kasparas12/is_organizational_model

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

This is a SetFit model that can be used for Text Classification. This SetFit model uses BAAI/bge-small-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-small-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
ORGANIZATIONAL	'cryptonewton Shelby BitGet partner ' 'trezor Trezor Crypto security made easy' 'forbes Forbes Sign up now for Forbes free daily newsletter for unmatched insights and exclusive reporting '
INDIVIDUAL	'anbessa100 ANBESSA No paid service Never DM u' 'sbf_ftx SBF ' 'machibigbrother Machi Big Brother '

Evaluation

Metrics

Label	Accuracy
all	0.99

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("kasparas12/is_organizational_model")
# Run inference
preds = model("tradermayne Mayne ")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	3	15.7338	35

Label	Training Sample Count
INDIVIDUAL	423
ORGANIZATIONAL	377

Training Hyperparameters

• batch_size: (32, 32)
• num_epochs: (1, 1)
• max_steps: -1
• sampling_strategy: oversampling
• body_learning_rate: (2e-05, 1e-05)
• head_learning_rate: 0.01
• loss: CosineSimilarityLoss
• distance_metric: cosine_distance
• margin: 0.25
• end_to_end: False
• use_amp: False
• warmup_proportion: 0.1
• seed: 42
• eval_max_steps: -1
• load_best_model_at_end: False

Training Results

Epoch	Step	Training Loss	Validation Loss
0.0016	1	0.2511	-
0.0789	50	0.2505	-
0.1577	100	0.2225	-
0.2366	150	0.2103	-
0.3155	200	0.1383	-
0.3943	250	0.0329	-
0.4732	300	0.0098	-
0.5521	350	0.0034	-
0.6309	400	0.0019	-
0.7098	450	0.0015	-
0.7886	500	0.0014	-
0.8675	550	0.0012	-
0.0001	1	0.2524	-
0.0050	50	0.2115	-
0.0099	100	0.193	-
0.0001	1	0.2424	-
0.0050	50	0.2038	-
0.0099	100	0.1782	-
0.0001	1	0.2208	-
0.0050	50	0.1931	-
0.0099	100	0.1629	-
0.0149	150	0.2716	-
0.0199	200	0.18	-
0.0249	250	0.2504	-
0.0298	300	0.1936	-
0.0348	350	0.1764	-
0.0398	400	0.1817	-
0.0447	450	0.0624	-
0.0497	500	0.1183	-
0.0547	550	0.0793	-
0.0596	600	0.0281	-
0.0646	650	0.0876	-
0.0696	700	0.1701	-
0.0746	750	0.0468	-
0.0795	800	0.0525	-
0.0845	850	0.0783	-
0.0895	900	0.0342	-
0.0944	950	0.0158	-
0.0994	1000	0.0286	-
0.1044	1050	0.0016	-
0.1094	1100	0.0014	-
0.1143	1150	0.0298	-
0.1193	1200	0.018	-
0.1243	1250	0.0299	-
0.1292	1300	0.0019	-
0.1342	1350	0.0253	-
0.1392	1400	0.0009	-
0.1441	1450	0.0009	-
0.1491	1500	0.0011	-
0.1541	1550	0.0006	-
0.1591	1600	0.0006	-
0.1640	1650	0.0008	-
0.1690	1700	0.0005	-
0.1740	1750	0.0007	-
0.1789	1800	0.0006	-
0.1839	1850	0.0006	-
0.1889	1900	0.0006	-
0.1939	1950	0.0012	-
0.1988	2000	0.0004	-
0.2038	2050	0.0006	-
0.2088	2100	0.0005	-
0.2137	2150	0.0005	-
0.2187	2200	0.0005	-
0.2237	2250	0.0004	-
0.2287	2300	0.0005	-
0.2336	2350	0.0004	-
0.2386	2400	0.0004	-
0.2436	2450	0.0003	-
0.2485	2500	0.0004	-
0.2535	2550	0.0004	-
0.2585	2600	0.0004	-
0.2634	2650	0.0004	-
0.2684	2700	0.0004	-
0.2734	2750	0.0004	-
0.2784	2800	0.0056	-
0.2833	2850	0.0004	-
0.2883	2900	0.0003	-
0.2933	2950	0.0003	-
0.2982	3000	0.0004	-
0.3032	3050	0.0003	-
0.3082	3100	0.0003	-
0.3132	3150	0.0003	-
0.3181	3200	0.0003	-
0.3231	3250	0.0004	-
0.3281	3300	0.0003	-
0.3330	3350	0.0003	-
0.3380	3400	0.0003	-
0.3430	3450	0.0003	-
0.3479	3500	0.0003	-
0.3529	3550	0.0003	-
0.3579	3600	0.0003	-
0.3629	3650	0.0003	-
0.3678	3700	0.0003	-
0.3728	3750	0.0004	-
0.3778	3800	0.0004	-
0.3827	3850	0.0003	-
0.3877	3900	0.0003	-
0.3927	3950	0.0003	-
0.3977	4000	0.0003	-
0.4026	4050	0.0003	-
0.4076	4100	0.0003	-
0.4126	4150	0.0003	-
0.4175	4200	0.0003	-
0.4225	4250	0.0003	-
0.4275	4300	0.0003	-
0.4324	4350	0.0003	-
0.4374	4400	0.0002	-
0.4424	4450	0.0003	-
0.4474	4500	0.0003	-
0.4523	4550	0.0003	-
0.4573	4600	0.0003	-
0.4623	4650	0.0003	-
0.4672	4700	0.0002	-
0.4722	4750	0.0002	-
0.4772	4800	0.0003	-
0.4822	4850	0.0002	-
0.4871	4900	0.0002	-
0.4921	4950	0.0002	-
0.4971	5000	0.0003	-
0.5020	5050	0.0003	-
0.5070	5100	0.0002	-
0.5120	5150	0.0003	-
0.5169	5200	0.0002	-
0.5219	5250	0.0002	-
0.5269	5300	0.0002	-
0.5319	5350	0.0002	-
0.5368	5400	0.0003	-
0.5418	5450	0.0002	-
0.5468	5500	0.0002	-
0.5517	5550	0.0002	-
0.5567	5600	0.0002	-
0.5617	5650	0.0002	-
0.5667	5700	0.0002	-
0.5716	5750	0.0002	-
0.5766	5800	0.0002	-
0.5816	5850	0.0002	-
0.5865	5900	0.0002	-
0.5915	5950	0.0002	-
0.5965	6000	0.0002	-
0.6015	6050	0.0002	-
0.6064	6100	0.0002	-
0.6114	6150	0.0002	-
0.6164	6200	0.0002	-
0.6213	6250	0.0002	-
0.6263	6300	0.0002	-
0.6313	6350	0.0002	-
0.6362	6400	0.0002	-
0.6412	6450	0.0002	-
0.6462	6500	0.0002	-
0.6512	6550	0.0002	-
0.6561	6600	0.0002	-
0.6611	6650	0.0002	-
0.6661	6700	0.0002	-
0.6710	6750	0.0002	-
0.6760	6800	0.0002	-
0.6810	6850	0.0002	-
0.6860	6900	0.0002	-
0.6909	6950	0.0002	-
0.6959	7000	0.0002	-
0.7009	7050	0.0002	-
0.7058	7100	0.0002	-
0.7108	7150	0.0002	-
0.7158	7200	0.0002	-
0.7207	7250	0.0002	-
0.7257	7300	0.0002	-
0.7307	7350	0.0002	-
0.7357	7400	0.0002	-
0.7406	7450	0.0002	-
0.7456	7500	0.0002	-
0.7506	7550	0.0002	-
0.7555	7600	0.0002	-
0.7605	7650	0.0002	-
0.7655	7700	0.0248	-
0.7705	7750	0.0002	-
0.7754	7800	0.0002	-
0.7804	7850	0.0002	-
0.7854	7900	0.0002	-
0.7903	7950	0.0002	-
0.7953	8000	0.0002	-
0.8003	8050	0.0002	-
0.8052	8100	0.0002	-
0.8102	8150	0.0002	-
0.8152	8200	0.0002	-
0.8202	8250	0.0002	-
0.8251	8300	0.0002	-
0.8301	8350	0.0002	-
0.8351	8400	0.0002	-
0.8400	8450	0.0001	-
0.8450	8500	0.0002	-
0.8500	8550	0.0002	-
0.8550	8600	0.0001	-
0.8599	8650	0.0002	-
0.8649	8700	0.0002	-
0.8699	8750	0.0002	-
0.8748	8800	0.0002	-
0.8798	8850	0.0002	-
0.8848	8900	0.0002	-
0.8898	8950	0.0003	-
0.8947	9000	0.0002	-
0.8997	9050	0.0001	-
0.9047	9100	0.0002	-
0.9096	9150	0.0002	-
0.9146	9200	0.0002	-
0.9196	9250	0.0002	-
0.9245	9300	0.0002	-
0.9295	9350	0.0002	-
0.9345	9400	0.0002	-
0.9395	9450	0.0002	-
0.9444	9500	0.0002	-
0.9494	9550	0.0001	-
0.9544	9600	0.0001	-
0.9593	9650	0.0002	-
0.9643	9700	0.0002	-
0.9693	9750	0.0002	-
0.9743	9800	0.0001	-
0.9792	9850	0.0002	-
0.9842	9900	0.0002	-
0.9892	9950	0.0002	-
0.9941	10000	0.0002	-
0.9991	10050	0.0002	-

Framework Versions

• Python: 3.10.12
• SetFit: 1.0.3
• Sentence Transformers: 2.3.1
• Transformers: 4.35.2
• PyTorch: 2.1.0+cu121
• Datasets: 2.17.0
• Tokenizers: 0.15.1

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}
}