imatag/Llama2_watermarking

try again

Default to a model that doesn't need a special authorization

Update README.md

simple prompt

Back to llama

Merge branch 'main' of https://huggingface.co/spaces/NohTow/Llama2_watermarking into main

Test debug using mistral

Remove prompt

Remove prompt

Adding self to get prompt

Fixes

Fixing indent

Adding system prompt in decoding

debug

Fixing comma

fixing prompt

Tuning WM parameters

Handling prompt

Adapting max_length to the prompt

Adding padding token

Removing prompt from the generated text

Remove window size from embed call

Forging prompt in the function rather than in the call

Adding AutoTokenizer

Moving the auth token to where the model is loaded

Creating model and tokenizer and passing it to watermarker init

Loading the model only once and adding LLama2 prompt

Adapting requirements.txt

aligning code with official demo

Initial commit

initial commit