Update README.md

README.md

@@ -51,14 +51,15 @@ This model is based on OpenAI's GPT2 model that has been specifically fine-tuned
 <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
 
 
-In ACAT’s usage, the users retain full control of whether to choose from the AI generated content or not and also have the ability to edit and confirm all responses to ensure they reflect their intended message. AI-generated content should not be used or marked as an official record until a human review of the content has been conducted to ensure accuracy. It is users’ responsibility to edit for tone was well as facts. ​​Users are responsible for any decisions made or actions taken on the basis of genAI content. 
+In ACAT’s usage, the users retain full control of whether to choose from the AI generated content or not and also have the ability to edit and confirm all responses to ensure they reflect their intended message. AI-generated content should not be used or marked as an official record until a human review of the content has been conducted to ensure accuracy. It is users’ responsibility to edit for tone was well as facts. 
+
+Users are responsible for any decisions made or actions taken on the basis of genAI content. 
 
 
 # Training Details
 
 ## Training Data
 
-<!-- This should link to a Data Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
 
 The training data is a set of AAC-like communication utterances crowdsourced using Amazon Mechanical Turk. 
 (Keith Vertanen and Per Ola Kristensson. The Imagination of Crowds: Conversational AAC Language Modeling using Crowdsourcing and Large Data Sources. In Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP). ACL: 700-711, 2011.)
@@ -68,7 +69,7 @@ The training data is a set of AAC-like communication utterances crowdsourced usi
 
 We incorporate this model in the ACAT framework with additional post-processing to make it more suitable for our use-case. Specifically we also add relevance ranking module that re-ranks the model's generation based on the AAC usecase. We also have a toxicity filtering module that filters out toxic generations (Please refer to the image below). 
 
-![image](https://github.com/intel-sandbox/ConvAssist/assets/89480559/d93ac63a-1e3d-454b-920f-54c7f0bfd631)
+![image](sentenceMode.jpg)
 
 
 ## Model Evaluation on Toxicity