Update README.md
Browse files
README.md
CHANGED
@@ -30,16 +30,10 @@ configs:
|
|
30 |
---
|
31 |
|
32 |
# Experiments for training Auto Regressive models for text-to-image generation
|
33 |
-
This dataset is derived from [conceptual captions](https://huggingface.co/datasets/pixparse/cc3m-wds) (CC3M) which contains roughly 3.3M image and caption pairs
|
34 |
-
|
35 |
-
For images we use [1d-tokenizer](https://github.com/bytedance/1d-tokenizer) by [bytedance](https://www.bytedance.com/en/) which tokenizes a 256 * 256 image into 32 tokens while still achieving SOTA fidelity ratio
|
36 |
-
|
37 |
-
For text we train a BPE based tokenizer on the image captions dataset with a vocab size set to 30K, where 4096 tokens where used to represent images, 9 to represent some special tokens and the remaining 25895 tokens for text
|
38 |
|
39 |
## Training Procedure
|
40 |
-
For training we prompt the model to generate an image based on a text such as: "a river has burst it 's banks and has spread out onto arable farmland alongside<|startofimage|><|image:2931|><|image:560|><|image:763|><|image:1539|><|image:3161|><|image:1997|><|image:3376|><|image:510|><|image:3036|><|image:1585|><|image:1853|><|image:1970|><|image:2687|><|image:1436|><|image:2213|><|image:3968|><|image:3999|><|image:877|><|image:725|><|image:3013|><|image:438|><|image:3159|><|image:2936|><|image:3003|><|image:2261|><|image:2137|><|image:3821|><|image:1513|><|image:3536|><|image:311|><|image:494|><|image:413|><|endofimage|>"
|
41 |
-
|
42 |
-
We use use cross entropy loss with logits masked for the audio tokens as it showed performance improvements for speech-to-text tasks and employ the standard cross entorpy loss over the masked logits
|
43 |
|
44 |
|
45 |
| Train Iter | hard rock artist performing music | football player during a match | concept vector illustration showing a flag | police officer and soldiers arrest military combatant | bird on a tree |
|
|
|
30 |
---
|
31 |
|
32 |
# Experiments for training Auto Regressive models for text-to-image generation
|
33 |
+
This dataset is derived from [conceptual captions](https://huggingface.co/datasets/pixparse/cc3m-wds) (CC3M) which contains roughly 3.3M image and caption pairs. For images we use [1d-tokenizer](https://github.com/bytedance/1d-tokenizer) by [bytedance](https://www.bytedance.com/en/) which tokenizes a 256 * 256 image into 32 tokens while still achieving SOTA fidelity ratio. For text we train a BPE based tokenizer on the image captions dataset with a vocab size set to 30K, where 4096 tokens where used to represent images, 9 to represent some special tokens and the remaining 25895 tokens for text
|
|
|
|
|
|
|
|
|
34 |
|
35 |
## Training Procedure
|
36 |
+
For training we prompt the model to generate an image based on a text such as: "a river has burst it 's banks and has spread out onto arable farmland alongside<|startofimage|><|image:2931|><|image:560|><|image:763|><|image:1539|><|image:3161|><|image:1997|><|image:3376|><|image:510|><|image:3036|><|image:1585|><|image:1853|><|image:1970|><|image:2687|><|image:1436|><|image:2213|><|image:3968|><|image:3999|><|image:877|><|image:725|><|image:3013|><|image:438|><|image:3159|><|image:2936|><|image:3003|><|image:2261|><|image:2137|><|image:3821|><|image:1513|><|image:3536|><|image:311|><|image:494|><|image:413|><|endofimage|>". We use use cross entropy loss with logits masked for the audio tokens as it showed performance improvements for speech-to-text tasks and employ the standard cross entorpy loss over the masked logits
|
|
|
|
|
37 |
|
38 |
|
39 |
| Train Iter | hard rock artist performing music | football player during a match | concept vector illustration showing a flag | police officer and soldiers arrest military combatant | bird on a tree |
|