v0.15.20 - Chatbot Model

What's Changed

• Enhance model selection persistence in chatbot by @jasonacox in #20
• Model selection behavior improved. Session will persist with selected model in that session even when refreshed. Last model selection will be remembered and auto-selected for any new sessions but will not impact existing sessions.

Full Changelog: v0.15.19...v0.15.20

v0.15.19 - Chatbot Docs

What's Changed

• v0.15.19 Chatbot Docs by @jasonacox in #19
• Update URL reader to display reading status to user while processing document.

Full Changelog: v0.15.18...v0.15.19

v0.15.18 - Chatbot Updates

What's Changed

• Chatbot update - model and image handling by @jasonacox in #18
• Model selection will now be stored as a cookie to allow it to persist between sessions.
• Image handling has been updated to recover when switching between vision models and language models. A new MAX_IMAGES setting has been added to allow persisting more than one image in the same conversation context (must be supported by model or the images will be pruned by chatbot)
• Model selection option /model list will display list of available models in chat windows.

Full Changelog: v0.15.17...v0.15.18

v0.15.17 - Model Selector

What's Changed

• Add model selection UI feature by @jasonacox in #16
• Chatbot - The /model command will now initiate a UI popup window and dropdown to allow the use to select a model from the list of available models. Alternatively, the user can specify the model with the command (e.g. /model mixtral) to select it immediately without the popup.

Full Changelog: v0.15.16...v0.15.17

v0.15.16 - Think Tags

What's Changed

• Chatbot - Add /think filter command and THINK_FILTER envrionmental setting to have chatbot filter out (no display) the <think></think> content from models that have built in CoT reasoning like Deepseek R1.
• Add /think filter command to chatbot by @jasonacox in #15

Full Changelog: v0.15.15...v0.15.16

v0.15.15 - Multi-model Support

What's Changed

• Chatbot: Add multi-model support with LiteLLM proxy instructions by @jasonacox in #13
• Docker Compose Quickstart by @jasonacox in #14

0.15.15 - Docker Compose

• Quick Start using Docker compose for Chatbot.
• Chatbot - Bug Fix: Remove token limit on response. The MAXTOKENS setting is used to prune content sent to LLM. If not set, no pruning will happen.
• Chatbot - Added additional LiteLLM support with the environmental settings LITELLM_PROXY and LITELLM_KEY. If set, these will override the OpenAI API settings to use LiteLLM and will remove EXTRA_BODY defaults that conflict with LiteLLM.
• LiteLLM - Added docker compose to start LiteLLM, PostgreSQL, and Chatbot.

0.15.14 - Multi-model Support

• Chatbot - Add /model command to list available models and dynamically set models during the session.
• LiteLLM - Added instructions to use LiteLLM proxy to combine local LLMs, AWS Bedrock, OpenAI, and other LLM options.

Full Changelog: v0.15.13...v0.15.15

v0.15.13 - Chatbot Fix

0.15.13 - Resource Fix

• Chatbot - Add LLM connection closures for non-streaming ad-hoc calls (e.g. CoT calls). This has removed the resource warning as identified in Issue #12. Improved debug messages.
• Chatbot Documentation - Updated CoT prompts and added reasoning.md for additional prompt options.

Full Changelog: v0.15.12...v0.15.13

v0.15.12 - CoT Updates

• Chatbot - Update Chain of Thought (CoT) to check request before routing all prompts through the CoT process. Using /think always will force CoT for all requests. Additionally, CoT prompts updated for better responses. Bug fixes and other minor improvements including documentation.
• Chatbot updated docker image: jasonacox/chatbot:0.15.12

Full Changelog: v0.15.11...v0.15.12

0.15.11 - Chain of Thought

• Chatbot - Add Chain of Thought (CoT) thinking option using the /think on or /think off toggles to the UI. When activated, queries will be passed through an out-of-band CoT loop to allow the LLM to thoughtfully explore answer and then provide a conclusion summary to the user. Set environmental variable "THINKING" to "true" to default all conversations to CoT mode.

Full Changelog: v0.15.10...v0.15.11

v0.15.10 - Bug Fix

• Chatbot - Fix error handling bug used to auto-detect max content length of LLM. Updated user input UI rendering to better handle indention.
• News Bot Script - Added logic to verify news summary from LLM to help prevent hallucinations.

Full Changelog: v0.15.9...v0.15.10