Spaces:
Sleeping
Sleeping
Commit History
13 days ago
-
Sync backend to GitHub 246f221: Merge pull request #26 from siddhm11/feat/auto-context-switch
github-actions[bot]Automated by .github/workflows/sync-hf-space.yml8706d24 -
Sync backend to GitHub 912a16f: Video: 10-second clip prompts with start and end frames
github-actions[bot]Automated by .github/workflows/sync-hf-space.ymld730ddb
-
Sync backend to GitHub 7a32980: Merge pull request #9: fix tester feedback and guide new users (4.7.0)
github-actions[bot]Automated by .github/workflows/sync-hf-space.yml149dd33
-
Sync backend to GitHub f8d0b96: consent-gated memory, analytics API, structured logging
siddhm11Brings the Space up to date with 9 backend commits from GitHub main (ad58940..f8d0b96): - Passive memory written only on /enhance/accept, retrieval filtered on approved=true; existing unapproved vectors stop matching - Retrieval thresholds calibrated per layer (saved 0.24, passive 0.20) - New /api/user/analytics, /api/evaluate/prompt-improvement, /api/user/prompt-history/{id} - Account deletion ordered, retryable, profile deleted last - Prompt text removed from console logs; print -> JSON logger - Embedding LRU cache shared across memory layers Dockerfile, README.md and other Space-only files unchanged.10ce33e
-
Sync backend to GitHub d419447: voice transcription, builder dashboard, analytics
siddhm11 +1: Claude Opus 5Brings the Space up to date with siddhm11/prompt_engineering_skeleton@d419447. Adds POST /voice-transcribe (the extension has been calling it and getting 404), the private builder dashboard API, and analytics telemetry. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01NoNHrDfqGj7BLKagHkRM3Zb1f9886
-
Replace dead Groq models with a provider fallback chain
Verified signaturellama-3.3-70b-versatile and llama-3.1-8b-instant were decommissioned by Groq on 2026-08-16. The hardcoded model id meant every /enhance call failed with a 404, which the handler swallowed (it only rotated keys on 429), so the endpoint returned HTTP 200 with an empty enhancement. - services/providers.py: ordered fallback chain across models and providers, with dead-model detection, key rotation and a pooled HTTP client. Primary is qwen/qwen3.8-27b (reasoning off, best romanised-Hinglish fidelity of the surviving models), falling back to qwen3.6-27b, gpt-oss-120b and gpt-oss-20b. - Failures now surface as 429/503 with diagnostics instead of a silent 200. Empty completions are treated as failures, not successes. - BYOK: /enhance accepts a per-request user API key, never stored. - Romanised Hinglish is now detected and preserved (previously it was classified as English, and the prompt then ordered an English reply). - New /health/llm endpoint reports the chain and any dead models.8715f2a
-
Update backend from polished-with-website branch - new routers (feedback, saved_prompts), updated services and configs
siddhm11c0dd1c5
-
feat: update backend with complete prompt engine v4.0 - Enhanced main.py with request logging, CORS config, embedding preload - Updated config.py with tier-based model routing, rate limits, Stripe stubs - Expanded prompts.py with streaming, voice-enhance, feedback, history, usage endpoints - Added llm_service.py with Groq client pool and key rotation - Updated memory_service.py with passive learning, feedback analysis - Added migrate_embeddings.py for multilingual model migration - Updated schemas.py, auth.py, saved_prompts.py, requirements.txt - Removed prompt_engineering_skeleton directory
siddhm11257fe44
-
Update backend: mode-aware enhance, saved prompts CRUD, voice-enhance, feedback endpoint
siddhm114d5cbc7
-
refactor for HF spaces: moved code to backend/
siddhm11515a3fb