Netta1994/setfit_baai_cybereason_gpt-4o_cot-few_shot_only_reasoning_1726751740.333197

SetFit with BAAI/bge-base-en-v1.5

This is a SetFit model that can be used for Text Classification. This SetFit model uses BAAI/bge-base-en-v1.5 as the Sentence Transformer embedding model. A LogisticRegression instance is used for classification.

The model has been trained using an efficient few-shot learning technique that involves:

1. Fine-tuning a Sentence Transformer with contrastive learning.
2. Training a classification head with features from the fine-tuned Sentence Transformer.

Model Details

Model Description

• Model Type: SetFit
• Sentence Transformer body: BAAI/bge-base-en-v1.5
• Classification head: a LogisticRegression instance
• Maximum Sequence Length: 512 tokens
• Number of Classes: 2 classes

Model Sources

• Repository: SetFit on GitHub
• Paper: Efficient Few-Shot Learning Without Prompts
• Blogpost: SetFit: Efficient Few-Shot Learning Without Prompts

Model Labels

Label	Examples
0	"The answer provided is vague and does not directly address the question asked. The question specifically seeks to understand the significance of considering all the answers together when determining if the behavior in a MalOp is malicious. The document suggests that the combined answers help in determining whether the behavior requires remediation, identifying vital machines, addressing severe activities, and determining the significance of users involved. This nuanced consideration is critical to accurately identifying and prioritizing threats. The answer lacks these specific references and explanations and fails to provide a clear connection to the document's content.\n\nFinal evaluation: Bad" 'Evaluation:\nThe answer given by the user does not directly respond to the question. The provided document details the steps involved in excluding a MalOp during the remediation phase, but the user’s answer states that the information is insufficient and suggests seeking additional sources. This is incorrect, as the document contains the necessary steps to answer the question.\n\nFinal evaluation: Bad' 'The answer provided is accurate and correctly grounded in the document. The query asks what should be done if a file is quarantined, and the given response clearly states that the file should be un-quarantined before submitting it to the relevant organization. This matches the specific instruction found in the article, fulfilling the requirement of the evaluation.\n\nFinal Verdict: Good'
1	"Evaluation:\n\nThe given answer is accurate and correctly reflects the content of the document. It specifies what the computer will generate (a dump file containing the entire contents of the sensor's RAM) in the event of a system failure, which aligns with the information provided in the document.\n\nFinal evaluation: Good" 'Evaluation:\nThe answer directly addresses the question by stating that the purpose of the platform's threat detection abilities is "to identify cyber security threats," which aligns with the information provided in the document. The document elaborates on the capabilities of the platform's threat detection, including the identification of cyber security threats using various methods like artificial intelligence, machine learning, and behavioral analysis.\n\nThe final evaluation: Good' 'The answer provided is unhelpful and does not directly address the question of identifying the severity score for the fifth scenario. The document briefing detailed examples of scoring for different scenarios, but the response fails to utilize that information. Instead, it defers to seeking additional sources or context, which is unnecessary here.\n\nFinal evaluation: Bad'

Evaluation

Metrics

Label	Accuracy
all	0.5915

Uses

Direct Use for Inference

First install the SetFit library:

pip install setfit

Then you can load this model and run inference.

from setfit import SetFitModel

# Download from the 🤗 Hub
model = SetFitModel.from_pretrained("Netta1994/setfit_baai_cybereason_gpt-4o_cot-few_shot_only_reasoning_1726751740.333197")
# Run inference
preds = model("The given answer is \"..\/..\/_images\/hunting_http://www.flores.net/\". However, the document clearly outlines the URLs for image examples relating to the queries. For the second query, the URL provided in the document is ..\/..\/_images\/hunting_http://miller.co. The answer provided does not match the correct URL from the document.

Final evaluation: Bad")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	20	58.5072	183

Label	Training Sample Count
0	34
1	35

Training Hyperparameters

• batch_size: (16, 16)
• num_epochs: (5, 5)
• max_steps: -1
• sampling_strategy: oversampling
• num_iterations: 20
• body_learning_rate: (2e-05, 2e-05)
• head_learning_rate: 2e-05
• loss: CosineSimilarityLoss
• distance_metric: cosine_distance
• margin: 0.25
• end_to_end: False
• use_amp: False
• warmup_proportion: 0.1
• l2_weight: 0.01
• seed: 42
• eval_max_steps: -1
• load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0058	1	0.2075	-
0.2890	50	0.2549	-
0.5780	100	0.2371	-
0.8671	150	0.084	-
1.1561	200	0.0034	-
1.4451	250	0.0021	-
1.7341	300	0.0019	-
2.0231	350	0.0016	-
2.3121	400	0.0016	-
2.6012	450	0.0014	-
2.8902	500	0.0013	-
3.1792	550	0.0013	-
3.4682	600	0.0013	-
3.7572	650	0.0013	-
4.0462	700	0.0013	-
4.3353	750	0.0012	-
4.6243	800	0.0012	-
4.9133	850	0.0012	-

Framework Versions

• Python: 3.10.14
• SetFit: 1.1.0
• Sentence Transformers: 3.1.0
• Transformers: 4.44.0
• PyTorch: 2.4.1+cu121
• Datasets: 2.19.2
• Tokenizers: 0.19.1

Citation

BibTeX

@article{https://doi.org/10.48550/arxiv.2209.11055,
    doi = {10.48550/ARXIV.2209.11055},
    url = {https://arxiv.org/abs/2209.11055},
    author = {Tunstall, Lewis and Reimers, Nils and Jo, Unso Eun Seo and Bates, Luke and Korat, Daniel and Wasserblat, Moshe and Pereg, Oren},
    keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},
    title = {Efficient Few-Shot Learning Without Prompts},
    publisher = {arXiv},
    year = {2022},
    copyright = {Creative Commons Attribution 4.0 International}
}