added bloom

demo_watermark.py

@@ -34,6 +34,7 @@ from watermark_processor import WatermarkLogitsProcessor, WatermarkDetector
 
 # FIXME correct lengths for all models
 API_MODEL_MAP = {
+    "bigscience/bloom"        : {"max_length": 2048, "gamma": 0.5, "delta": 2.0},
     "bigscience/bloomz"       : {"max_length": 2048, "gamma": 0.5, "delta": 2.0},
     "google/flan-ul2"         : {"max_length": 2048, "gamma": 0.5, "delta": 2.0},
     "google/flan-t5-xxl"      : {"max_length": 2048, "gamma": 0.5, "delta": 2.0},
@@ -461,7 +462,7 @@ def run_gradio(args, model=None, device=None, tokenizer=None):
                             Longer prompts that end mid-sentence will result in more fluent generations.
 
                             The larger models available in this demo are fine-tuned to follow instructions but have different strengths and will showcase different
-                            types of watermark behavior. [BLOOMZ (175B)](https://huggingface.co/bigscience/bloomz) is an instruction tuned variant of BLOOM capable of following instructions in dozens of languages zero-shot 
+                            types of watermark behavior. [BLOOMZ](https://huggingface.co/bigscience/bloomz) is an instruction tuned variant of [BLOOM (175B)](https://huggingface.co/bigscience/bloom) capable of following instructions in dozens of languages zero-shot 
                             and can generate long and coherent paragraphs and stories given the right prompt.
                             The FLAN models [FLAN-t5-xxl (11B)](https://huggingface.co/google/flan-t5-xxl) and [FLAN-UL2 (20B)](https://huggingface.co/google/flan-ul2) are fine-tuned on a variety of in-context few-shot learning NLP tasks, 
                             such as reasoning, and question answering.