magnifi/bge-small-en-v1.5-ft-orc-0930-dates

SentenceTransformer based on BAAI/bge-small-en-v1.5

This is a sentence-transformers model finetuned from BAAI/bge-small-en-v1.5. It maps sentences & paragraphs to a 384-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: BAAI/bge-small-en-v1.5
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 384 tokens
• 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': True}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 384, '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:

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 have I done in US equity [DATES]?',
    '[{"get_portfolio(None)": "portfolio"}, {"factor_contribution(\'portfolio\',\'<DATES>\',\'asset_class\',\'us equity\',\'returns\')": "portfolio"}]',
    '[{"get_portfolio(None)": "portfolio"}, {"factor_contribution(\'portfolio\',\'<DATES>\',\'asset_class\',\'us equity\',\'returns\')": "portfolio"}]',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

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

Evaluation

Metrics

Information Retrieval

• Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.6644
cosine_accuracy@3	0.8151
cosine_accuracy@5	0.8767
cosine_accuracy@10	0.9315
cosine_precision@1	0.6644
cosine_precision@3	0.2717
cosine_precision@5	0.1753
cosine_precision@10	0.0932
cosine_recall@1	0.0185
cosine_recall@3	0.0226
cosine_recall@5	0.0244
cosine_recall@10	0.0259
cosine_ndcg@10	0.1752
cosine_mrr@10	0.7525
cosine_map@100	0.021
dot_accuracy@1	0.6644
dot_accuracy@3	0.8151
dot_accuracy@5	0.8767
dot_accuracy@10	0.9315
dot_precision@1	0.6644
dot_precision@3	0.2717
dot_precision@5	0.1753
dot_precision@10	0.0932
dot_recall@1	0.0185
dot_recall@3	0.0226
dot_recall@5	0.0244
dot_recall@10	0.0259
dot_ndcg@10	0.1752
dot_mrr@10	0.7525
dot_map@100	0.021

Training Details

Training Dataset

Unnamed Dataset

• Size: 734 training samples
• Columns: sentence_0 and sentence_1
• Approximate statistics based on the first 1000 samples:

  	sentence_0	sentence_1
  type	string	string
  details	min: 5 tokens mean: 11.94 tokens max: 26 tokens	min: 24 tokens mean: 84.1 tokens max: 194 tokens

• Samples:

  sentence_0	sentence_1
  what is my portfolio [DATES] cagr?	[{"get_portfolio(None)": "portfolio"}, {"get_attribute('portfolio',['gains'],'')": "portfolio"}, {"sort('portfolio','gains','desc')": "portfolio"}]
  what is my [DATES] rate of return	[{"get_portfolio(None)": "portfolio"}, {"get_attribute('portfolio',['gains'],'')": "portfolio"}, {"sort('portfolio','gains','desc')": "portfolio"}]
  show backtest of my performance [DATES]?	[{"get_portfolio(None)": "portfolio"}, {"get_attribute('portfolio',['gains'],'')": "portfolio"}, {"sort('portfolio','gains','desc')": "portfolio"}]

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 10
• per_device_eval_batch_size: 10
• num_train_epochs: 6
• multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 10
• per_device_eval_batch_size: 10
• 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: 6
• 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: 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: False
• hub_always_push: False
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• include_inputs_for_metrics: False
• 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
• dispatch_batches: None
• split_batches: 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
• eval_use_gather_object: False
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: round_robin

Training Logs

Click to expand

Epoch	Step	cosine_map@100
0.0270	2	0.0136
0.0541	4	0.0138
0.0811	6	0.0140
0.1081	8	0.0142
0.1351	10	0.0144
0.1622	12	0.0146
0.1892	14	0.0147
0.2162	16	0.0150
0.2432	18	0.0152
0.2703	20	0.0157
0.2973	22	0.0165
0.3243	24	0.0168
0.3514	26	0.0167
0.3784	28	0.0170
0.4054	30	0.0174
0.4324	32	0.0180
0.4595	34	0.0181
0.4865	36	0.0181
0.5135	38	0.0182
0.5405	40	0.0182
0.5676	42	0.0182
0.5946	44	0.0183
0.6216	46	0.0183
0.6486	48	0.0183
0.6757	50	0.0183
0.7027	52	0.0182
0.7297	54	0.0185
0.7568	56	0.0186
0.7838	58	0.0189
0.8108	60	0.0190
0.8378	62	0.0191
0.8649	64	0.0193
0.8919	66	0.0197
0.9189	68	0.0198
0.9459	70	0.0196
0.9730	72	0.0196
1.0	74	0.0198
1.0270	76	0.0198
1.0541	78	0.0198
1.0811	80	0.0199
1.1081	82	0.0199
1.1351	84	0.0199
1.1622	86	0.0199
1.1892	88	0.0199
1.2162	90	0.0199
1.2432	92	0.0199
1.2703	94	0.0200
1.2973	96	0.0199
1.3243	98	0.0197
1.3514	100	0.0198
1.3784	102	0.0198
1.4054	104	0.0198
1.4324	106	0.0200
1.4595	108	0.0201
1.4865	110	0.0202
1.5135	112	0.0202
1.5405	114	0.0203
1.5676	116	0.0203
1.5946	118	0.0201
1.6216	120	0.0201
1.6486	122	0.0202
1.6757	124	0.0201
1.7027	126	0.0201
1.7297	128	0.0201
1.7568	130	0.0200
1.7838	132	0.0200
1.8108	134	0.0202
1.8378	136	0.0201
1.8649	138	0.0202
1.8919	140	0.0202
1.9189	142	0.0202
1.9459	144	0.0201
1.9730	146	0.0202
2.0	148	0.0202
2.0270	150	0.0204
2.0541	152	0.0204
2.0811	154	0.0203
2.1081	156	0.0203
2.1351	158	0.0204
2.1622	160	0.0204
2.1892	162	0.0202
2.2162	164	0.0202
2.2432	166	0.0201
2.2703	168	0.0202
2.2973	170	0.0202
2.3243	172	0.0202
2.3514	174	0.0202
2.3784	176	0.0202
2.4054	178	0.0202
2.4324	180	0.0203
2.4595	182	0.0203
2.4865	184	0.0203
2.5135	186	0.0204
2.5405	188	0.0204
2.5676	190	0.0203
2.5946	192	0.0203
2.6216	194	0.0203
2.6486	196	0.0202
2.6757	198	0.0202
2.7027	200	0.0202
2.7297	202	0.0202
2.7568	204	0.0201
2.7838	206	0.0201
2.8108	208	0.0201
2.8378	210	0.0201
2.8649	212	0.0202
2.8919	214	0.0202
2.9189	216	0.0203
2.9459	218	0.0203
2.9730	220	0.0204
3.0	222	0.0204
3.0270	224	0.0204
3.0541	226	0.0206
3.0811	228	0.0205
3.1081	230	0.0205
3.1351	232	0.0205
3.1622	234	0.0206
3.1892	236	0.0206
3.2162	238	0.0206
3.2432	240	0.0206
3.2703	242	0.0206
3.2973	244	0.0205
3.3243	246	0.0205
3.3514	248	0.0204
3.3784	250	0.0204
3.4054	252	0.0204
3.4324	254	0.0205
3.4595	256	0.0205
3.4865	258	0.0205
3.5135	260	0.0205
3.5405	262	0.0204
3.5676	264	0.0204
3.5946	266	0.0204
3.6216	268	0.0203
3.6486	270	0.0203
3.6757	272	0.0204
3.7027	274	0.0204
3.7297	276	0.0206
3.7568	278	0.0206
3.7838	280	0.0206
3.8108	282	0.0206
3.8378	284	0.0206
3.8649	286	0.0205
3.8919	288	0.0206
3.9189	290	0.0207
3.9459	292	0.0206
3.9730	294	0.0206
4.0	296	0.0207
4.0270	298	0.0207
4.0541	300	0.0207
4.0811	302	0.0208
4.1081	304	0.0208
4.1351	306	0.0207
4.1622	308	0.0207
4.1892	310	0.0207
4.2162	312	0.0208
4.2432	314	0.0208
4.2703	316	0.0208
4.2973	318	0.0208
4.3243	320	0.0208
4.3514	322	0.0208
4.3784	324	0.0208
4.4054	326	0.0208
4.4324	328	0.0207
4.4595	330	0.0207
4.4865	332	0.0207
4.5135	334	0.0207
4.5405	336	0.0207
4.5676	338	0.0207
4.5946	340	0.0207
4.6216	342	0.0208
4.6486	344	0.0208
4.6757	346	0.0208
4.7027	348	0.0208
4.7297	350	0.0208
4.7568	352	0.0209
4.7838	354	0.0209
4.8108	356	0.0210

Framework Versions

• Python: 3.10.9
• Sentence Transformers: 3.0.1
• Transformers: 4.44.0
• PyTorch: 2.4.0+cu121
• Accelerate: 0.33.0
• Datasets: 2.20.0
• Tokenizers: 0.19.1

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}
}