zazaman/guardrails-final

Simplify translation prompt to reduce reasoning time

Move re import to top of file

Filter out reasoning tokens and extract actual translation from Qwen output

Fix timeout error message to match actual timeout

Optimize translation: reduce max_tokens and context_size, add no-display-prompt flag

Remove --stop argument (not supported in llama.cpp CLI)

Build llama.cpp statically to avoid shared library dependencies

Build llama-cli target explicitly and improve binary detection

Fix binary name detection for llama.cpp (handle different binary names)

Disable CURL in llama.cpp build (not needed for local inference)

Compile llama.cpp in Dockerfile for architecture compatibility

Add automatic architecture detection and binary selection for llama.cpp

Add binary architecture compatibility check and better error handling

Add comprehensive logging with flush for translation debugging

Fix translation: add OS detection, better error handling and logging

Replace llama-cpp-python with pre-built llama.cpp binary for Qwen translator

Merge with HF Spaces and resolve README conflict

initial commit

Add multilingual translation support with Qwen3-0.6B-GGUF and optimize for Hugging Face Spaces deployment

Initial commit