Spaces:
Running
Running
Commit History
-
FIX: Replace invalid torch-cpu with --extra-index-url for CPU-only PyTorch install on HF Spaces Running
b42b77c -
CODEFIX: Added lazy loading and specific requirements, trying to solve problem with this not running via HuggingFace spaces
baeb731
-
CODEFIX: Added new logic in chat app to match most recent available models. Added .pyc to gitignore. Added new script start.ps1 for easy start of streamlit app from powershell.
85f2838
-
CODEFIX: Updated touse Gemma4 as default, but qwen3.5 if already loaded. Also shows what model you are chatting with.
0c96cc7
-
CODEFIX: Made the intro text shorter, as it would overlat the history box.
3432868 -
CODEFIX: Add token counter, context management, and UI improvements
- Add dynamic token counter with color-coded warnings (green/yellow/orange/red) - Add truncate_history() function to remove oldest messages when limit reached - Add truncation warning display when older messages are removed - Add chat_truncated flag to track truncation state - Update build_context() to use MAX_HISTORY_CHARS (~6k tokens) - Improve summarization with proper available space calculation - Add datasets.logging.set_verbosity_error() to hide info messages - Add CSS/JS to hide deploy button and info icon from header - Use st.set_page_config() for page title - Make user messages have semi-transparent gray background - Update system prompt from 512 to 1024 tokens - Update token counter to refresh after each response - Remove unused imports and duplicate constants - Cast response.content to str() for type consistency45dc510 -
CODEFIX: Set 12k num_ctx. Introduced history. Added history counter.. Put input control at the bottom and made it fixed. Lowered temperature.
0a460ad
-
CODEFIX: Removed the clear chat button
2b439ff -
CODEIFX: Try fix vector store retriever
bf8582b -
CODEFIX: Try fix runtime problem in HF Space
7ba8ef0 -
CODEFIX: Added specific requirements.
f586b15 -
CODEFIX: Added missing reference
6fabf64 -
Bug 20523: Web.Help: AI Chat stopped working. Changed to non-versioned requirements. Completely removed unused HF and Together references.
93b82d5 -
CODEFIX: Try to update langchain version to get this compiling on hf spaces
4c42b97 -
Bug 20523: Web.Help: AI Chat stopped working. Another try to get this working on HF Spaces
5ff5053 -
Update requirements.txt
Verified signatureed8a0c9 -
Update requirements.txt
Verified signatureTry #24f87477 -
Update app.py
Verified signatureTry upgrade to langchain 0.3.0c649b0f -
Update requirements.txt
Verified signatureupgrade to langchain 0.348fcdb4 -
Update requirements.txt
Verified signatureTry again0408caf -
Update requirements.txt
Verified signatureRemoved streamlit09f2410 -
Create .force-rebuild
Verified signature48b6565 -
Update requirements.txt
Verified signaturef1727c2 -
Bug 20523: Web.Help: AI Chat stopped working
5edef6b -
Update requirements.txt
Verified signatureNew try, fix imports2563943
-
Update requirements.txt
Verified signaturenew try rebuild requirements33164c4 -
Update requirements.txt
Verified signatureUpdated requirementsa18536c -
Update app.py
Verified signatureAnother try. Changed reference to docs #2b87a0f1 -
Update app.py
Verified signatureUpdated invalid Doc reference. It must have been modernized.82bd26b -
Update app.py
Verified signatureTry #3468274b -
Update app.py
Verified signatureTry#23a9e531 -
Update app.py
Verified signatureTry changing import to fix deprecated langchain imports56a94d6 -
Update app.py
Verified signatureRetry adding token to adrega ollama installationc44f7b8 -
Update app.py
Verified signatureAI Edited to allow passing of api_key to adrega442c1a6
-
CODEFIX: Try to solve problem where HF has problems generating contents due to missing GPU access.
4036ee3 -
CODEFIX: Added Adrega styling to fonts and improved breaking after bulletpoints etc
1af7545
-
CODEFIX: Made the text more readable by introducing markup
01e1fc7 -
CODEFIX: Added adrega first and together last, because now we have our own fully working inference client
3528ff4
-
CODEFIX: Narrowed padding left and right to remove whitespace in Adrega chat
c5caedf -
CODEFIX: Try to remove padding from main container in Streamlit to better fit inside Adrega AI Chat window.
b50deba -
CODEFIX: Removed HuggingFace as provider, as it always displayed thought process. Improved responces. Improved removal of thought process in general.
5c9b053 -
CODEFIX: Improved responses from Adrega and Hugging Face, to prevent them from revealing thought process.
c63b0b5 -
CODEFIX: Improved the models to act in a more traditional chat way, instead of being in q&a mode.
7a5082a -
CODEFIX: Added css file for style rule. Added 1 rule, to narrow the padding on the first control. Added memory for the model
b2cd6f6
-
CODEFIX: Set new Adrega model to Qwen3:1.7b, disapling the thinking specifically for all models. Made quesitons appear in grey
59bae64 -
CODEFIX: Updated to use new json based context
5e4ad3f
-
CODEFIX: Try3 - Added langchain-openai library
dccd131 -
CODEFIX: Try2 - Add langchain-huggingface specifically to requirements
e1515fc -
CODEFIX: Try fix error with incorrect requirements.txt
b1baa36 -
CODEFIX: Implemented Together and Adrega as additional providers. Defaults to Together. Implemented langchain to split requests into chunks
7e234ff