fix env

Dockerfile

@@ -8,7 +8,7 @@ WORKDIR /home/user/app
 
 RUN sed -i 's http://deb.debian.org http://cdn-aws.deb.debian.org g' /etc/apt/sources.list && sed -i 's http://archive.ubuntu.com http://us-east-1.ec2.archive.ubuntu.com g' /etc/apt/sources.list && sed -i '/security/d' /etc/apt/sources.list && apt-get update && apt-get install -y 	git 	git-lfs 	ffmpeg 	libsm6 	libxext6 	cmake 	libgl1-mesa-glx 	&& rm -rf /var/lib/apt/lists/* 	&& git lfs install
 
-RUN pip install --no-cache-dir Cython "gradio==3.28.3" "torch==1.10.1"
+RUN pip install --no-cache-dir Cython "gradio==3.28.3" "torch==1.10.1"  jieba subword-nmt
 
 RUN git clone --recurse-submodules https://github.com/thu-coai/DA-Transformer.git && cd DA-Transformer && git checkout demo && pip install -e . && cd dag_search && python3 setup.py build_ext --inplace && pip install -e . && cd ../..
 

app.py

@@ -28,7 +28,7 @@ from mass.s2s_model import TransformerMASSModel
 from transformer.hub_interface import TransformerHubInterface
 
 notice_markdown = ("""
-# Directed Acyclic Transformer: A Non-Autoregressive Sequence-to-Sequence Model designed for Parallel Text Generation.
+# ⚡ Directed Acyclic Transformer: A Non-Autoregressive Sequence-to-Sequence Model designed for Parallel Text Generation.
 - **Fast Generation**: DA-Transformer offers faster inference compared to autoregressive Transformers (with fairseq implementation), with a reduction in latency by 7~14x and an increase in throughput by ~20x.
 - **High Quality**: DA-Transformer performs competitively with autoregressive Transformers, even with pre-trained models like BART, in a variety of text generation tasks.
 - **Easy Training**: DA-Transformer can be trained end-to-end without requiring knowledge distillation, making it simple and straightforward to train.
@@ -527,7 +527,7 @@ def build_demo():
 
         gr.Markdown(learn_more_markdown)
 
-        compare_load(demo.load)
+        detail_load(demo.load)
 
         demo.load(None,None,None,_js=open("global.js").read())
     return demo