add ner and pos_tagging

app.py

@@ -8,6 +8,8 @@ import apps.paraphrasing
 import apps.title_generation
 import apps.sentiment
 import apps.categorization
+import apps.ner
+import apps.pos_tagging
 
 st.set_page_config(
         page_title="Turna",
@@ -16,15 +18,16 @@ st.set_page_config(
     )
 
 PAGES = {
-    "Turna": apps.home,
+    "TURNA": apps.home,
     "Text Summarization": apps.summarization,
     "Text Paraphrasing": apps.paraphrasing,
     "News Title Generation": apps.title_generation,
     "Sentiment Classification": apps.sentiment,
     "Text Categorization": apps.categorization,
+    "Named Entity Recognition": apps.ner,
+    "Part-of-Speech Tagging": apps.pos_tagging
 }
 
-
 st.sidebar.title("Navigation")
 selection = st.sidebar.radio("Pages", list(PAGES.keys()))
 

apps/ner.py

@@ -0,0 +1,60 @@
+import requests
+import streamlit as st
+import time 
+from transformers import pipeline
+import os
+from .utils import query
+
+HF_AUTH_TOKEN = os.getenv('HF_AUTH_TOKEN')
+headers = {"Authorization": f"Bearer {HF_AUTH_TOKEN}"}
+
+def write():
+
+	st.markdown("# Named Entity Recognition")
+	st.sidebar.header("Named Entity Recognition")
+	st.write(
+		'''Here, you can detect named entities in your text using the fine-tuned TURNA NER models.'''
+	)
+
+	# Sidebar
+
+	# Taken from https://huggingface.co/spaces/flax-community/spanish-gpt2/blob/main/app.py
+	st.sidebar.subheader("Configurable parameters")
+
+	model_name = st.sidebar.selectbox(
+		"Model Selector",
+		options=[
+			"turna_ner_wikiann",
+			"turna_ner_milliyet"
+		],
+		index=0,
+	)
+	max_new_tokens = st.sidebar.number_input(
+		"Maximum length",
+		min_value=0,
+		max_value=64,
+		value=64,
+		help="The maximum length of the sequence to be generated.",
+	)
+
+	length_penalty = st.sidebar.number_input(
+		"Length penalty",
+		value=2.0,
+		help=" length_penalty > 0.0 promotes longer sequences, while length_penalty < 0.0 encourages shorter sequences. ",
+	)
+
+	no_repeat_ngram_size = st.sidebar.number_input(
+		"No Repeat N-Gram Size",
+		min_value=0,
+		value=3,
+		help="If set to int > 0, all ngrams of that size can only occur once.",
+	)
+
+	input_text = st.text_area(label='Enter a text: ', height=100, 
+			value="Ecevit, Irak hükümetinin de Ankara Büyükelçiliği için agreman istediğini belirtti.")
+	url = ("https://api-inference.huggingface.co/models/boun-tabi-LMG/" + model_name.lower())
+	params = {"length_penalty": length_penalty, "no_repeat_ngram_size": no_repeat_ngram_size, "max_new_tokens": max_new_tokens }
+	if st.button("Generate"):
+		with st.spinner('Generating...'):
+			output = query(input_text, url, params)
+			st.success(output)

apps/pos_tagging.py

@@ -0,0 +1,60 @@
+import requests
+import streamlit as st
+import time 
+from transformers import pipeline
+import os
+from .utils import query
+
+HF_AUTH_TOKEN = os.getenv('HF_AUTH_TOKEN')
+headers = {"Authorization": f"Bearer {HF_AUTH_TOKEN}"}
+
+def write():
+
+	st.markdown("# Part-of-Speech Tagging")
+	st.sidebar.header("Part-of-Speech Tagging")
+	st.write(
+		'''Here, you can detect part-of-speech tags in your text using the fine-tuned TURNA POS models.'''
+	)
+
+	# Sidebar
+
+	# Taken from https://huggingface.co/spaces/flax-community/spanish-gpt2/blob/main/app.py
+	st.sidebar.subheader("Configurable parameters")
+
+	model_name = st.sidebar.selectbox(
+		"Model Selector",
+		options=[
+			"turna_pos_boun",
+			"turna_pos_imst"
+		],
+		index=0,
+	)
+	max_new_tokens = st.sidebar.number_input(
+		"Maximum length",
+		min_value=0,
+		max_value=64,
+		value=64,
+		help="The maximum length of the sequence to be generated.",
+	)
+
+	length_penalty = st.sidebar.number_input(
+		"Length penalty",
+		value=2.0,
+		help=" length_penalty > 0.0 promotes longer sequences, while length_penalty < 0.0 encourages shorter sequences. ",
+	)
+
+	no_repeat_ngram_size = st.sidebar.number_input(
+		"No Repeat N-Gram Size",
+		min_value=0,
+		value=3,
+		help="If set to int > 0, all ngrams of that size can only occur once.",
+	)
+
+	input_text = st.text_area(label='Enter a text: ', height=100, 
+			value="Çünkü her kişinin bir başka yolu, bir başka yöntemi olmak gerektir.")
+	url = ("https://api-inference.huggingface.co/models/boun-tabi-LMG/" + model_name.lower())
+	params = {"length_penalty": length_penalty, "no_repeat_ngram_size": no_repeat_ngram_size, "max_new_tokens": max_new_tokens }
+	if st.button("Generate"):
+		with st.spinner('Generating...'):
+			output = query(input_text, url, params)
+			st.success(output)