taesunwhang
commited on
Commit
•
d788394
1
Parent(s):
4c10cab
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,38 @@
|
|
1 |
-
---
|
2 |
-
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
base_model:
|
3 |
+
- lmsys/vicuna-7b-v1.5
|
4 |
+
- meta-math/MetaMath-Llemma-7B
|
5 |
+
library_name: transformers
|
6 |
+
tags:
|
7 |
+
- mergekit
|
8 |
+
- merge
|
9 |
+
license: llama2
|
10 |
+
---
|
11 |
+
# merge
|
12 |
+
|
13 |
+
This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
|
14 |
+
Model merge (slerp) based on [lmsys/vicuna-7b-v1.5](https://huggingface.co/lmsys/vicuna-7b-v1.5) and [meta-math/MetaMath-Llemma-7B](https://huggingface.co/meta-math/MetaMath-Llemma-7B)
|
15 |
+
|
16 |
+
1. Vicuna
|
17 |
+
|
18 |
+
## Model Details
|
19 |
+
|
20 |
+
Vicuna is a chat assistant trained by fine-tuning Llama 2 on user-shared conversations collected from ShareGPT.
|
21 |
+
|
22 |
+
- **Developed by:** [LMSYS](https://lmsys.org/)
|
23 |
+
- **Model type:** An auto-regressive language model based on the transformer architecture
|
24 |
+
- **License:** Llama 2 Community License Agreement
|
25 |
+
- **Finetuned from model:** [Llama 2](https://arxiv.org/abs/2307.09288)
|
26 |
+
|
27 |
+
### Model Sources
|
28 |
+
|
29 |
+
- **Repository:** https://github.com/lm-sys/FastChat
|
30 |
+
- **Blog:** https://lmsys.org/blog/2023-03-30-vicuna/
|
31 |
+
- **Paper:** https://arxiv.org/abs/2306.05685
|
32 |
+
- **Demo:** https://chat.lmsys.org/
|
33 |
+
|
34 |
+
2. MetaMath Llemma
|
35 |
+
|
36 |
+
## Model Details
|
37 |
+
|
38 |
+
MetaMath-Llemma-7B is fully fine-tuned on the MetaMathQA datasets and based on the powerful Llemma-7B model. It is glad to see using MetaMathQA datasets and change the base model from llama-2-7B to Llemma-7B can boost the MATH performance from 19.8 to **30.0**.
|