fyerfyer/finetune-jina-transformers-v1

SentenceTransformer based on jinaai/jina-embeddings-v2-base-en

This is a sentence-transformers model finetuned from jinaai/jina-embeddings-v2-base-en. 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: jinaai/jina-embeddings-v2-base-en
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity

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': 512, 'do_lower_case': False, 'architecture': 'JinaBertModel'})
  (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): 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("sentence_transformers_model_id")
# Run inference
sentences = [
    'How can I generate text with OLMo3 directly from the command line?',
    'Context: OLMo3\n\n# OLMo3\n\nOlmo3 is an improvement on OLMo2. More details will be released on *soon*.\n\n> !TIP\n> Click on the OLMo3 models in the right sidebar for more examples of how to apply OLMo3 to different language tasks.\n\nThe example below demonstrates how to generate text with `Pipeline`, `AutoModel` and from the command line.',
    'Context: Text generation\n\n# Text generation\n\n[open-in-colab]\n\nText generation is the most popular application for large language models (LLMs). A LLM is trained to generate the next word (token) given some initial text (prompt) along with its own generated outputs up to a predefined length or when it reaches an end-of-sequence (`EOS`) token.\n\nIn Transformers, the `~GenerationMixin.generate` API handles text generation, and it is available for all models with generative capabilities. This guide will show you the basics of text generation with `~GenerationMixin.generate` and some common pitfalls to avoid.\n\n> !TIP\n> You can also chat with a model directly from the command line. (reference(transformers-endocs/conversations#transformers))\n>\n> ```shell\n> transformers chat Qwen/Qwen2.5-0.5B-Instruct\n> ```',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.6064, 0.4927],
#         [0.6064, 1.0000, 0.3680],
#         [0.4927, 0.3680, 1.0000]])

Training Details

Training Dataset

Unnamed Dataset

• Size: 15,019 training samples
• Columns: sentence_0, sentence_1, and sentence_2
• Approximate statistics based on the first 1000 samples:

  	sentence_0	sentence_1	sentence_2
  type	string	string	string
  details	min: 12 tokens mean: 20.33 tokens max: 54 tokens	min: 7 tokens mean: 190.29 tokens max: 512 tokens	min: 9 tokens mean: 170.79 tokens max: 512 tokens

• Samples:

  sentence_0	sentence_1	sentence_2
  What does the documentation say about 'GPTBigCodeForCausalLM'?	Context: GPTBigCode > GPTBigCodeForCausalLM ## GPTBigCodeForCausalLM [autodoc] GPTBigCodeForCausalLM - forward	Context: Glossary > N > Natural language processing (NLP) ### Natural language processing (NLP) A generic way to say "deal with texts".
  How do I use or implement 'Local Self Attention' according to the provided text?	Context: Reformer > Usage tips > Local Self Attention ### Local Self Attention Local self attention is essentially a "normal" self attention layer with key, query and value projections, but is chunked so that in each chunk of length config.local_chunk_length the query embedding vectors only attends to the key embedding vectors in its chunk and to the key embedding vectors of config.local_num_chunks_before previous neighboring chunks and config.local_num_chunks_after following neighboring chunks. Using Local self attention, the memory and time complexity of the query-key matmul operation can be reduced from $\mathcal{O}(n_s \times n_s)$ to $\mathcal{O}(n_s \times \log(n_s))$, which usually represents the memory and time bottleneck in a transformer model, with $n_s$ being the sequence length.	Context: Contributing a new model to Transformers > Implementing a modular file > Attention ### Attention The modular Olmo2Attention is shown below. ```py from ..llama.modeling_llama import eager_attention_forward from ..olmo.modeling_olmo import OlmoAttention, apply_rotary_pos_emb # Olmo2 attention is identical to OLMo attention except: # - Norm is applied to attention queries and keys. # - No qkv clipping. class Olmo2Attention(OlmoAttention): def init(self, config: Olmo2Config, layer_idx: Optionalint = None): super().init(config, layer_idx=layer_idx) self.q_norm = Olmo2RMSNorm(config.num_attention_heads * self.head_dim, config.rms_norm_eps) self.k_norm = Olmo2RMSNorm(config.num_key_value_heads * self.head_dim, config.rms_norm_eps) def forward( self, hidden_states: torch.Tensor, position_embeddings: tupletorch.Tensor, torch.Tensor, attention_mask: Optionaltorch.Tensor, past_key_values: OptionalCache...
  Explain the 'model-specific feature extractor' section within 'Feature extractors > Feature extractor classes > AutoFeatureExtractor'.	Context: Feature extractors > Feature extractor classes > AutoFeatureExtractor > model-specific feature extractor #### model-specific feature extractor Every pretrained audio model has a specific associated feature extractor for correctly processing audio data. When you load a feature extractor, it retrieves the feature extractors configuration (feature size, chunk length, etc.) from preprocessor_config.json(https://hf.co/openai/whisper-tiny/blob/main/preprocessor_config.json). A feature extractor can be loaded directly from its model-specific class. py<br>from transformers import WhisperFeatureExtractor<br><br>feature_extractor = WhisperFeatureExtractor.from_pretrained("openai/whisper-tiny")<br>	Context: Feature extractors > Feature extractor classes > AutoFeatureExtractor #### AutoFeatureExtractor The AutoClass API automatically loads the correct feature extractor for a given model. Use ~AutoFeatureExtractor.from_pretrained to load a feature extractor. py<br>from transformers import AutoFeatureExtractor<br><br>feature_extractor = AutoFeatureExtractor.from_pretrained("openai/whisper-tiny")<br>

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• num_train_epochs: 1
• fp16: True
• multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: no
• prediction_loss_only: True
• per_device_train_batch_size: 8
• per_device_eval_batch_size: 8
• 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: 1
• 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
• 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
• project: huggingface
• trackio_space_id: trackio
• 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: no
• 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: True
• prompts: None
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: round_robin
• router_mapping: {}
• learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss
0.2662	500	0.9802
0.5325	1000	0.278
0.7987	1500	0.2026

Framework Versions

• Python: 3.12.12
• Sentence Transformers: 5.1.2
• Transformers: 4.57.1
• PyTorch: 2.9.0+cu126
• Accelerate: 1.12.0
• Datasets: 4.0.0
• Tokenizers: 0.22.1

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",
}

MultipleNegativesRankingLoss

@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}
}