updated readme

Author: bhavsarpratik

.github/workflows/classification.yml

@@ -3,7 +3,7 @@ name: classification
 on:
   pull_request: 
     branches:
-      - main
+      - master # this can be main
     paths:
       - "classification/**"
 

README.md

@@ -1,2 +1,37 @@
-# serverless-transformers-on-aws
-Deploy transformers serverless on AWS Lambda
+# End2End Serverless Transformers On AWS Lambda for NLP 🚀
+
+Deploy transformers serverless on AWS Lambda with ease 💆‍♂️
+
+Current available pipelines
+1. classification
+1. translation (coming soon)
+1. token classification (need contribution)
+1. text generation (need contribution)
+1. zero shot classification (need contribution)
+
+## What you get with this?
+- ability to run transformers without servers
+- complete CI/CD
+- concurrency upto 1000 (default AWS limit)
+## How to use this?
+- clone the repo
+- keep the pipeline folder you want to use
+- modify the source and tests
+- keep the corresponding github action in `.github/workflows`
+- modify directory, registry and lambda function name in workflow
+- create repository in AWS ECR
+- set up secrets in repo (needed for access to AWS; this creds should have access to ECR and Lambda)
+    - AWS_ACCESS_KEY_ID
+    - AWS_SECRET_ACCESS_KEY
+- push the code
+- create PR
+    - this will build the container
+    - run all the tests
+    - push container to ECR registry
+    - update lambda with the new container (this will not happen when you push the first time)
+- create lambda function if it does not exist
+    - give appropriate IAM role
+    - set timeout and RAM
+- create API in API gateway and link to lambda
+
+Done! Now you can call the lambda using the API

classification/README.MD

@@ -1,2 +1,27 @@
 ## Classification service   
-Classification using AWS Lambda & Transformers
+Classification using Transformers on AWS Lambda. Check root readme for complete setup info.
+
+## Request format
+```
+{
+        "texts": ["food was great", "food was bad", "i am going out for food"],
+        "model_name": "cardiffnlp/twitter-roberta-base-sentiment",
+        "tokenizer_name": "roberta-base"
+    }
+```
+
+## Response format
+```
+{
+        'predictions': [{
+            'label': 'POSITIVE',
+            'score': 0.97
+        }, {
+            'label': 'NEGATIVE',
+            'score': 0.95
+        }, {
+            'label': 'NEUTRAL',
+            'score': 0.69
+        }]
+    }
+```