Funnyworld1412/review_game_absa-aspect

SetFit Aspect Model

This is a SetFit model that can be used for Aspect Based Sentiment Analysis (ABSA). A LogisticRegression instance is used for classification. In particular, this model is in charge of filtering aspect span candidates.

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.

This model was trained within the context of a larger system for ABSA, which looks like so:

1. Use a spaCy model to select possible aspect span candidates.
2. Use this SetFit model to filter these possible aspect span candidates.
3. Use a SetFit model to classify the filtered aspect span candidates.

Model Details

Model Description

• Model Type: SetFit
• Classification head: a LogisticRegression instance
• spaCy Model: id_core_news_trf
• SetFitABSA Aspect Model: Funnyworld1412/review_game_absa-aspect
• SetFitABSA Polarity Model: Funnyworld1412/review_game_absa-polarity
• 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
aspect	'story:saranku developer harus menciptakan sebuah story yang sangat menarik agar tidak kehilangan para player karena masalahnya banyak player yg tidak bertahan lama karena repetitif dan monoton tiap update size makin gede doang yg isinya cuma chest baru itupun sampah puzzle yg makin lama makin rumit tapi chest nya sampah story kebanyakan npc teyvat story utama punya mc dilupain gak difokusin map kalo udah kosong ya nyampah bikin size gede doang main 3 tahun rasanya monoton perkembangan buruk' 'reward:tolong ditambah lagi reward untuk gachanya untuk player lama kesulitan mendapatkan primo karena sudah tidak ada lagi quest dan eksplorasi juga sudah 100 dasar developer kapitalis game ini makin lama makin monoton dan tidak ramah untuk player lama yang kekurangan bahan untuk gacha karakter' 'event:cuman saran jangan terlalu pelit biar para player gak kabur sama game sebelah hadiah event quest di perbaiki udah nunggu event lama lama hadiah cuman gitu gitu aja sampek event selesai primogemnya buat 10 pull gacha gak cukup tingakat kesulitan beda hadiah sama saja lama lama yang main pada kabur kalok terlalu pelit dan 1 lagi jariang mohon di perbaiki untuk server indonya trimaksih'
no aspect	'saranku developer:saranku developer harus menciptakan sebuah story yang sangat menarik agar tidak kehilangan para player karena masalahnya banyak player yg tidak bertahan lama karena repetitif dan monoton tiap update size makin gede doang yg isinya cuma chest baru itupun sampah puzzle yg makin lama makin rumit tapi chest nya sampah story kebanyakan npc teyvat story utama punya mc dilupain gak difokusin map kalo udah kosong ya nyampah bikin size gede doang main 3 tahun rasanya monoton perkembangan buruk' 'story:saranku developer harus menciptakan sebuah story yang sangat menarik agar tidak kehilangan para player karena masalahnya banyak player yg tidak bertahan lama karena repetitif dan monoton tiap update size makin gede doang yg isinya cuma chest baru itupun sampah puzzle yg makin lama makin rumit tapi chest nya sampah story kebanyakan npc teyvat story utama punya mc dilupain gak difokusin map kalo udah kosong ya nyampah bikin size gede doang main 3 tahun rasanya monoton perkembangan buruk' 'player:saranku developer harus menciptakan sebuah story yang sangat menarik agar tidak kehilangan para player karena masalahnya banyak player yg tidak bertahan lama karena repetitif dan monoton tiap update size makin gede doang yg isinya cuma chest baru itupun sampah puzzle yg makin lama makin rumit tapi chest nya sampah story kebanyakan npc teyvat story utama punya mc dilupain gak difokusin map kalo udah kosong ya nyampah bikin size gede doang main 3 tahun rasanya monoton perkembangan buruk'

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 AbsaModel

# Download from the 🤗 Hub
model = AbsaModel.from_pretrained(
    "Funnyworld1412/review_game_absa-aspect",
    "Funnyworld1412/review_game_absa-polarity",
)
# Run inference
preds = model("The food was great, but the venue is just way too busy.")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Word count	4	46.6389	94

Label	Training Sample Count
no aspect	4189
aspect	990

Training Hyperparameters

• batch_size: (4, 4)
• num_epochs: (1, 1)
• max_steps: -1
• sampling_strategy: oversampling
• num_iterations: 1
• 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
0.0004	1	0.4229	-
0.0193	50	0.3888	-
0.0386	100	0.268	-
0.0579	150	0.3151	-
0.0772	200	0.0962	-
0.0965	250	0.2717	-
0.1158	300	0.2986	-
0.1351	350	0.1456	-
0.1544	400	0.3291	-
0.1737	450	0.4705	-
0.1931	500	0.162	-
0.2124	550	0.227	-
0.2317	600	0.105	-
0.2510	650	0.0809	-
0.2703	700	0.0608	-
0.2896	750	0.0804	-
0.3089	800	0.5065	-
0.3282	850	0.1868	-
0.3475	900	0.2777	-
0.3668	950	0.0483	-
0.3861	1000	0.0174	-
0.4054	1050	0.0361	-
0.4247	1100	0.0208	-
0.4440	1150	0.1162	-
0.4633	1200	0.3258	-
0.4826	1250	0.4762	-
0.5019	1300	0.009	-
0.5212	1350	0.0445	-
0.5405	1400	0.4436	-
0.5598	1450	0.036	-
0.5792	1500	0.2706	-
0.5985	1550	0.2454	-
0.6178	1600	0.0539	-
0.6371	1650	0.2127	-
0.6564	1700	0.174	-
0.6757	1750	0.0915	-
0.6950	1800	0.3465	-
0.7143	1850	0.2593	-
0.7336	1900	0.205	-
0.7529	1950	0.2425	-
0.7722	2000	0.1797	-
0.7915	2050	0.0083	-
0.8108	2100	0.0973	-
0.8301	2150	0.1209	-
0.8494	2200	0.0049	-
0.8687	2250	0.0028	-
0.8880	2300	0.1165	-
0.9073	2350	0.046	-
0.9266	2400	0.2102	-
0.9459	2450	0.1639	-
0.9653	2500	0.0114	-
0.9846	2550	0.3658	-

Framework Versions

• Python: 3.10.13
• SetFit: 1.0.3
• Sentence Transformers: 3.0.1
• spaCy: 3.7.5
• Transformers: 4.36.2
• PyTorch: 2.1.2
• Datasets: 2.19.2
• Tokenizers: 0.15.2

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}
}