Version 3.0
28 April, 2025
Important Note
Starting from v3.0, the text-language-v11 version must be used to enable shared usage of the text language detection module across all tenants. Since the existing tenant-specific pods will automatically timeout after a certain period, no additional action is required.
Important Note
Starting from v3.0, the text-manip service has been moved from the common namespace to the stai namespace. As part of this change, the new service address has been defined as an environment variable:
"SestekNerClient": {
"BaseUrl": "{{textmanipBaseUrl}}"
}
ai-agent Module Requirements
| CPU Request | CPU Limit | Memory Request | Memory Limit |
|---|---|---|---|
| 500m | 1000m | 2000Mi | 2000Mi |
Qdrant Vector DB Requirements
| CPU Request | CPU Limit | Memory Request | Memory Limit |
|---|---|---|---|
| 2000m | 2000m | 4000Mi | 4000Mi |
AI Module Note
Please use the following stable AI modules for the v3.0 update!
- ai-agent (v1) -- Optional service, requires Qdrant Vector Database
- sentence-embedding (v22)
- text-normalization (v29)
- text-classification-mlp (v25)
- question-answering (v17)
- text-sentiment (v17)
- text-clustering (v27)
- text-language (v11)
- text-summarization (v8) -- Requires Vllm for operation, cannot function independently.
Enhancements
- The ai-agent-v1 module has been tested and released. This module utilizes the Qdrant vector database to store created AI Agents.
Improvements
New Version of AI Modules:
- sentence-embedding-v22 has been tested and released.
- text-normalization-v29 has been tested and released.
- text-classification-mlp-v25 has been tested and released.
- question-answering-v17 has been tested and released.
- text-sentiment-v17 has been tested and released.
- text-clustering-v27 has been tested and released.
- text-language-v11 has been tested and released.