Download data during app startup

Download embedding model during app startup

Download model during app startup

Apply fix for ChromaDB ValueError

Update prompts

Update examples

Enable thinking for answer

Pin langchain-* and langgraph-* packages

Update langchain-chroma to fix ChromaDB ValueError

Add Loading Data textbox

Revert model downloading

Import boto3

Download data from S3

Download model before running workflow (try 2)

Revert model download before running workflow

Download model before running workflow

Increase @spaces.GPU duration

Revert "Attempt fix for ChromaDB ValueError"

Revert fix for ChromaDB ValueError

Download model before running workflow

Attempt fix for ChromaDB ValueError

Update examples depending on compute mode

Remove print statements

Support multiple tool calls for local models

Further simplify ToolCallingLLM

Improve args for thinking mode

Clean up ToolCallingLLM

Display thinking output

Download data from Dropbox

Change model to Qwen3-14B

Increase @spaces.GPU duration

Lower @spaces.GPU duration

Use session state for LangChain graph

Use transformers==4.51.3

Handle tool calls with thinking enabled

Use transformers==4.51.3

Use transformers==4.52.4

Use transformers==4.51.3

Print message summaries for debugging

Use transformers==4.50.0

Pin torch version

Prevent automatic selection of remote mode

Pin transformers version

Update requirements.txt

Download data from S3 bucket

Change title

Update retrieval prompt

Cleanup data directories

Use custom chat interface

Update fonts