Hugging Face's logo

genbio-ai
/
AIDO.DNA-300M

PyTorch
rnabert
Model card Files Community
1
DianLiI commited on
Commit
55e7080
·
verified ·
1 Parent(s): 96bce5f

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +18 -25
README.md CHANGED
@@ -1,52 +1,45 @@
1
- # Build any downstream models from this backbone
2
- ## Embedding
 
3
  ```python
4
  from genbio_finetune.tasks import Embed
5
-
6
- model = Embed.from_config({"model.backbone": "dna300m"})
7
-
8
- collated_batch = model.collate({"sequences": ["ACGT", "ACGT"]})
9
  embedding = model(collated_batch)
10
  print(embedding.shape)
11
  print(embedding)
12
  ```
13
- ## Sequence Level Classification
14
  ```python
15
  import torch
16
  from genbio_finetune.tasks import SequenceClassification
17
-
18
- model = SequenceClassification.from_config({"model.backbone": "dna300m", "model.n_classes": 2})
19
-
20
- collated_batch = model.collate({"sequences": ["ACGT", "ACGT"]})
21
  logits = model(collated_batch)
22
  print(logits)
23
  print(torch.argmax(logits, dim=-1))
24
  ```
25
- ## Token Level Classification
26
  ```python
27
  import torch
28
  from genbio_finetune.tasks import TokenClassification
29
-
30
- model = TokenClassification.from_config({"model.backbone": "dna300m", "model.n_classes": 3})
31
-
32
- collated_batch = model.collate({"sequences": ["ACGT", "ACGT"]})
33
  logits = model(collated_batch)
34
  print(logits)
35
  print(torch.argmax(logits, dim=-1))
36
  ```
37
- ## Regression
38
  ```python
39
  from genbio_finetune.tasks import SequenceRegression
40
-
41
- model = SequenceRegression.from_config({"model.backbone": "dna300m"})
42
-
43
- collated_batch = model.collate({"sequences": ["ACGT", "ACGT"]})
44
  logits = model(collated_batch)
45
  print(logits)
46
  ```
47
- ## Or use our one-liner CLI to finetune or evaluate any of the above!
48
  ```
49
- gbft fit --model SequenceClassification --model.backbone dna300m --data SequenceClassification --data.path <hf_or_local_path_to_your_dataset>
50
- gbft test --model SequenceClassification --model.backbone dna300m --data SequenceClassification --data.path <hf_or_local_path_to_your_dataset>
51
  ```
52
- For more information, visit: [Model Generator](https://github.com/genbio-ai/test)
 
1
+ ## How to Use
2
+ ### Build any downstream models from this backbone
3
+ #### Embedding
4
  ```python
5
  from genbio_finetune.tasks import Embed
6
+ model = Embed.from_config({"model.backbone": "dnafm"}).eval()
7
+ collated_batch = model.collate({"sequences": ["ACGT", "AGCT"]})
 
 
8
  embedding = model(collated_batch)
9
  print(embedding.shape)
10
  print(embedding)
11
  ```
12
+ #### Sequence Level Classification
13
  ```python
14
  import torch
15
  from genbio_finetune.tasks import SequenceClassification
16
+ model = SequenceClassification.from_config({"model.backbone": "dnafm", "model.n_classes": 2}).eval()
17
+ collated_batch = model.collate({"sequences": ["ACGT", "AGCT"]})
 
 
18
  logits = model(collated_batch)
19
  print(logits)
20
  print(torch.argmax(logits, dim=-1))
21
  ```
22
+ #### Token Level Classification
23
  ```python
24
  import torch
25
  from genbio_finetune.tasks import TokenClassification
26
+ model = TokenClassification.from_config({"model.backbone": "dnafm", "model.n_classes": 3}).eval()
27
+ collated_batch = model.collate({"sequences": ["ACGT", "AGCT"]})
 
 
28
  logits = model(collated_batch)
29
  print(logits)
30
  print(torch.argmax(logits, dim=-1))
31
  ```
32
+ #### Regression
33
  ```python
34
  from genbio_finetune.tasks import SequenceRegression
35
+ model = SequenceRegression.from_config({"model.backbone": "dnafm"}).eval()
36
+ collated_batch = model.collate({"sequences": ["ACGT", "AGCT"]})
 
 
37
  logits = model(collated_batch)
38
  print(logits)
39
  ```
40
+ #### Or use our one-liner CLI to finetune or evaluate any of the above!
41
  ```
42
+ gbft fit --model SequenceClassification --model.backbone dnafm --data SequenceClassification --data.path <hf_or_local_path_to_your_dataset>
43
+ gbft test --model SequenceClassification --model.backbone dnafm --data SequenceClassification --data.path <hf_or_local_path_to_your_dataset>
44
  ```
45
+ For more information, visit: [Model Generator](https://github.com/genbio-ai/modelgenerator)