Spaces:
Sleeping
Sleeping
Commit History
-
Fix: Handle list type in context_priority_used extraction
882becc -
Upgrade to Hybrid Architecture v3 (Fast Lane + Deep Dive)
e2d9873 -
Add 21 Groq API keys with fallback mechanism
f895e11 -
Update Groq API keys with rotation logic
3d6dd09
-
Update Groq API key
1190215 -
Update Groq API key
499a614 -
feat: migrate from Gemini to Groq API for LLM calls
f0c4cb2
-
Increase max_output_tokens to 4096 to prevent response truncation
dcb1582 -
Complete rewrite: direct API fetch, proper vegetation indices + cluster stress extraction
d6e5751 -
Fix: format specifier error in SAR section
65058ca -
Fix: extract Sentinel-2 vegetation_indices_summary and SAR crop_health
15baaaf -
Improve prompt: force data-specific responses, add context logging
f11af9a -
Fix: handle None values in SAR bands section
c187541 -
Fix: add defensive null checks in generate_response
258275e -
Add comprehensive context: all indices, SAR, weather, persona-based responses
c7df30e -
Switch to gemini-2.5-flash (2.0 quota exhausted)
bec40f7 -
Fix: use gemini-2.0-flash model
ee7723b -
Add /chat/stream endpoint for Flutter compatibility
b965fde -
Restore original simple chatbot from initial commit
cb547c1 -
Simplify chatbot: single LLM call, remove complex reasoning
aaf7391 -
Fix: record_api_call once per query, not per key rotation
83e9fb0 -
Add strict 5 calls/minute rate limit with smart fallback
87f35a1 -
Reduce API usage: default to simple reasoning (1 call), add backoff
dd7cfa9 -
Add SSE streaming endpoint /chat/stream for typewriter effect
5d82df6 -
Fix: add 400 to rotation triggers for invalid API keys
0f136d3 -
Add 5-key Gemini API fallback, conversation memory, historical trends, and zone recommendations
504a512
-
Lower confidence thresholds for satellite data fetch and reasoning
a7903f2 -
Full Developer Spec implementation: Priority-based Multi-Stage Reasoning
- models.py: Pydantic schemas for all data structures - prompts.py: Enhanced LLM prompts with JSON output formats - context_aggregator.py: Priority-based satellite data formatting - reasoning_engine.py: 4-stage pipeline with evidence chains - priority_mapper.py: Intent-to-context mapping per spec - intent_classifier.py: Enhanced with phrases and regional support - app.py: Spec-compliant response with suggested_followupsee8289e -
Add satellite band data injection from HF Space APIs (SAR, Sentinel-2)
2221290 -
Auto-fetch field data (coordinates, crop_type) from Supabase
3fad61c -
Implement multi-stage reasoning architecture (Claim→Validate→Contradict→Confirm)
e2fa5a7 -
Add auto model discovery to find working Gemini model
07a23c0 -
Try gemini-1.5-flash model
c7be993 -
Switch to gemini-1.5-flash-8b for higher rate limits
c36a5b7 -
Add retry logic for rate limits
99bff2e -
Try gemini-2.0-flash-exp model
126daae -
Fix: Use v1beta API and add better error logging
fcbe917 -
Fix: Use gemini-pro with v1 API endpoint
0a05e0c -
Fix: Use direct Gemini REST API instead of SDK
43f4a4c -
Fix: Use models/gemini-1.5-flash format
5673ba4 -
Fix: Try multiple Gemini model names for compatibility
336ef34 -
Fix: Simplify Gemini integration and update dependencies
d633941 -
Fix: Use gemini-pro model
ba6427c -
Initial commit: AGROW Chatbot with Gemini LLM
b3afc51