devilyouwei commited on
Commit
82c2aac
·
verified ·
1 Parent(s): 48a13b9

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +8 -0
README.md CHANGED
@@ -15,6 +15,8 @@ tags:
15
  - solidity
16
  - code-understanding
17
  library_name: transformers
 
 
18
  ---
19
 
20
  # SmartBERT V2 CodeBERT
@@ -37,6 +39,9 @@ SmartBERT V2 can be used for tasks such as:
37
  - Smart contract classification
38
  - Code embedding and retrieval
39
 
 
 
 
40
  ---
41
 
42
  ## Training Data
@@ -45,6 +50,9 @@ SmartBERT V2 was trained on a corpus of approximately **16,000 smart contracts**
45
 
46
  To better model smart contract behavior, contracts were processed at the **function level**, enabling the model to learn fine-grained semantic representations of smart contract functions.
47
 
 
 
 
48
  ---
49
 
50
  ## Preprocessing
 
15
  - solidity
16
  - code-understanding
17
  library_name: transformers
18
+ datasets:
19
+ - web3se/smart-contract-intent-vul-dataset
20
  ---
21
 
22
  # SmartBERT V2 CodeBERT
 
39
  - Smart contract classification
40
  - Code embedding and retrieval
41
 
42
+ SmartBERT V2 is a pre-trained model specifically developed for **[SmartIntent V2](https://github.com/web3se-lab/web3-sekit)**. It was trained on **16,000 smart contracts**, with no overlap with the SmartIntent V2 evaluation dataset to avoid data leakage.
43
+ For production use or general smart contract representation tasks, we recommend **SmartBERT V3**: https://huggingface.co/web3se/SmartBERT-v3
44
+
45
  ---
46
 
47
  ## Training Data
 
50
 
51
  To better model smart contract behavior, contracts were processed at the **function level**, enabling the model to learn fine-grained semantic representations of smart contract functions.
52
 
53
+ For benchmarking purposes in the **SmartIntent V2**, the pretraining corpus was intentionally limited to this **16,000-contract dataset**.
54
+ The **evaluation dataset (4,000 smart contracts)** was strictly held out and **not included in the pretraining data**, ensuring that downstream evaluations remain unbiased and free from data leakage.
55
+
56
  ---
57
 
58
  ## Preprocessing