czczup commited on
Commit
ef3aa9f
1 Parent(s): 614f6b8

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +3 -0
README.md CHANGED
@@ -33,6 +33,7 @@ It is _**the largest open-source vision/vision-language foundation model (14B)**
33
  - Pretraining Stage
34
  - Learnable Component: InternViT-6B
35
  - Data: Trained on 72M samples, including COYO, LAION, CC12M, CC3M, SBU, Wukong, GRIT, Objects365, OpenImages, and OCR data.
 
36
  - SFT Stage
37
  - Learnable Component: MLP + LLM
38
  - Data: A comprehensive collection of open-source SFT datasets, along with their Chinese translation versions, totaling approximately 10M.
@@ -61,6 +62,8 @@ If you find this project useful in your research, please consider cite:
61
 
62
  This project is released under the MIT license. Parts of this project contain code and models (e.g., LLaMA2) from other sources, which are subject to their respective licenses.
63
 
 
 
64
  ## Acknowledgement
65
 
66
  InternVL is built with reference to the code of the following projects: [OpenAI CLIP](https://github.com/openai/CLIP), [Open CLIP](https://github.com/mlfoundations/open_clip), [CLIP Benchmark](https://github.com/LAION-AI/CLIP_benchmark), [EVA](https://github.com/baaivision/EVA/tree/master), [InternImage](https://github.com/OpenGVLab/InternImage), [ViT-Adapter](https://github.com/czczup/ViT-Adapter), [MMSegmentation](https://github.com/open-mmlab/mmsegmentation), [Transformers](https://github.com/huggingface/transformers), [DINOv2](https://github.com/facebookresearch/dinov2), [BLIP-2](https://github.com/salesforce/LAVIS/tree/main/projects/blip2), [Qwen-VL](https://github.com/QwenLM/Qwen-VL/tree/master/eval_mm), and [LLaVA-1.5](https://github.com/haotian-liu/LLaVA). Thanks for their awesome work!
 
33
  - Pretraining Stage
34
  - Learnable Component: InternViT-6B
35
  - Data: Trained on 72M samples, including COYO, LAION, CC12M, CC3M, SBU, Wukong, GRIT, Objects365, OpenImages, and OCR data.
36
+ - Note: In this stage, we load the pretrained weights of InternViT-6B-224px and interpolate its position embedding to the size corresponding to 448x448 pixels. Moreover, in order to reduce the number of visual tokens, we use a pixel shuffle to reduce 1024 tokens to 256 tokens.
37
  - SFT Stage
38
  - Learnable Component: MLP + LLM
39
  - Data: A comprehensive collection of open-source SFT datasets, along with their Chinese translation versions, totaling approximately 10M.
 
62
 
63
  This project is released under the MIT license. Parts of this project contain code and models (e.g., LLaMA2) from other sources, which are subject to their respective licenses.
64
 
65
+ Llama 2 is licensed under the LLAMA 2 Community License, Copyright (c) Meta Platforms, Inc. All Rights Reserved.
66
+
67
  ## Acknowledgement
68
 
69
  InternVL is built with reference to the code of the following projects: [OpenAI CLIP](https://github.com/openai/CLIP), [Open CLIP](https://github.com/mlfoundations/open_clip), [CLIP Benchmark](https://github.com/LAION-AI/CLIP_benchmark), [EVA](https://github.com/baaivision/EVA/tree/master), [InternImage](https://github.com/OpenGVLab/InternImage), [ViT-Adapter](https://github.com/czczup/ViT-Adapter), [MMSegmentation](https://github.com/open-mmlab/mmsegmentation), [Transformers](https://github.com/huggingface/transformers), [DINOv2](https://github.com/facebookresearch/dinov2), [BLIP-2](https://github.com/salesforce/LAVIS/tree/main/projects/blip2), [Qwen-VL](https://github.com/QwenLM/Qwen-VL/tree/master/eval_mm), and [LLaVA-1.5](https://github.com/haotian-liu/LLaVA). Thanks for their awesome work!