Downloads · 30 days
2
7% of all-time downloads
BERT-36L-CSL-Sepsis-Diag
fansen/BERT-36L-CSL-Sepsis-Diag
BERT-36L-CSL-Sepsis-Diag is a machine learning model from fansen. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Sepsis in-hospital mortality prediction from ICD diagnosis text.
All-time downloads
27
Public
Repo size
122 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.bin122 MB · 99%
At a glance
- Model type
- bert
- Access
- Public
- Created
- Jul 13, 2026
- Updated
- Jul 13, 2026
- SHA
- ae845ca0
Model card
From the Hugging Face model README
BERT-36L-CSL
Sepsis in-hospital mortality prediction from ICD diagnosis text.
- Mode: diag
- Architecture: BERT-style Transformer (encoder_num=6, 6 layers per branch)
- Total parameters: 21,997,698
- Vocabulary size: 4805
- Hidden size: 128
- Num attention heads: 4
- Feed-forward dim: 2048
- Max sequence length: 30
Loading the model
import torch
from src.bert_classifier import BertClassifier
# Load checkpoint
ckpt = torch.load('pytorch_model.bin', map_location='cpu')
model = BertClassifier(
vocab_size=4805,
encoder_num=6,
)
model.load_state_dict(ckpt['model_state_dict'])
model.eval()
# Inference
input_ids = tokenizer(text, return_tensors='pt')['input_ids']
logits = model(input_ids)
pred = torch.argmax(logits, dim=-1).item()
Built with sepsis-diagnosis.
- Type
- bert
- Created
- Jul 13, 2026
- Updated
- Jul 13, 2026