Update README.md
Browse files
README.md
CHANGED
|
@@ -9,8 +9,6 @@ tags:
|
|
| 9 |
|
| 10 |
**This is repository for MutBERT (pretrained with mutation data in human genome)**.
|
| 11 |
|
| 12 |
-
**You can find all MutBERT variants at [here](https://huggingface.co/JadenLong).**
|
| 13 |
-
|
| 14 |
## Introduction
|
| 15 |
|
| 16 |
This is the official pre-trained model introduced in MutBERT: Probabilistic Genome Representation Improves Genomics Foundation Models.
|
|
@@ -31,8 +29,8 @@ MutBERT is a transformer-based genome foundation model trained only on Human gen
|
|
| 31 |
```python
|
| 32 |
from transformers import AutoTokenizer, AutoModel
|
| 33 |
|
| 34 |
-
model_name = "
|
| 35 |
-
# Optional:
|
| 36 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
| 37 |
model = AutoModel.from_pretrained(model_name, trust_remote_code=True)
|
| 38 |
```
|
|
@@ -47,8 +45,8 @@ import torch.nn.functional as F
|
|
| 47 |
|
| 48 |
from transformers import AutoTokenizer, AutoModel
|
| 49 |
|
| 50 |
-
model_name = "
|
| 51 |
-
# Optional:
|
| 52 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
| 53 |
model = AutoModel.from_pretrained(model_name, trust_remote_code=True)
|
| 54 |
|
|
@@ -73,8 +71,8 @@ print(embedding_max.shape) # expect to be 768
|
|
| 73 |
```python
|
| 74 |
from transformers import AutoModelForSequenceClassification
|
| 75 |
|
| 76 |
-
model_name = "
|
| 77 |
-
# Optional:
|
| 78 |
model = AutoModelForSequenceClassification.from_pretrained(model_name, trust_remote_code=True, num_labels=2)
|
| 79 |
```
|
| 80 |
|
|
@@ -85,8 +83,8 @@ Allowed types for RoPE scaling are: `linear` and `dynamic`. To extend the model'
|
|
| 85 |
If you want to scale your model context by 2x:
|
| 86 |
|
| 87 |
```python
|
| 88 |
-
model_name = "
|
| 89 |
-
# Optional:
|
| 90 |
model = AutoModel.from_pretrained(model_name,
|
| 91 |
trust_remote_code=True,
|
| 92 |
rope_scaling={'type': 'dynamic','factor': 2.0}
|
|
|
|
| 9 |
|
| 10 |
**This is repository for MutBERT (pretrained with mutation data in human genome)**.
|
| 11 |
|
|
|
|
|
|
|
| 12 |
## Introduction
|
| 13 |
|
| 14 |
This is the official pre-trained model introduced in MutBERT: Probabilistic Genome Representation Improves Genomics Foundation Models.
|
|
|
|
| 29 |
```python
|
| 30 |
from transformers import AutoTokenizer, AutoModel
|
| 31 |
|
| 32 |
+
model_name = "CompBioDSA/MutBERT"
|
| 33 |
+
# Optional: CompBioDSA/MutBERT-Huamn-Ref, CompBioDSA/MutBERT-Multi
|
| 34 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
| 35 |
model = AutoModel.from_pretrained(model_name, trust_remote_code=True)
|
| 36 |
```
|
|
|
|
| 45 |
|
| 46 |
from transformers import AutoTokenizer, AutoModel
|
| 47 |
|
| 48 |
+
model_name = "CompBioDSA/MutBERT"
|
| 49 |
+
# Optional: CompBioDSA/MutBERT-Huamn-Ref, CompBioDSA/MutBERT-Multi
|
| 50 |
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
| 51 |
model = AutoModel.from_pretrained(model_name, trust_remote_code=True)
|
| 52 |
|
|
|
|
| 71 |
```python
|
| 72 |
from transformers import AutoModelForSequenceClassification
|
| 73 |
|
| 74 |
+
model_name = "CompBioDSA/MutBERT"
|
| 75 |
+
# Optional: CompBioDSA/MutBERT-Huamn-Ref, CompBioDSA/MutBERT-Multi
|
| 76 |
model = AutoModelForSequenceClassification.from_pretrained(model_name, trust_remote_code=True, num_labels=2)
|
| 77 |
```
|
| 78 |
|
|
|
|
| 83 |
If you want to scale your model context by 2x:
|
| 84 |
|
| 85 |
```python
|
| 86 |
+
model_name = "CompBioDSA/MutBERT"
|
| 87 |
+
# Optional: CompBioDSA/MutBERT-Huamn-Ref, CompBioDSA/MutBERT-Multi
|
| 88 |
model = AutoModel.from_pretrained(model_name,
|
| 89 |
trust_remote_code=True,
|
| 90 |
rope_scaling={'type': 'dynamic','factor': 2.0}
|