Fix TypeScript error: remove invalid role check in user message branch b312316 Running Pulastya B commited on 32 minutes ago
Fix Mistral message ordering: remove orphaned tool messages during pruning b54c2e5 Pulastya B commited on about 4 hours ago
Add comprehensive final summary instructions to system prompt b7bc364 Pulastya B commited on about 4 hours ago
Fix enhanced summary crash: handle best_model dict structure and add fallback 7f3326d Pulastya B commited on about 4 hours ago
Fix session memory: preserve uploaded files for follow-up requests b213e33 Pulastya B commited on about 5 hours ago
Fix plot rendering: extract from workflow_history, remove debug code, update system prompt 5992afe Pulastya B commited on about 6 hours ago
fix: Extract plots from workflow_history like reports 55d3b25 Pulastya B commited on about 7 hours ago
Fix: Handle infinity/NaN values in JSON serialization 84d9eaa Pulastya B commited on about 8 hours ago
Fix: Add mock tool responses for all blocking points 24ea09f Pulastya B commited on about 8 hours ago
Fix Mistral package version - use >=0.0.7 instead of >=1.0.0 2d15e4f Pulastya B commited on about 9 hours ago
Add Mistral AI integration - 1B tokens/month (33x Groq limits) be4eb33 Pulastya B commited on about 9 hours ago
Fix Phase 1 errors: schema extraction and message type handling 7d775b3 Pulastya B commited on about 19 hours ago
Refactor: Move workflow context out of LLM prompts into structured state 2797314 Pulastya B commited on about 19 hours ago
Fix TPM rate limiting: Add automatic retry with delay parsing 3e672a1 Pulastya B commited on about 20 hours ago
Remove Gemini fallback and improve rate limit error messages 7c2ff18 Pulastya B commited on about 20 hours ago
Fix syntax error: Proper try-except indentation in compression function 5fb3ea4 Pulastya B commited on about 20 hours ago
Fix compression crash: Add defensive checks for model data structure 230b136 Pulastya B commited on about 20 hours ago
Fix TPM rate limiting: Add token budget management for Groq 12K/min limit c206ce6 Pulastya B commited on about 20 hours ago
fix: Correct indentation in _compress_tool_result method 74cd041 Pulastya B commited on about 20 hours ago
feat: Add production-grade tool result compression for Groq f35ddc4 Pulastya B commited on about 20 hours ago
fix: Suppress harmless multiprocessing warnings from ydata-profiling 1ab1ded Pulastya B commited on 2 days ago
fix: Configure high-quality ydata profiling for HuggingFace 16GB RAM bdf714a Pulastya B commited on 2 days ago
feat: Add dynamic prompt system for small context window models (Groq support) b8bcf55 Pulastya B commited on 2 days ago
fix: Use stable gemini-2.5-flash instead of experimental model with stricter rate limits 562b130 Pulastya B commited on 2 days ago
fix: Support both GOOGLE_API_KEY and GEMINI_API_KEY env vars for HuggingFace compatibility fde5dd3 Pulastya B commited on 2 days ago
fix: Handle output_dir parameter mismatch in ydata_profiling 03b24f8 Pulastya B commited on 2 days ago
fix: Correct ToolCall object structure for Gemini text parser 943877f Pulastya B commited on 2 days ago
feat: Implement text-based tool calling for Gemini (ReAct pattern) 0310ce5 Pulastya B commited on 2 days ago
fix: Remove system_instruction parameter from chat endpoint eeeb8b7 Pulastya B commited on 2 days ago
fix: Add Gemini safety settings and handle blocked responses f8fdbd6 Pulastya B commited on 2 days ago