File size: 7,488 Bytes
405bce1 0700252 d841716 8f23f2d d841716 0700252 acdc403 0700252 acdc403 0700252 8f23f2d 0700252 8f23f2d 0700252 8f23f2d ff8cb47 8f23f2d 0700252 acdc403 0700252 070912c 3e2b3c7 9da2e57 070912c 9da2e57 3e2b3c7 9da2e57 070912c 3e2b3c7 070912c 9da2e57 070912c 9da2e57 070912c 3e2b3c7 070912c 9da2e57 3e2b3c7 070912c 3e2b3c7 070912c 0700252 8f23f2d 0700252 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 |
---
license: apache-2.0
---
# OmniFusion
**OmniFusion** is an advanced multimodal AI model designed to extend the capabilities of traditional language processing systems by integrating additional data modalities such as images, and potentially audio, 3D and video content.
### ChangeLog
[10/04/2024] OmniFusion-1.1 [weights](https://huggingface.co/AIRI-Institute/OmniFusion/tree/main/OmniMistral-v1_1) uploaded. The new model can speak Russian
[01/04/2024] Model training [source code](https://github.com/AIRI-Institute/OmniFusion/tree/main/OmniFusion/train_src) for OmniFusion-1.1 released
[22/11/2023] OmniFusion weights are available on [Huggingface](https://huggingface.co/AIRI-Institute/OmniFusion)
### Architecture
<p align="left">
<img src="https://raw.githubusercontent.com/AIRI-Institute/OmniFusion/main/content/architecture.png" width="100%">
</p>
OmniFusion open source version core is Mistral-7B. Initially focusing on images, we selected the CLIP-ViT-L as the visual encoder for its efficient information transfer capabilities. The most important component of OmniFusion is its adapter, a mechanism allowing the language model to interpret and incorporate information from different modalities. The adapter is a single-layer, four-headed transformer, which has shown superior performance compared to simpler linear layers or MLP structures.
This adapter takes embeddings from the visual encoder (excluding the CLS token) and maps them into textual embeddings compatible with the language model.
To further enhance the model's multimodal capabilities, we employ trainable special tokens to mark the beginning and end of visual data within the text sequence.
### Training Process consists of two stages
1. Pre-training the adapter on Image Captioning tasks (LAION, CC-4M).
2. Once the adapter has learned to map ViT's visual embeddings to the language model's textual space, we proceed to unfreeze Mistral for improved understanding of dialog formats and complex queries.
<p align="left">
<img src="https://raw.githubusercontent.com/AIRI-Institute/OmniFusion/main/content/datasets.png" width="70%">
</p>
### Results
OmniFusion-1.1 was benchmarked against the latest multimodal SOTA models. It excelled in generative metrics and classification benchmarks like Text-VQA.
<p align="left">
<img src="https://github.com/AIRI-Institute/OmniFusion/blob/main/content/radar_plot_gigachat.png" width="70%">
</p>
**OmniFusion-1.1** (Mistral version) results (April, 2024 update):
| Model | textvqa| scienceqa | pope | gqa | ok_vqa |
| -------------------------------------- | ------ | ---------- | --------- | -------- | ------- |
| OmniFusion-1.1 (one encoder, Mistral) | **0.4893** | **0.6802** | 0.7818 | 0.4600 | 0.5187 |
| OmniFusion-1.1 (two encoders, Mistral) | 0.4755 | 0.6732 | **0.8153** | **0.4761** | **0.5317** |
OmniFusion-1 (previous Mistral version) Performance on Visual Dialog Benchmark
| Model | NDCG | MRR | Recall@1 | Recall@5 | Recall@10 |
| ------------ | ---- | ---- | -------- | -------- | --------- |
| OmniFusion | 25.91| 10.78| 4.74 | 13.80 | 20.53 |
| LLaVA-13B | 24.74| 8.91 | 2.98 | 10.80 | 18.02 |
### Examples
<p align="left">
<img src="https://raw.githubusercontent.com/AIRI-Institute/OmniFusion/main/content/examples.png" width="100%">
</p>
### How to Use
```python
import torch
from PIL import Image
from transformers import AutoTokenizer, AutoModelForCausalLM
from urllib.request import urlopen
import torch.nn as nn
from huggingface_hub import hf_hub_download
# Loading some sources of the projection adapter and image encoder
hf_hub_download(repo_id="AIRI-Institute/OmniFusion", filename="models.py", local_dir='./')
from models import CLIPVisionTower
DEVICE = "cuda:0"
PROMPT = "This is a dialog with AI assistant.\n"
tokenizer = AutoTokenizer.from_pretrained("AIRI-Institute/OmniFusion", subfolder="OmniMistral-tokenizer", use_fast=False)
model = AutoModelForCausalLM.from_pretrained("AIRI-Institute/OmniFusion", subfolder="OmniMistral-model", torch_dtype=torch.bfloat16, device_map=DEVICE)
hf_hub_download(repo_id="AIRI-Institute/OmniFusion", filename="projection", local_dir='./')
hf_hub_download(repo_id="AIRI-Institute/OmniFusion", filename="special_embeddings.pt", local_dir='./')
projection = torch.load("projection", map_location=DEVICE)
special_embs = torch.load("special_embeddings.pt", map_location=DEVICE)
clip = CLIPVisionTower("openai/clip-vit-large-patch14-336")
clip.load_model()
clip = clip.to(device=DEVICE, dtype=torch.bfloat16)
def gen_answer(model, tokenizer, clip, projection, query, special_embs, image=None):
bad_words_ids = tokenizer(["\n", "</s>", ":"], add_special_tokens=False).input_ids + [[13]]
gen_params = {
"do_sample": False,
"max_new_tokens": 50,
"early_stopping": True,
"num_beams": 3,
"repetition_penalty": 1.0,
"remove_invalid_values": True,
"eos_token_id": 2,
"pad_token_id": 2,
"forced_eos_token_id": 2,
"use_cache": True,
"no_repeat_ngram_size": 4,
"bad_words_ids": bad_words_ids,
"num_return_sequences": 1,
}
with torch.no_grad():
image_features = clip.image_processor(image, return_tensors='pt')
image_embedding = clip(image_features['pixel_values']).to(device=DEVICE, dtype=torch.bfloat16)
projected_vision_embeddings = projection(image_embedding).to(device=DEVICE, dtype=torch.bfloat16)
prompt_ids = tokenizer.encode(f"{PROMPT}", add_special_tokens=False, return_tensors="pt").to(device=DEVICE)
question_ids = tokenizer.encode(query, add_special_tokens=False, return_tensors="pt").to(device=DEVICE)
prompt_embeddings = model.model.embed_tokens(prompt_ids).to(torch.bfloat16)
question_embeddings = model.model.embed_tokens(question_ids).to(torch.bfloat16)
embeddings = torch.cat(
[
prompt_embeddings,
special_embs['SOI'][None, None, ...],
projected_vision_embeddings,
special_embs['EOI'][None, None, ...],
special_embs['USER'][None, None, ...],
question_embeddings,
special_embs['BOT'][None, None, ...]
],
dim=1,
).to(dtype=torch.bfloat16, device=DEVICE)
out = model.generate(inputs_embeds=embeddings, **gen_params)
out = out[:, 1:]
generated_texts = tokenizer.batch_decode(out)[0]
return generated_texts
img_url = "https://i.pinimg.com/originals/32/c7/81/32c78115cb47fd4825e6907a83b7afff.jpg"
question = "who is the author?"
img = Image.open(urlopen(img_url))
answer = gen_answer(
model,
tokenizer,
clip,
projection,
query=question,
special_embs=special_embs,
image=img
)
img.show()
print(question)
print(answer)
```
### Future Plans
Work is underway on a version that uses ImageBind encoders and accepts more modalities (sound, 3D, video). Stay tuned for updates on GitHub!
### Authors
The FusionBrain scientific group from the AIRI Institute, in collaboration with scientists from Sber AI, led the model's development.
Main contributors:
+ Anton Razzhigaev: [Blog](https://t.me/abstractDL)
+ Elizaveta Goncharova
+ Matvey Mihkalchuk
+ Maxim Kurkin
+ Irina Abdullaeva
+ Denis Dimitrov [Blog](https://t.me/dendi_math_ai)
+ Andrey Kuznetsov [Blog](https://t.me/complete_ai) |