yuyijiong commited on
Commit
03a79c2
1 Parent(s): 668cd2a

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +8 -8
README.md CHANGED
@@ -6,18 +6,18 @@ language:
6
  - en
7
  license: cc-by-sa-4.0
8
  ---
9
- * 2024.6.4 Update:
10
- 1. Add a slim version. The sample number is reduced from about 20k to 10k.
11
 
12
- * 2024.5.28 Update:
13
- 1. The data format is converted from "chatml" to "messages", which is more convenient to use ```tokenizer.apply_chat_template```.
14
- 2. And the version without "Original text paraphrasing" is added.
15
- 3. The old version has been moved to "legacy" branch.
16
 
 
 
 
17
  * [Paper](https://arxiv.org/abs/2312.11193)
18
  * [Github](https://github.com/yuyijiong/train_with_paraphrasing)
19
-
20
- # Long Context Instruction-tuning dataset with "Original text paraphrasing"
21
  * consist of multiple tasks
22
  * Chinese and English
23
  * sample length ranging from 4k to 32k
 
6
  - en
7
  license: cc-by-sa-4.0
8
  ---
9
+ ## 🔥 Updates
10
+ **\[2024.6.4\]** Add a slim version. The sample number is reduced from about 20k to 10k.
11
 
12
+ **\[2024.5.28\]**
13
+ 1. The data format is converted from "chatml" to "messages", which is more convenient to use ```tokenizer.apply_chat_template```. The old version has been moved to "legacy" branch.
14
+ 2. The version without "Original text paraphrasing" is added.
 
15
 
16
+
17
+
18
+ # 📊 Long Context Instruction-tuning dataset with "Original text paraphrasing"
19
  * [Paper](https://arxiv.org/abs/2312.11193)
20
  * [Github](https://github.com/yuyijiong/train_with_paraphrasing)
 
 
21
  * consist of multiple tasks
22
  * Chinese and English
23
  * sample length ranging from 4k to 32k