TFS v0.2 | THINKING FAST AND SLOW

TFS is a classification model trained on over 5 million entries to classify the given sequences into:

<fast_thinking>
<slow_thinking>
<calculative_thinking>

from transformers import AutoModelForSequenceClassification, AutoTokenizer
import torch

model_id = "brahmairesearch/TFS-v0.2"
model = AutoModelForSequenceClassification.from_pretrained(model_id)
tokenizer = AutoTokenizer.from_pretrained(model_id)

def classify(sequence):
    inputs = tokenizer(sequence, return_tensors="pt")
    with torch.no_grad():
        logits = model(**inputs).logits
    predicted_class_id = logits.argmax().item()
    print(model.config.id2label[predicted_class_id])


classify("How many R are there in STRAWBERRY?")
# <calculative_thinking>

brahmairesearch
/

TFS-v0.2

TFS v0.2 | THINKING FAST AND SLOW

DO NOT OPENSOURCE ### | BRAHMAI