weicaijaden commited on
Commit
cd005b4
·
verified ·
1 Parent(s): c912820

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +8 -10
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 = "JadenLong/MutBERT"
35
- # Optional: JadenLong/MutBERT-Huamn-Ref, JadenLong/MutBERT-Multi
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 = "JadenLong/MutBERT"
51
- # Optional: JadenLong/MutBERT-Huamn-Ref, JadenLong/MutBERT-Multi
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 = "JadenLong/MutBERT"
77
- # Optional: JadenLong/MutBERT-Huamn-Ref, JadenLong/MutBERT-Multi
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 = "JadenLong/MutBERT"
89
- # Optional: JadenLong/MutBERT-Huamn-Ref, JadenLong/MutBERT-Multi
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}