NeuralSeek Change Log
September 2023
- IBM watsonx Granite Models Supported
- AWS Bedrock Integration and Models Supported
- NEW FEATURE – Explore: An Open-Ended Retrieval Augmented Generation Playground
- Kore.ai Round Trip Monitoring
- Refinements to Vector Similarity Matching
- Llama 2 Chat Model Support
- HuggingFace Integration for Supported Models
- Vector Similarity for Intent Matching
- AWS OpenSearch Integration
August 2023
- IBM Watson Assistant Dialog Multi-Turn Conversation Templates
- New ‘Seek’ Parameter Call to Indicate LLM Performance
- BYO-LLM – IBM watsonx language translation
- Summarization of document passage results
- Link Summarization of NeuralSeek Results, 1-5 Result Links
- Ability to set specific language on each LLM – e.g., “use <this> model for Spanish Seek / Translation”
- AWS Kendra Search/Index KnowledgeBase connector
- ‘Bring Your Own’ Large Language Model (BYO-LLM) cards – ability to use multiple LLMs for a specific task
- AWS Lex Multi-Turn Conversation Generation Templates
July 2023
- Ability to auto-fill slots when gathering information
- KnowledgeBase translation updated – questions now get translated to KnowledgeBase source language for summarization
- Offline spreadsheet editing with upload to Curate tab
- ConsoleAPI under Integrate tab
- Streaming Endpoint – users can now enable streaming responses from NeuralSeek
- Streaming Responses from LLM
- Cross-lingual support when using language code “xx” (Match Input) enhanced
- Translate Endpoint
- Curate to CSV / Upload Curated QA from CSV
- Semantic Match Analysis to describe the logic for the Semantic Score enhanced
- IBM watsonx Model Compatibility
- Local / On-Prem / Kubernetes (K8s) deployments supported
- Identify Language Endpoint
- AWS Lex Round-Trip Monitoring
- Entity Extraction feature, Custom Entity Creation
- GitHub Knowledge Repository
June 2023
- AWS Partnership Announcement
- IBM watsonx (LLM) connector
- Improvements to Caching
- Confidence and Coverage Score Graphs added to Curate tab
May 2023
- Analytics API endpoint
- Data Cleanser for non-HTML enabled
- Table Extraction model to enable answers from tabular data
March 2023
- Personal Identifiable Information (PII) Detection
- Round-Trip Logging
- Sentiment Analysis
- Source Document Monitoring and Answer Regeneration
- User-specified input length enabled
February 2023
- Additional languages enabled (Chinese, Czech, Dutch, Indonesian, Japanese)
- Auto-Build Multi-Step Action
- Enhanced API to allow run-time modification of all parameters
- KB tuning parameters enabled
- Large Language Model (LLM) tuning
- Personalization of generated answers