Commit
•
a3a74c7
1
Parent(s):
eb15d06
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- zh
|
| 5 |
+
- en
|
| 6 |
+
tags:
|
| 7 |
+
- code
|
| 8 |
+
---
|
| 9 |
+
# Chinese-CodeLlama-7B-PT
|
| 10 |
+
|
| 11 |
+
We have further expanded the vocabulary based on Chinese-LLaMA-2-7B which from 599296 to 75548, and we pre-trained the model based on LoRA, including `embed_tokens` and `lm_head` layers.
|
| 12 |
+
|
| 13 |
+
The training data contains approximately 400 million tokens, which from high-quality code dataset on HuggingFace.
|
| 14 |
+
|
| 15 |
+
In addition, we applied `memory_efficient_attention` to the pre-training, which saves us a lot of GPU memory space.
|
| 16 |
+
|
| 17 |
+
Our model can be used for SFT, and we hope to contribute more valuable work in the Chinese field.
|