abhinav90parmar/bge-small-en-v1.5-erpguard

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 retrieval.

Model Details

Model Description

• —Model Type: Sentence Transformer
• —Base model: BAAI/bge-small-en-v1.5 <!-- at revision 5c38ec7c405ec4b44b94cc5a9bb96e735b38267a -->
• —Maximum Sequence Length: 512 tokens
• —Output Dimensionality: 384 dimensions
• —Similarity Function: Cosine Similarity
• —Supported Modality: Text <!-- - Training Dataset: Unknown --> <!-- - Language: Unknown --> <!-- - License: Unknown -->

Model Sources

• —Documentation: Sentence Transformers Documentation
• —Repository: Sentence Transformers on GitHub
• —Hugging Face: Sentence Transformers on Hugging Face

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'transformer_task': 'feature-extraction', 'modality_config': {'text': {'method': 'forward', 'method_output_name': 'last_hidden_state'}}, 'module_output_name': 'token_embeddings', 'architecture': 'BertModel'})
  (1): Pooling({'embedding_dimension': 384, 'pooling_mode': 'cls', '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
queries = [
    'Represent this sentence for searching relevant passages: The LC-5500 is producing bad cuts with heavy dross on the bottom, which component should I replace?',
]
documents = [
    '[maintenance_log] log_LC-5500_01 - 2025-04-22 | Corrective Maintenance | Technician A. Deshmukh\nUnscheduled intervention during Shift B following operator report of poor cut quality. Inspection of the cut components revealed heavy dross on underside of cut edge, matching failure mode LC5500-FM-01. Further examination of the cutting head indicated nozzle orifice degradation and wear, which disrupted the gas flow dynamics for the assist gas types (Nitrogen and Oxygen). Action taken: Removed damaged nozzle and replaced with spare part SP-7001. Verified nozzle standoff_mm maintained at 0.8 mm relative to the sheet surface. Performed test cut on 10 mm mild steel, verifying gas pressure against the material parameter table. Total downtime: 45 minutes. Parts consumed: SP-7001 (1 unit). Technician remarks: Cut edge quality restored to acceptable manufacturing standards. No further dross observed.',
    '[maintenance_log] log_LC-5500_03 - 2026-03-22 | Corrective Maintenance | Technician S. Deshmukh\nObserved condition: Shift B operators reported poor cut quality during processing of 12 mm structural carbon steel sheet, well within the max_sheet_thickness_mm rating of 20 mm. Inspection of cut components revealed severe dross accumulation on the underside of the cut edge, corresponding to failure mode LC5500-FM-01. Edge geometry degraded due to inconsistent assist gas flow and orifice degradation.\n\nAction taken: Powered down LC-5500 and locked out electrical supply. Removed inspection access cover on the cutting head assembly. Extracted worn cutting nozzle and cleaned seating surfaces with lint-free ethanol wipes. Installed replacement cutting nozzle part SP-7001. Performed nozzle centering routine and executed test cuts on reference material. Verified assist gas pressure alignment against standard cutting parameter table.\n\nParts consumed: SP-7001.\n\nDowntime in minutes: 45.\n\nTechnician remarks: Worn nozzle orifice caused turbulent gas jet stream, leading to incomplete slag ejection. Post-repair test cuts showed clean edges free of dross. Handed over machine to Shift C production supervisor without further anomalies.',
    '[maintenance_log] log_HP-450_01 - 2025-04-05 | Corrective Maintenance | Technician R. Kulkarni\nObserved condition: Shift A operator reported audible pump cavitation and vibration during initial cold startup of asset HP-450. Reservoir temperature was recorded at 21 degC. Hydraulic system parameter checks indicated pump starvation on the suction side, corresponding to failure mode HP450-FM-04.\n\nAction taken: Isolated system power and drained suction line. Inspected tank interior and found heavy contamination on the suction strainer. Removed and cleaned the suction strainer using solvent degreaser. Verified proper functioning of the reservoir immersion heater circuit. Replaced clogged line filter element with fresh SP-2004 stock. Refilled suction line, primed pump, and restarted system during Shift B.\n\nParts consumed: SP-2004 (1 ea)\n\nDowntime: 45 minutes\n\nTechnician remarks: Cavitation noise eliminated upon restart. Oil tank heater maintained standby oil temperature above minimum threshold. Pump operation verified smooth with static reservoir fluid level normal at 380 L capacity. Plant location: MPW-PUN-01.',
]
query_embeddings = model.encode_query(queries)
document_embeddings = model.encode_document(documents)
print(query_embeddings.shape, document_embeddings.shape)
# [1, 384] [3, 384]

# Get the similarity scores for the embeddings
similarities = model.similarity(query_embeddings, document_embeddings)
print(similarities)
# tensor([[0.7618, 0.7514, 0.2863]])

<!--

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

Information Retrieval

• —Dataset: val
• —Evaluated with <code>InformationRetrievalEvaluator</code>

<!--

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: 480 training samples
• —Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
• —Approximate statistics based on the first 100 samples: | | anchor | positive | negative | |:---------|:-----------------------------------------------------------------------------------|:--------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------| | type | string | string | string | | modality | text | text | text | | details | <ul><li>min: 23 tokens</li><li>mean: 32.75 tokens</li><li>max: 41 tokens</li></ul> | <ul><li>min: 164 tokens</li><li>mean: 221.46 tokens</li><li>max: 283 tokens</li></ul> | <ul><li>min: 54 tokens</li><li>mean: 205.46 tokens</li><li>max: 324 tokens</li></ul> |
• —Samples: | anchor | positive | negative | |:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| | <code>Represent this sentence for searching relevant passages: What procedures should technicians follow when replacing the SP-1001 on the CNC-XR200?</code> | <code>[sparepartscatalog] catalog01 - SP-1001 \| Turret Index Motor<br>Compatible Equipment: CNC-XR200. Unit Price: INR 84500. Supplier Lead Time: 21 days. Current Stock: 1. Reorder Level: 1. Supplier: Arcturus Spares India. <br><br>The Turret Index Motor is an electromechanical drive unit responsible for the precise rotational positioning of the tool turret on the CNC-XR200 machining center. It receives directional and step commands from the machine control unit to rotate the tool disc to the programmed tool position during automated tool change sequences. Failure of this motor results in tool change errors, turret indexing faults, and complete stoppage of machining operations. Maintenance technicians must verify encoder feedback, brake release function, and electrical connections during routine inspections. Replacement requires mechanical alignment of the coupling and parameter calibration within the CNC drive controller.</code> | <code>[manual] manualCNC-XR200 - Overview<br>This manual governs the operation, maintenance, and technical specifications for the XR-200 CNC Turning Center, designated under asset ID CNC-XR200. Manufactured by Arcturus Machine Tools, the machine was installed on 2021-03-14 on Line-1 Machining at the Meridian Precision Works facility (MPW-PUN-01), located in the Chakan Industrial Area, Pune. Operational schedules across the plant follow three continuous working periods: Shift A (06:00-14:00), Shift B (14:00-22:00), and Shift C (22:00-06:00). Line operators and maintenance personnel across all three shifts must adhere strictly to the procedures defined in this document to maintain operational efficiency and work quality. The XR-200 CNC Turning Center is designed for high-precision turning operations, requiring systematic tracking of operational parameters, regular maintenance intervals, and immediate intervention upon fault generation.</code> | | <code>Represent this sentence for searching relevant passages: The CNC-XR200 is throwing tool change errors due to a faulty SP-1001, what specific components need to be checked during inspection?</code> | <code>[sparepartscatalog] catalog01 - SP-1001 \| Turret Index Motor<br>Compatible Equipment: CNC-XR200. Unit Price: INR 84500. Supplier Lead Time: 21 days. Current Stock: 1. Reorder Level: 1. Supplier: Arcturus Spares India. <br><br>The Turret Index Motor is an electromechanical drive unit responsible for the precise rotational positioning of the tool turret on the CNC-XR200 machining center. It receives directional and step commands from the machine control unit to rotate the tool disc to the programmed tool position during automated tool change sequences. Failure of this motor results in tool change errors, turret indexing faults, and complete stoppage of machining operations. Maintenance technicians must verify encoder feedback, brake release function, and electrical connections during routine inspections. Replacement requires mechanical alignment of the coupling and parameter calibration within the CNC drive controller.</code> | <code>[maintenancelog] logIM-78003 - 2026-04-19 \| Corrective Maintenance \| Technician R. Kulkarni<br>Addressed mould stoppage during Shift A. Scheduled routine tool changeover completed with mould change time of 35 min, exactly matching reference parameters. Following tool installation, secondary hydraulic core pull failed to complete stroke, corresponding to failure mode IM780-FM-04. Checked pilot manifold pressure and traced fault to electrical actuation failure on directional control manifold. Replaced faulty coil using spare part core-pull solenoid SP-3003. Executed manual core-pull validation cycles. Total downtime recorded at 110 minutes. Technician remarks: Solenoid replacement restored full hydraulic actuation to auxiliary core circuits. Mould sequence verified through five consecutive automatic dry cycles.</code> | | <code>Represent this sentence for searching relevant passages: What maintenance tasks are required to prevent clamping alarms and seal degradation in part SP-1002?</code> | <code>[sparepartscatalog] catalog01 - SP-1002 \| Turret Clamp Assembly<br>Compatible Equipment: CNC-XR200. Unit Price: INR 32800. Supplier Lead Time: 14 days. Current Stock: 2. Reorder Level: 1. Supplier: Arcturus Spares India. <br><br>The Turret Clamp Assembly provides the mechanical locking force required to secure the tool turret disc in a rigid, stationary position during heavy cutting operations on the CNC-XR200. It prevents angular deflection and chatter when high torsional loads are applied to the cutting tools. The assembly consists of internal hydraulic or pneumatic pistons, Hirth couplings, and proximity sensors that confirm the locked or unlocked status of the turret. Operational failure or seal degradation inside this assembly leads to clamping alarms, loss of machining accuracy, and potential tool breakage. Regular checks should focus on hydraulic line pressure, actuation speed, and sensor alignment.</code> | <code>[manual] manualWR-3000 - Routine Maintenance Schedule<br>Maintenance tasks for the WR-3000 are divided into shift, weekly, and periodic intervals to prevent unexpected stoppages. Operators and maintenance personnel must execute checks at the start of Shifts A, B, and C. Inspect the welding torch nozzle for spatter accumulation, clean or replace contact tips at the designated contacttiplifeh interval, and verify gas diffuser condition. Check wire spool tension and feed roll integrity. Every 1000 hours of operation, grease all robot arm axes in accordance with the axisgreaseintervalh specification using approved industrial lubricant. Inspect internal cabling for wear, test emergency stop response times, and verify backup battery voltage within the controller cabinet annually. Log all completed service tasks in the plant maintenance management system.</code> |
• —Loss: <code>MultipleNegativesRankingLoss</code> with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim",
      "gather_across_devices": false,
      "directions": [
          "query_to_doc"
      ],
      "partition_mode": "joint",
      "hardness_mode": null,
      "hardness_strength": 0.0
  }

Training Hyperparameters

Non-Default Hyperparameters

• —per_device_train_batch_size: 32
• —num_train_epochs: 4
• —learning_rate: 2e-05
• —warmup_steps: 0.1
• —fp16: True
• —batch_sampler: no_duplicates

All Hyperparameters

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

• —per_device_train_batch_size: 32
• —num_train_epochs: 4
• —max_steps: -1
• —learning_rate: 2e-05
• —lr_scheduler_type: linear
• —lr_scheduler_kwargs: None
• —warmup_steps: 0.1
• —optim: adamwtorchfused
• —optim_args: None
• —weight_decay: 0.0
• —adam_beta1: 0.9
• —adam_beta2: 0.999
• —adam_epsilon: 1e-08
• —optim_target_modules: None
• —gradient_accumulation_steps: 1
• —average_tokens_across_devices: True
• —max_grad_norm: 1.0
• —label_smoothing_factor: 0.0
• —bf16: False
• —fp16: True
• —bf16_full_eval: False
• —fp16_full_eval: False
• —tf32: None
• —gradient_checkpointing: False
• —gradient_checkpointing_kwargs: None
• —torch_compile: False
• —torch_compile_backend: None
• —torch_compile_mode: None
• —use_liger_kernel: False
• —liger_kernel_config: None
• —use_cache: False
• —neftune_noise_alpha: None
• —torch_empty_cache_steps: None
• —auto_find_batch_size: False
• —log_on_each_node: True
• —logging_nan_inf_filter: True
• —include_num_input_tokens_seen: no
• —log_level: passive
• —log_level_replica: warning
• —disable_tqdm: False
• —project: huggingface
• —trackio_space_id: None
• —trackio_bucket_id: None
• —trackio_static_space_id: None
• —per_device_eval_batch_size: 8
• —prediction_loss_only: True
• —eval_on_start: False
• —eval_do_concat_batches: True
• —eval_use_gather_object: False
• —eval_accumulation_steps: None
• —include_for_metrics: []
• —batch_eval_metrics: False
• —save_only_model: False
• —save_on_each_node: False
• —enable_jit_checkpoint: False
• —push_to_hub: False
• —hub_private_repo: None
• —hub_model_id: None
• —hub_strategy: every_save
• —hub_always_push: False
• —hub_revision: None
• —load_best_model_at_end: False
• —ignore_data_skip: False
• —restore_callback_states_from_checkpoint: False
• —full_determinism: False
• —seed: 42
• —data_seed: None
• —use_cpu: False
• —accelerator_config: {'splitbatches': False, 'dispatchbatches': None, 'evenbatches': True, 'useseedablesampler': True, 'nonblocking': False, 'gradientaccumulationkwargs': None}
• —parallelism_config: None
• —dataloader_drop_last: False
• —dataloader_num_workers: 0
• —dataloader_pin_memory: True
• —dataloader_persistent_workers: False
• —dataloader_prefetch_factor: None
• —remove_unused_columns: True
• —label_names: None
• —train_sampling_strategy: random
• —length_column_name: length
• —ddp_find_unused_parameters: None
• —ddp_bucket_cap_mb: None
• —ddp_broadcast_buffers: False
• —ddp_static_graph: None
• —ddp_backend: None
• —ddp_timeout: 1800
• —fsdp: None
• —fsdp_config: None
• —deepspeed: None
• —debug: []
• —skip_memory_metrics: True
• —do_predict: False
• —resume_from_checkpoint: None
• —warmup_ratio: None
• —local_rank: -1
• —prompts: None
• —batch_sampler: no_duplicates
• —multi_dataset_batch_sampler: proportional
• —router_mapping: {}
• —learning_rate_mapping: {}

</details>

Training Logs

Training Time

• —Training: 4.1 minutes

Framework Versions

• —Python: 3.12.13
• —Sentence Transformers: 5.6.0
• —Transformers: 5.13.1
• —PyTorch: 2.11.0+cu128
• —Accelerate: 1.14.0
• —Datasets: 4.0.0
• —Tokenizers: 0.22.2

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{oord2019representationlearningcontrastivepredictive,
      title={Representation Learning with Contrastive Predictive Coding},
      author={Aaron van den Oord and Yazhe Li and Oriol Vinyals},
      year={2019},
      eprint={1807.03748},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/1807.03748},
}

<!--

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. -->