Commit
·
622d3c8
1
Parent(s):
558face
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
datasets:
|
| 3 |
+
- bleugreen/typescript-instruct
|
| 4 |
+
language:
|
| 5 |
+
- en
|
| 6 |
+
tags:
|
| 7 |
+
- code
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
This model is a fune-tuned version of codet5-large on Typescript instruct-code pairs.
|
| 11 |
+
|
| 12 |
+
To run this model, you can use following example:
|
| 13 |
+
|
| 14 |
+
```
|
| 15 |
+
import torch
|
| 16 |
+
device = torch.device('cuda:0') if torch.cuda.is_available() else None
|
| 17 |
+
from transformers import AutoTokenizer, T5ForConditionalGeneration
|
| 18 |
+
|
| 19 |
+
def generate_code(task_description):
|
| 20 |
+
# Prepare the task description
|
| 21 |
+
input_ids = tokenizer.encode(task_description, return_tensors='pt').to(device)
|
| 22 |
+
|
| 23 |
+
# Generate the output
|
| 24 |
+
with torch.no_grad():
|
| 25 |
+
output_ids = model.generate(input_ids, max_length=200, temperature=0.7, num_beams=5)
|
| 26 |
+
|
| 27 |
+
# Decode the output
|
| 28 |
+
output = tokenizer.decode(output_ids[0], skip_special_tokens=True)
|
| 29 |
+
|
| 30 |
+
return output
|
| 31 |
+
|
| 32 |
+
model = T5ForConditionalGeneration.from_pretrained('mishasadhaker/codet5_large_typescript').to(device)
|
| 33 |
+
tokenizer = AutoTokenizer.from_pretrained('mishasadhaker/codet5_large_typescript')
|
| 34 |
+
|
| 35 |
+
print(generate_code('write function for sum of two numbers and return it'))
|
| 36 |
+
```
|