 0.2406 \| 2500 \| 0.8212 \| - \|
	\| 0.2888 \| 3000 \| 0.8204 \| - \|
	\| 0.3369 \| 3500 \| 0.7696 \| - \|
	\| 0.3850 \| 4000 \| 0.7501 \| 0.8089 \|
	\| 0.4332 \| 4500 \| 0.7118 \| - \|
	\| 0.4813 \| 5000 \| 0.7073 \| - \|
	\| 0.5294 \| 5500 \| 0.6772 \| - \|
	\| 0.5775 \| 6000 \| 0.6637 \| 0.8085 \|
	\| 0.6257 \| 6500 \| 0.6507 \| - \|
	\| 0.6738 \| 7000 \| 0.605 \| - \|
	\| 0.7219 \| 7500 \| 0.6076 \| - \|
	\| 0.7700 \| 8000 \| 0.6076 \| 0.8060 \|
	\| 0.8182 \| 8500 \| 0.5594 \| - \|
	\| 0.8663 \| 9000 \| 0.5928 \| - \|
	\| 0.9144 \| 9500 \| 0.5587 \| - \|
	\| 0.9626 \| 10000 \| 0.5736 \| 0.8099 \|
	\| 1.0 \| 10389 \| - \| 0.8122 \|
	\| 1.0107 \| 10500 \| 0.555 \| - \|
	\| 1.0588 \| 11000 \| 0.5233 \| - \|
	\| 1.1069 \| 11500 \| 0.5216 \| - \|
	\| 1.1551 \| 12000 \| 0.5176 \| 0.8015 \|
	\| 1.2032 \| 12500 \| 0.4865 \| - \|
	\| 1.2513 \| 13000 \| 0.4907 \| - \|
	\| 1.2995 \| 13500 \| 0.5079 \| - \|
	\| 1.3476 \| 14000 \| 0.4991 \| 0.8027 \|
	\| 1.3957 \| 14500 \| 0.4834 \| - \|
	\| 1.4438 \| 15000 \| 0.4626 \| - \|
	\| 1.4920 \| 15500 \| 0.4442 \| - \|
	\| 1.5401 \| 16000 \| 0.4768 \| 0.8079 \|
	\| 1.5882 \| 16500 \| 0.4459 \| - \|
	\| 1.6363 \| 17000 \| 0.4409 \| - \|
	\| 1.6845 \| 17500 \| 0.4434 \| - \|
	\| 1.7326 \| 18000 \| 0.4264 \| 0.8041 \|
	\| 1.7807 \| 18500 \| 0.4341 \| - \|
	\| 1.8289 \| 19000 \| 0.4143 \| - \|
	\| 1.8770 \| 19500 \| 0.4304 \| - \|
	\| 1.9251 \| 20000 \| 0.4314 \| 0.8133 \|
	\| 1.9732 \| 20500 \| 0.448 \| - \|
	\| 2.0 \| 20778 \| - \| 0.8116 \|
	\| 2.0214 \| 21000 \| 0.3985 \| - \|
	\| 2.0695 \| 21500 \| 0.3854 \| - \|
	\| 2.1176 \| 22000 \| 0.3875 \| 0.8095 \|
	\| 2.1658 \| 22500 \| 0.4139 \| - \|
	\| 2.2139 \| 23000 \| 0.3956 \| - \|
	\| 2.2620 \| 23500 \| 0.3856 \| - \|
	\| 2.3101 \| 24000 \| 0.3816 \| 0.8110 \|
	\| 2.3583 \| 24500 \| 0.3732 \| - \|
	\| 2.4064 \| 25000 \| 0.3662 \| - \|
	\| 2.4545 \| 25500 \| 0.3773 \| - \|
	\| 2.5026 \| 26000 \| 0.3703 \| 0.8058 \|
	\| 2.5508 \| 26500 \| 0.3666 \| - \|
	\| 2.5989 \| 27000 \| 0.369 \| - \|
	\| 2.6470 \| 27500 \| 0.3612 \| - \|
	\| 2.6952 \| 28000 \| 0.3444 \| 0.8135 \|
	\| 2.7433 \| 28500 \| 0.3667 \| - \|
	\| 2.7914 \| 29000 \| 0.3707 \| - \|
	\| 2.8395 \| 29500 \| 0.3698 \| - \|
	\| 2.8877 \| 30000 \| 0.3658 \| 0.8156 \|


	### Framework Versions
	- Python: 3.12.7
	- Sentence Transformers: 3.3.1
	- Transformers: 4.51.3
	- PyTorch: 2.6.0+cu124
	- Accelerate: 1.4.0
	- Datasets: 3.3.2
	- 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}
	}
	```

	#### MultipleNegativesRankingLoss
	```bibtex
	@misc{henderson2017efficient,
	title={Efficient Natural Language Response Suggestion for Smart Reply},
	author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
	year={2017},
	eprint={1705.00652},
	archivePrefix={arXiv},
	primaryClass={cs.CL}
	}
	```

	<!--
	## Glossary

	Clearly define terms in order to be accessible across audiences.
	-->

	<!--
	## Model Card Authors

	Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.
	-->

	<!--
	## Model Card Contact

	Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.
	-->