Jingjing Zhai
commited on
Commit
•
c0420f1
1
Parent(s):
d62653f
Update README
Browse files
README.md
CHANGED
@@ -4,7 +4,7 @@ license: apache-2.0
|
|
4 |
|
5 |
## Model Overview
|
6 |
|
7 |
-
PlantCaduceus is a DNA language model pre-trained on 16 Angiosperm genomes. Utilizing the Caduceus
|
8 |
|
9 |
- **PlantCaduceus_l20**: 20 layers, 384 hidden size, 20M parameters
|
10 |
- **PlantCaduceus_l24**: 24 layers, 512 hidden size, 40M parameters
|
@@ -14,7 +14,8 @@ PlantCaduceus is a DNA language model pre-trained on 16 Angiosperm genomes. Util
|
|
14 |
## How to use
|
15 |
```python
|
16 |
from transformers import AutoModel, AutoModelForMaskedLM, AutoTokenizer
|
17 |
-
|
|
|
18 |
device = "cuda:0" if torch.cuda.is_available() else "cpu"
|
19 |
model = AutoModelForMaskedLM.from_pretrained(model_path, trust_remote_code=True).to(device)
|
20 |
model.eval()
|
@@ -30,4 +31,21 @@ encoding = tokenizer.encode_plus(
|
|
30 |
input_ids = encoding["input_ids"].to(device)
|
31 |
with torch.inference_mode():
|
32 |
outputs = model(input_ids=input_ids, output_hidden_states=True)
|
33 |
-
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4 |
|
5 |
## Model Overview
|
6 |
|
7 |
+
PlantCaduceus is a DNA language model pre-trained on 16 Angiosperm genomes. Utilizing the [Caduceus](https://caduceus-dna.github.io/) and [Mamba](https://arxiv.org/abs/2312.00752) architectures and a masked language modeling objective, PlantCaduceus is designed to learn evolutionary conservation and DNA sequence grammar from 16 species spanning a history of 160 million years. We have trained a series of PlantCaduceus models with varying parameter sizes:
|
8 |
|
9 |
- **PlantCaduceus_l20**: 20 layers, 384 hidden size, 20M parameters
|
10 |
- **PlantCaduceus_l24**: 24 layers, 512 hidden size, 40M parameters
|
|
|
14 |
## How to use
|
15 |
```python
|
16 |
from transformers import AutoModel, AutoModelForMaskedLM, AutoTokenizer
|
17 |
+
import torch
|
18 |
+
model_path = 'kuleshov-group/PlantCaduceus_l24'
|
19 |
device = "cuda:0" if torch.cuda.is_available() else "cpu"
|
20 |
model = AutoModelForMaskedLM.from_pretrained(model_path, trust_remote_code=True).to(device)
|
21 |
model.eval()
|
|
|
31 |
input_ids = encoding["input_ids"].to(device)
|
32 |
with torch.inference_mode():
|
33 |
outputs = model(input_ids=input_ids, output_hidden_states=True)
|
34 |
+
```
|
35 |
+
|
36 |
+
## Citation
|
37 |
+
```bibtex
|
38 |
+
@article {Zhai2024.06.04.596709,
|
39 |
+
author = {Zhai, Jingjing and Gokaslan, Aaron and Schiff, Yair and Berthel, Ana and Liu, Zong-Yan and Miller, Zachary R and Scheben, Armin and Stitzer, Michelle C and Romay, Cinta and Buckler, Edward S. and Kuleshov, Volodymyr},
|
40 |
+
title = {Cross-species plant genomes modeling at single nucleotide resolution using a pre-trained DNA language model},
|
41 |
+
elocation-id = {2024.06.04.596709},
|
42 |
+
year = {2024},
|
43 |
+
doi = {10.1101/2024.06.04.596709},
|
44 |
+
URL = {https://www.biorxiv.org/content/early/2024/06/05/2024.06.04.596709},
|
45 |
+
eprint = {https://www.biorxiv.org/content/early/2024/06/05/2024.06.04.596709.full.pdf},
|
46 |
+
journal = {bioRxiv}
|
47 |
+
}
|
48 |
+
```
|
49 |
+
|
50 |
+
## Contact
|
51 |
+
Jingjing Zhai (jz963@cornell.edu)
|