distilbert-no-mans-sky-sentiment-v2

DistilBERT – No Man’s Sky Sentiment (v2)

Fine-tuned distilbert-base-uncased on Steam reviews (2016–2024) for binary sentiment classification.

Training Details

• 70/30 train–test split
• Class-weighted CrossEntropyLoss to correct 78/22 imbalance
• 2 epochs, lr = 2e-5, batch size = 16
• Hardware: RTX 3070

Metrics

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
tokenizer = AutoTokenizer.from_pretrained("bkhang/distilbert-no-mans-sky-sentiment-v2")
model = AutoModelForSequenceClassification.from_pretrained("bkhang/distilbert-no-mans-sky-sentiment-v2")
text = "The update finally made the game worth playing!"
inputs = tokenizer(text, return_tensors="pt")
with torch.no_grad():
    logits = model(**inputs).logits
    prob = torch.softmax(logits, dim=1)[0,1].item()
print(prob)