Refactor pattern configurations and update default settings

Author: Bob Strahan

CHANGELOG.md

@@ -32,6 +32,9 @@ SPDX-License-Identifier: MIT-0
   - Backward compatibility maintained - old parameter pattern still supported with deprecation warning
   - Updated all lambda functions and notebooks to use new simplified pattern
 - Removed fixed image target_height and target_width from default configurations, so images are processed in original resolution by default.
+- **Updated Default Configuration for Pattern1 and Pattern2**
+  - Changed default configuration for new stacks from "default" to "lending-package-sample" for both Pattern1 and Pattern2
+  - Maintains backward compatibility for stack updates by keeping the parameter value "default" mapped to the rvl-cdip-sample for pattern-2.
 
 
 ### Fixed

VERSION

@@ -1 +1 @@
-0.3.8-wip6
+0.3.8-alpha

config_library/pattern-1/README.md

@@ -27,4 +27,4 @@ See the main [README.md](../README.md) for more detailed instructions on creatin
 
 ## Available Configurations
 
-Currently, only the default configuration is available for Pattern 1. Contributions are welcome!
+Currently, only the default lending-package-sample configuration is available for Pattern 1. Contributions are welcome!

…nfig_library/pattern-1/default/README.md …ttern-1/lending-package-sample/README.mdconfig_library/pattern-1/default/README.md renamed to config_library/pattern-1/lending-package-sample/README.md config_library/pattern-1/default/README.md renamed to config_library/pattern-1/lending-package-sample/README.md

@@ -1,7 +1,7 @@
 Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
 SPDX-License-Identifier: MIT-0
 
-# Default Configuration
+# Default Configuration (lending-package-sample)
 
 This directory contains the default configuration for the GenAI IDP Accelerator. This configuration serves as the baseline for all document processing tasks and can be used as a starting point for creating custom configurations.
 

…ig_library/pattern-1/default/config.yaml …ern-1/lending-package-sample/config.yamlconfig_library/pattern-1/default/config.yaml renamed to config_library/pattern-1/lending-package-sample/config.yaml config_library/pattern-1/default/config.yaml renamed to config_library/pattern-1/lending-package-sample/config.yaml

@@ -1,7 +1,7 @@
 # Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
 # SPDX-License-Identifier: MIT-0
 
-notes: Default settings
+notes: Default settings for lending-package-sample configuration
 ocr:
   backend: "textract"  # Default to Textract for backward compatibility
   model_id: "us.anthropic.claude-3-7-sonnet-20250219-v1:0"
@@ -983,7 +983,7 @@ classification:
     DOCUMENT_IMAGE: Visual representation of the document page that provides layout, formatting, and visual structure information
     CLASS_NAMES_AND_DESCRIPTIONS: List of valid document types with their descriptions that the document must be classified into
     </variables>
-  classificationMethod: textbasedHolisticClassification
+  classificationMethod: multimodalPageLevelClassification
 extraction:
   image:
     target_width: '951'

…rn-1/default/samples/lending_package.pdf …ckage-sample/samples/lending_package.pdfconfig_library/pattern-1/default/samples/lending_package.pdf renamed to config_library/pattern-1/lending-package-sample/samples/lending_package.pdf config_library/pattern-1/default/samples/lending_package.pdf renamed to config_library/pattern-1/lending-package-sample/samples/lending_package.pdf

@@ -1,7 +1,7 @@
 # Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
 # SPDX-License-Identifier: MIT-0
 
-notes: Default settings
+notes: Default settings for the rvl-cdip-sample-with-few-shot-examples config
 ocr:
   backend: "textract"  # Default to Textract for backward compatibility
   model_id: "us.anthropic.claude-3-7-sonnet-20250219-v1:0"
@@ -657,58 +657,49 @@ classification:
   top_k: '5'
   max_tokens: '4096'
   system_prompt: >-
-    You are a document classification system that analyzes business documents,
-    forms, and publications. Your sole task is to classify documents into
-    categories based on their visual layout and textual content. You must:
+    You are a multimodal document classification expert that analyzes business documents using both visual layout and textual content. Your task is to classify single-page documents into predefined categories based on their structural patterns, visual features, and text content. Your output must be valid JSON according to the requested format.
 
-    1. Output only a JSON object containing a single "class" field with the
-    classification label
-
-    2. Use exactly one of the predefined categories, using the exact spelling
-    and case provided
-
-    3. Never include explanations, reasoning, or additional text in your
-    response
-
-    4. Respond with nothing but the JSON containing the classification
-
-
-    Example correct response:
-
-    {"class": "letter"}
+    <variables>
+    DOCUMENT_TEXT: OCR-extracted text content from the document page that provides textual information for classification
+    DOCUMENT_IMAGE: Visual representation of the document page that provides layout, formatting, and visual structure information
+    CLASS_NAMES_AND_DESCRIPTIONS: List of valid document types with their descriptions that the document must be classified into
+    </variables>
   task_prompt: >-
-    Classify this document into exactly one of these categories:
-
-
-    {CLASS_NAMES_AND_DESCRIPTIONS}
-
-
-    <few_shot_example_with_multimodal_page_classifications> 
-
-    {few_shot_example_with_multimodal_page_classificationS}
-
-    </few_shot_example_with_multimodal_page_classifications>
-
+    <reasoning-guidelines>
+    When determining the document type:
+    - First identify the document's primary purpose and function
+    - Note specific visual elements (letterhead, forms, tables, signatures)
+    - Identify key textual indicators (terminology, phrases, structure)
+    - Consider the document's intended audience and use case
+    - Provide specific evidence from both visual and textual analysis
+    </reasoning-guidelines>
+
+    <output-format>
+    Return your classification as valid JSON following this exact structure:
+    {
+      "classification_reason": "Detailed reasoning including specific visual and textual evidence that led to this classification",
+      "class": "exact_document_type_from_list"
+    }
+    </output-format>
 
     <<CACHEPOINT>>
 
+    <document-ocr-data>
+    {DOCUMENT_TEXT}
+    </document-ocr-data>
 
-    <document_ocr_data> 
-
-    {DOCUMENT_TEXT} 
-
-    </document_ocr_data>
-
-
-    <document_image> 
-
-    {DOCUMENT_IMAGE} 
-
-    </document_image>
-
+    <document-image>
+    {DOCUMENT_IMAGE}
+    </document-image>
 
-    Respond only with a JSON object containing the class label. For example:
-    {{"class": "letter"}}
+    <final-instructions>
+    Analyze the document above by:
+    1. Applying the <classification-instructions> to examine both visual and textual features
+    2. Following the <reasoning-guidelines> to build your classification rationale
+    3. Selecting ONLY from document types in <document-types>
+    4. Providing clear reasoning with specific evidence before the classification
+    5. Outputting in the exact JSON format specified in <output-format>
+    </final-instructions>
 extraction:
   image:
     target_height: ''