Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -4,7 +4,11 @@ from transformers import (
|
|
4 |
GPT2LMHeadModel, GPT2Tokenizer,
|
5 |
pipeline
|
6 |
)
|
|
|
|
|
7 |
st.title("Multi Chatbot")
|
|
|
|
|
8 |
models = {
|
9 |
"English to French": {
|
10 |
"name": "Helsinki-NLP/opus-mt-en-fr",
|
@@ -19,39 +23,62 @@ models = {
|
|
19 |
"description": "Generate creative stories based on input."
|
20 |
}
|
21 |
}
|
|
|
|
|
22 |
st.sidebar.header("Choose a Model")
|
23 |
selected_model_key = st.sidebar.radio("Select a Model:", list(models.keys()))
|
24 |
model_name = models[selected_model_key]["name"]
|
25 |
model_description = models[selected_model_key]["description"]
|
|
|
26 |
st.sidebar.markdown(f"### Model Description\n{model_description}")
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
45 |
user_input = st.text_input("Enter your query:")
|
|
|
46 |
if user_input:
|
47 |
if selected_model_key == "English to French":
|
48 |
try:
|
49 |
-
inputs =
|
50 |
-
outputs =
|
51 |
-
|
52 |
-
st.write(f"Translated Text: {
|
53 |
except Exception as e:
|
54 |
st.error(f"Error during translation: {e}")
|
|
|
55 |
elif selected_model_key == "Sentiment Analysis":
|
56 |
try:
|
57 |
result = sentiment_analyzer(user_input)[0]
|
@@ -59,12 +86,19 @@ if user_input:
|
|
59 |
st.write(f"Confidence: {result['score']:.2f}")
|
60 |
except Exception as e:
|
61 |
st.error(f"Error during sentiment analysis: {e}")
|
|
|
62 |
elif selected_model_key == "Story Generator":
|
63 |
try:
|
64 |
-
inputs =
|
65 |
-
outputs =
|
66 |
-
|
67 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
68 |
except Exception as e:
|
69 |
st.error(f"Error during story generation: {e}")
|
70 |
-
|
|
|
4 |
GPT2LMHeadModel, GPT2Tokenizer,
|
5 |
pipeline
|
6 |
)
|
7 |
+
|
8 |
+
# App title
|
9 |
st.title("Multi Chatbot")
|
10 |
+
|
11 |
+
# Define models and descriptions
|
12 |
models = {
|
13 |
"English to French": {
|
14 |
"name": "Helsinki-NLP/opus-mt-en-fr",
|
|
|
23 |
"description": "Generate creative stories based on input."
|
24 |
}
|
25 |
}
|
26 |
+
|
27 |
+
# Sidebar: Model selection
|
28 |
st.sidebar.header("Choose a Model")
|
29 |
selected_model_key = st.sidebar.radio("Select a Model:", list(models.keys()))
|
30 |
model_name = models[selected_model_key]["name"]
|
31 |
model_description = models[selected_model_key]["description"]
|
32 |
+
|
33 |
st.sidebar.markdown(f"### Model Description\n{model_description}")
|
34 |
+
|
35 |
+
# Cache model loading for efficiency
|
36 |
+
@st.cache_resource
|
37 |
+
def load_english_to_french():
|
38 |
+
tokenizer = MarianTokenizer.from_pretrained("Helsinki-NLP/opus-mt-en-fr")
|
39 |
+
model = MarianMTModel.from_pretrained("Helsinki-NLP/opus-mt-en-fr")
|
40 |
+
return tokenizer, model
|
41 |
+
|
42 |
+
@st.cache_resource
|
43 |
+
def load_sentiment_analysis():
|
44 |
+
return pipeline("sentiment-analysis", model="distilbert-base-uncased-finetuned-sst-2-english")
|
45 |
+
|
46 |
+
@st.cache_resource
|
47 |
+
def load_story_generator():
|
48 |
+
tokenizer = GPT2Tokenizer.from_pretrained("distilgpt2")
|
49 |
+
model = GPT2LMHeadModel.from_pretrained("distilgpt2")
|
50 |
+
tokenizer.pad_token = tokenizer.eos_token # Set pad token to EOS token
|
51 |
+
return tokenizer, model
|
52 |
+
|
53 |
+
# Load the selected model
|
54 |
+
if selected_model_key == "English to French":
|
55 |
+
st.write("Loading English to French model...")
|
56 |
+
en_fr_tokenizer, en_fr_model = load_english_to_french()
|
57 |
+
st.write("English to French model loaded successfully.")
|
58 |
+
|
59 |
+
elif selected_model_key == "Sentiment Analysis":
|
60 |
+
st.write("Loading Sentiment Analysis model...")
|
61 |
+
sentiment_analyzer = load_sentiment_analysis()
|
62 |
+
st.write("Sentiment Analysis model loaded successfully.")
|
63 |
+
|
64 |
+
elif selected_model_key == "Story Generator":
|
65 |
+
st.write("Loading Story Generator model...")
|
66 |
+
story_gen_tokenizer, story_gen_model = load_story_generator()
|
67 |
+
st.write("Story Generator model loaded successfully.")
|
68 |
+
|
69 |
+
# User input
|
70 |
user_input = st.text_input("Enter your query:")
|
71 |
+
|
72 |
if user_input:
|
73 |
if selected_model_key == "English to French":
|
74 |
try:
|
75 |
+
inputs = en_fr_tokenizer(user_input, return_tensors="pt", truncation=True, padding=True)
|
76 |
+
outputs = en_fr_model.generate(inputs["input_ids"], max_length=150, num_return_sequences=1)
|
77 |
+
translated_text = en_fr_tokenizer.decode(outputs[0], skip_special_tokens=True)
|
78 |
+
st.write(f"Translated Text: {translated_text}")
|
79 |
except Exception as e:
|
80 |
st.error(f"Error during translation: {e}")
|
81 |
+
|
82 |
elif selected_model_key == "Sentiment Analysis":
|
83 |
try:
|
84 |
result = sentiment_analyzer(user_input)[0]
|
|
|
86 |
st.write(f"Confidence: {result['score']:.2f}")
|
87 |
except Exception as e:
|
88 |
st.error(f"Error during sentiment analysis: {e}")
|
89 |
+
|
90 |
elif selected_model_key == "Story Generator":
|
91 |
try:
|
92 |
+
inputs = story_gen_tokenizer(user_input, return_tensors="pt", truncation=True, padding=True)
|
93 |
+
outputs = story_gen_model.generate(
|
94 |
+
inputs["input_ids"],
|
95 |
+
attention_mask=inputs["attention_mask"], # Pass the attention mask
|
96 |
+
max_length=200,
|
97 |
+
num_return_sequences=1,
|
98 |
+
temperature=0.7,
|
99 |
+
no_repeat_ngram_size=2
|
100 |
+
)
|
101 |
+
story = story_gen_tokenizer.decode(outputs[0], skip_special_tokens=True)
|
102 |
+
st.write(f"Generated Story: {story}")
|
103 |
except Exception as e:
|
104 |
st.error(f"Error during story generation: {e}")
|
|