Update README.md

README.md

@@ -13,7 +13,7 @@ datasets:
 ---
 # DeciDiffusion 2.0
 
-DeciDiffusion 2.0 is a 725 million parameter text-to-image latent diffusion model trained on the LAION-v2 dataset and fine-tuned on the LAION-ART dataset. Advanced training techniques were used to speed up training, improve training performance, and achieve better inference quality. 
+DeciDiffusion 2.0 is a 732 million parameter text-to-image latent diffusion model trained on the LAION-v2 dataset and fine-tuned on the LAION-ART dataset. Advanced training techniques were used to speed up training, improve training performance, and achieve better inference quality. 
 
 ## Model Details
 
@@ -24,8 +24,8 @@ DeciDiffusion 2.0 is a 725 million parameter text-to-image latent diffusion mode
 - **Weights License:** The weights are released under the [CreativeML Open RAIL++-M License](https://huggingface.co/Deci/DeciDiffusion-v1-0/blob/main/LICENSE-WEIGHTS.md)
 
 ### Model Sources
-- **Blog:** [A technical overview and comparison to Stable Diffusion 1.5](https://deci.ai/blog/decidiffusion-1-0-3x-faster-than-stable-diffusion-same-quality/)CHANGE
-- **Demo:** [Experience DeciDiffusion in action](https://huggingface.co/spaces/Deci/DeciDiffusion-v1-0)CHANGE
+- **Blog:** [A technical overview](https://deci.ai/blog/decidiffusion-2-0-text-to-image-generation-optimized-for-cost-effective-hardware/)
+- **Demo:** [Experience DeciDiffusion in action](https://huggingface.co/spaces/Deci/DeciDiffusion-v2-0)
 
 ## Model Architecture
 
@@ -63,26 +63,26 @@ The following techniques were used to shorten training time:
 
 ### Additional Details
 #### Phase 1
-- **Hardware:** 8 x 8 x A100 (80gb)
-- **Optimizer:** AdamW
-- **Batch:** 8192
-- **Learning rate:** 1e-4
+- **Hardware:** 6 x 8 x H100 (80GB)
+- **Optimizer:** LAMB
+- **Batch:** 8432
+- **Learning rate:** 5e-03
   
 #### Phases 2-4
-- **Hardware:** 8 x 8 x H100 (80gb)
+- **Hardware:** 8 x 8 x H100 (80GB)
 - **Optimizer:** LAMB
-- **Batch:** 6144
-- **Learning rate:** 5e-3
+- **Batch:** 7168
+- **Learning rate:** 5e-03
 
 ## Runtime Benchmarks
 
-The following tables provide an image latency comparison between DeciDiffusion 1.0 and Stable Diffusion v1.5. 
+The following tables provide an image latency comparison between DeciDiffusion 2.0 and Stable Diffusion v1.5. 
 
-DeciDiffusion 1.0 vs. Stable Diffusion v1.5 at FP16 precision
-|Implementation + Iterations| DeciDiffusion 1.0 on A10 (seconds/image) | Stable Diffusion v1.5 on A10 (seconds/image) |
+DeciDiffusion 2.0 vs. Stable Diffusion v1.5 at FP16 precision
+|Implementation + Iterations| DeciDiffusion 2.0 on AI 100 (seconds/image) | Stable Diffusion v1.5 on AI 100 (seconds/image) |
 |:----------|:----------|:----------|
-| PyTorch 16 Iterations  | 1.358 | 3.3216 |
-| PyTorch 10 Iterations  | 1.0059 |2.2459 |
+| Compiled 16 Iterations  | 1.358 | 3.3216 |
+| Compiled 10 Iterations  | 1.0059 |2.2459 |
 
 ## How to Use
 
@@ -136,7 +136,7 @@ The model has certain limitations and may not function optimally in the followin
 - The autoencoding component of the model is lossy.
 
 ### Bias
-The remarkable abilities of image generation models can unintentionally amplify societal biases. DeciDiffusion was mainly trained on subsets of LAION-v2, focused on English descriptions. Consequently, non-English communities and cultures might be underrepresented, leading to a bias towards white and western norms. Outputs from non-English prompts are notably less accurate. Given these biases, users should approach DeciDiffusion with discretion, regardless of input.
+The remarkable abilities of image-generation models can unintentionally amplify societal biases. DeciDiffusion was mainly trained on subsets of LAION-v2, focused on English descriptions. Consequently, non-English communities and cultures might be underrepresented, leading to a bias towards white and western norms. Outputs from non-English prompts are notably less accurate. Given these biases, users should approach DeciDiffusion with discretion, regardless of input.
 
 
 ## How to Cite
@@ -147,7 +147,7 @@ Please cite this model using this format.
 @misc{DeciFoundationModels,
 title = {DeciDiffusion 2.0},
 author = {DeciAI Research Team},
-year = {2023}
+year = {2024}
 url={[https://huggingface.co/deci/decidiffusion-v2-0](https://huggingface.co/deci/decidiffusion-v2-0)},
 }
 ```