liuhaotian
commited on
Commit
•
bcfd034
1
Parent(s):
bc1d421
Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
inference: false
|
3 |
+
license: apache-2.0
|
4 |
+
---
|
5 |
+
|
6 |
+
<br>
|
7 |
+
<br>
|
8 |
+
|
9 |
+
# LLaVA Model Card
|
10 |
+
|
11 |
+
## Model details
|
12 |
+
|
13 |
+
**Model type:**
|
14 |
+
LLaVA is an open-source chatbot trained by fine-tuning LLM on multimodal instruction-following data.
|
15 |
+
It is an auto-regressive language model, based on the transformer architecture.
|
16 |
+
Base LLM: [NousResearch/Nous-Hermes-2-Yi-34B](https://huggingface.co/NousResearch/Nous-Hermes-2-Yi-34B)
|
17 |
+
|
18 |
+
**Model date:**
|
19 |
+
LLaVA-v1.6-34B was trained in December 2023.
|
20 |
+
|
21 |
+
**Paper or resources for more information:**
|
22 |
+
https://llava-vl.github.io/
|
23 |
+
|
24 |
+
## License
|
25 |
+
[NousResearch/Nous-Hermes-2-Yi-34B](https://huggingface.co/NousResearch/Nous-Hermes-2-Yi-34B) license.
|
26 |
+
|
27 |
+
**Where to send questions or comments about the model:**
|
28 |
+
https://github.com/haotian-liu/LLaVA/issues
|
29 |
+
|
30 |
+
## Intended use
|
31 |
+
**Primary intended uses:**
|
32 |
+
The primary use of LLaVA is research on large multimodal models and chatbots.
|
33 |
+
|
34 |
+
**Primary intended users:**
|
35 |
+
The primary intended users of the model are researchers and hobbyists in computer vision, natural language processing, machine learning, and artificial intelligence.
|
36 |
+
|
37 |
+
## Training dataset
|
38 |
+
- 558K filtered image-text pairs from LAION/CC/SBU, captioned by BLIP.
|
39 |
+
- 158K GPT-generated multimodal instruction-following data.
|
40 |
+
- 500K academic-task-oriented VQA data mixture.
|
41 |
+
- 50K GPT-4V data mixture.
|
42 |
+
- 40K ShareGPT data.
|
43 |
+
|
44 |
+
## Evaluation dataset
|
45 |
+
A collection of 12 benchmarks, including 5 academic VQA benchmarks and 7 recent benchmarks specifically proposed for instruction-following LMMs.
|