Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,52 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
inference: false
|
4 |
+
---
|
5 |
+
|
6 |
+
**NOTE: This GGML conversion is primarily for use with llama.cpp.**
|
7 |
+
- 13B parameters
|
8 |
+
- Based on version 1.1
|
9 |
+
- Used PR "More accurate Q4_0 and Q4_1 quantizations #896" (should be closer in quality to unquantized)
|
10 |
+
|
11 |
+
<br>
|
12 |
+
<br>
|
13 |
+
|
14 |
+
# Vicuna Model Card
|
15 |
+
|
16 |
+
## Model details
|
17 |
+
|
18 |
+
**Model type:**
|
19 |
+
Vicuna is an open-source chatbot trained by fine-tuning LLaMA on user-shared conversations collected from ShareGPT.
|
20 |
+
It is an auto-regressive language model, based on the transformer architecture.
|
21 |
+
|
22 |
+
**Model date:**
|
23 |
+
Vicuna was trained between March 2023 and April 2023.
|
24 |
+
|
25 |
+
**Organizations developing the model:**
|
26 |
+
The Vicuna team with members from UC Berkeley, CMU, Stanford, and UC San Diego.
|
27 |
+
|
28 |
+
**Paper or resources for more information:**
|
29 |
+
https://vicuna.lmsys.org/
|
30 |
+
|
31 |
+
**License:**
|
32 |
+
Apache License 2.0
|
33 |
+
|
34 |
+
**Where to send questions or comments about the model:**
|
35 |
+
https://github.com/lm-sys/FastChat/issues
|
36 |
+
|
37 |
+
## Intended use
|
38 |
+
**Primary intended uses:**
|
39 |
+
The primary use of Vicuna is research on large language models and chatbots.
|
40 |
+
|
41 |
+
**Primary intended users:**
|
42 |
+
The primary intended users of the model are researchers and hobbyists in natural language processing, machine learning, and artificial intelligence.
|
43 |
+
|
44 |
+
## Training dataset
|
45 |
+
70K conversations collected from ShareGPT.com.
|
46 |
+
|
47 |
+
## Evaluation dataset
|
48 |
+
A preliminary evaluation of the model quality is conducted by creating a set of 80 diverse questions and utilizing GPT-4 to judge the model outputs. See https://vicuna.lmsys.org/ for more details.
|
49 |
+
|
50 |
+
## Major updates of weights v1.1
|
51 |
+
- Refactor the tokenization and separator. In Vicuna v1.1, the separator has been changed from `"###"` to the EOS token `"</s>"`. This change makes it easier to determine the generation stop criteria and enables better compatibility with other libraries.
|
52 |
+
- Fix the supervised fine-tuning loss computation for better model quality.
|