Commit
Β·
097b91f
1
Parent(s):
819c99b
testing new generation strategy
Browse files
README.md
CHANGED
@@ -5,7 +5,7 @@ colorFrom: gray
|
|
5 |
colorTo: purple
|
6 |
sdk: gradio
|
7 |
sdk_version: 3.37.0
|
8 |
-
app_file: app
|
9 |
pinned: true
|
10 |
license: bigcode-openrail-m
|
11 |
tags:
|
|
|
5 |
colorTo: purple
|
6 |
sdk: gradio
|
7 |
sdk_version: 3.37.0
|
8 |
+
app_file: app.py
|
9 |
pinned: true
|
10 |
license: bigcode-openrail-m
|
11 |
tags:
|
app.py
CHANGED
@@ -1,15 +1,109 @@
|
|
1 |
import os
|
2 |
-
|
|
|
|
|
|
|
|
|
|
|
3 |
from transformers import (
|
4 |
AutoModelForCausalLM,
|
5 |
AutoTokenizer,
|
6 |
StoppingCriteria,
|
7 |
StoppingCriteriaList,
|
8 |
-
TextIteratorStreamer
|
9 |
)
|
10 |
-
import
|
|
|
|
|
|
|
|
|
|
|
11 |
import torch
|
12 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
13 |
|
14 |
model_name = os.getenv("HF_MODEL_NAME", None)
|
15 |
tok = AutoTokenizer.from_pretrained(model_name)
|
@@ -24,22 +118,20 @@ m = AutoModelForCausalLM.from_pretrained(
|
|
24 |
#load_in_8bit=True,
|
25 |
)
|
26 |
|
27 |
-
m.config.pad_token_id = m.config.eos_token_id
|
28 |
-
m.generation_config.pad_token_id = m.config.eos_token_id
|
29 |
-
|
30 |
-
stop_tokens = [";", "###", "Result"]
|
31 |
-
stop_token_ids = tok.convert_tokens_to_ids(stop_tokens)
|
32 |
|
33 |
print(f"Successfully loaded the model {model_name} into memory")
|
34 |
|
35 |
-
class StopOnTokens(StoppingCriteria):
|
36 |
-
def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> bool:
|
37 |
-
for stop_id in stop_token_ids:
|
38 |
-
if input_ids[0][-1] == stop_id:
|
39 |
-
return True
|
40 |
-
return False
|
41 |
|
42 |
-
def
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
43 |
stop = StopOnTokens()
|
44 |
|
45 |
# Format the user's input message
|
@@ -52,12 +144,14 @@ def bot(input_message: str, db_info="", temperature=0.1, top_p=0.9, top_k=0, rep
|
|
52 |
input_ids=input_ids,
|
53 |
max_new_tokens=max_new_tokens,
|
54 |
temperature=temperature,
|
55 |
-
do_sample=temperature > 0.0,
|
56 |
top_p=top_p,
|
57 |
top_k=top_k,
|
58 |
repetition_penalty=repetition_penalty,
|
59 |
streamer=streamer,
|
60 |
stopping_criteria=StoppingCriteriaList([stop]),
|
|
|
|
|
|
|
61 |
)
|
62 |
|
63 |
stream_complete = Event()
|
@@ -73,64 +167,92 @@ def bot(input_message: str, db_info="", temperature=0.1, top_p=0.9, top_k=0, rep
|
|
73 |
for new_text in streamer:
|
74 |
partial_text += new_text
|
75 |
|
76 |
-
|
77 |
-
try:
|
78 |
-
final_query = partial_text.split("|")[1].strip()
|
79 |
-
except Exception:
|
80 |
-
final_query = partial_text
|
81 |
|
82 |
-
|
83 |
-
#
|
84 |
-
|
85 |
-
except Exception:
|
86 |
-
# If formatting fails, use the original, unformatted query
|
87 |
-
formatted_query = final_query
|
88 |
|
89 |
-
|
90 |
-
final_query_markdown = f"{formatted_query}"
|
91 |
-
return final_query_markdown
|
92 |
|
|
|
93 |
with gr.Blocks(theme='gradio/soft') as demo:
|
|
|
94 |
header = gr.HTML("""
|
95 |
-
<h1 style="text-align: center">SQL
|
96 |
-
<h3 style="text-align: center"
|
|
|
|
|
|
|
97 |
""")
|
98 |
|
99 |
-
output_box = gr.Code(label="Generated SQL", lines=2, interactive=
|
|
|
|
|
|
|
|
|
|
|
100 |
input_text = gr.Textbox(lines=3, placeholder='Write your question here...', label='NL Input')
|
101 |
-
db_info = gr.Textbox(lines=4, placeholder='Example: | table_01 : column_01 , column_02 | table_02 : column_01 , column_02 | ...', label='Database Info')
|
|
|
|
|
|
|
|
|
|
|
102 |
|
103 |
-
with gr.Accordion("
|
104 |
-
temperature = gr.Slider(label="Temperature", minimum=0.0, maximum=1.0, value=0.
|
105 |
top_p = gr.Slider(label="Top-p (nucleus sampling)", minimum=0.0, maximum=1.0, value=0.9, step=0.01)
|
106 |
top_k = gr.Slider(label="Top-k", minimum=0, maximum=200, value=0, step=1)
|
107 |
repetition_penalty = gr.Slider(label="Repetition Penalty", minimum=1.0, maximum=2.0, value=1.08, step=0.01)
|
|
|
|
|
|
|
|
|
|
|
108 |
|
109 |
-
|
110 |
-
|
111 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
112 |
examples = gr.Examples([
|
113 |
-
["What is the average
|
|
|
|
|
114 |
["Show location and name for all stadiums with a capacity between 5000 and 10000.", "| stadium : stadium_id , location , name , capacity , highest , lowest , average | singer : singer_id , name , country , song_name , song_release_year , age , is_male | concert : concert_id , concert_name , theme , stadium_id , year | singer_in_concert : concert_id , singer_id | concert.stadium_id = stadium.stadium_id | singer_in_concert.singer_id = singer.singer_id | singer_in_concert.concert_id = concert.concert_id |"],
|
115 |
["What are the number of concerts that occurred in the stadium with the largest capacity ?", "| stadium : stadium_id , location , name , capacity , highest , lowest , average | singer : singer_id , name , country , song_name , song_release_year , age , is_male | concert : concert_id , concert_name , theme , stadium_id , year | singer_in_concert : concert_id , singer_id | concert.stadium_id = stadium.stadium_id | singer_in_concert.singer_id = singer.singer_id | singer_in_concert.concert_id = concert.concert_id |"],
|
116 |
-
["
|
117 |
-
["List the names of all singers who performed in a concert with the theme 'Rock'", "| stadium : stadium_id , location , name , capacity , highest , lowest , average | singer : singer_id , name , country , song_name , song_release_year , age , is_male | concert : concert_id , concert_name , theme , stadium_id , year | singer_in_concert : concert_id , singer_id | concert.stadium_id = stadium.stadium_id | singer_in_concert.singer_id = singer.singer_id | singer_in_concert.concert_id = concert.concert_id |"]
|
118 |
-
|
119 |
-
|
120 |
-
bitsandbytes_model = "richardr1126/spider-skeleton-wizard-coder-8bit"
|
121 |
-
merged_model = "richardr1126/spider-skeleton-wizard-coder-merged"
|
122 |
-
initial_model = "WizardLM/WizardCoder-15B-V1.0"
|
123 |
-
finetuned_model = "richardr1126/spider-skeleton-wizard-coder-qlora"
|
124 |
-
dataset = "richardr1126/spider-skeleton-context-instruct"
|
125 |
-
|
126 |
-
footer = gr.HTML(f"""
|
127 |
-
<p>π οΈ If you want you can <strong>duplicate this Space</strong>, then change the HF_MODEL_REPO spaces env varaible to use any Transformers model.</p>
|
128 |
-
<p>π Leveraging the <a href='https://huggingface.co/{bitsandbytes_model}'><strong>bitsandbytes 8-bit version</strong></a> of <a href='https://huggingface.co/{merged_model}'><strong>{merged_model}</strong></a> model.</p>
|
129 |
-
<p>π How it's made: <a href='https://huggingface.co/{initial_model}'><strong>{initial_model}</strong></a> was finetuned to create <a href='https://huggingface.co/{finetuned_model}'><strong>{finetuned_model}</strong></a>, then merged together to create <a href='https://huggingface.co/{merged_model}'><strong>{merged_model}</strong></a>.</p>
|
130 |
-
<p>π Fine-tuning was performed using QLoRA techniques on the <a href='https://huggingface.co/datasets/{dataset}'><strong>{dataset}</strong></a> dataset. You can view training metrics on the <a href='https://huggingface.co/{finetuned_model}'><strong>QLoRa adapter HF Repo</strong></a>.</p>
|
131 |
-
""")
|
132 |
|
133 |
|
134 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
135 |
|
136 |
-
demo.queue(concurrency_count=1, max_size=
|
|
|
1 |
import os
|
2 |
+
import gradio as gr
|
3 |
+
import sqlparse
|
4 |
+
import requests
|
5 |
+
from time import sleep
|
6 |
+
import re
|
7 |
+
import platform
|
8 |
from transformers import (
|
9 |
AutoModelForCausalLM,
|
10 |
AutoTokenizer,
|
11 |
StoppingCriteria,
|
12 |
StoppingCriteriaList,
|
13 |
+
TextIteratorStreamer
|
14 |
)
|
15 |
+
from threading import Event, Thread
|
16 |
+
# Additional Firebase imports
|
17 |
+
import firebase_admin
|
18 |
+
from firebase_admin import credentials, firestore
|
19 |
+
import json
|
20 |
+
import base64
|
21 |
import torch
|
22 |
+
|
23 |
+
|
24 |
+
print(f"Running on {platform.system()}")
|
25 |
+
|
26 |
+
if platform.system() == "Windows" or platform.system() == "Darwin":
|
27 |
+
from dotenv import load_dotenv
|
28 |
+
load_dotenv()
|
29 |
+
|
30 |
+
quantized_model = "richardr1126/spider-skeleton-wizard-coder-8bit"
|
31 |
+
merged_model = "richardr1126/spider-skeleton-wizard-coder-merged"
|
32 |
+
initial_model = "WizardLM/WizardCoder-15B-V1.0"
|
33 |
+
lora_model = "richardr1126/spider-skeleton-wizard-coder-qlora"
|
34 |
+
dataset = "richardr1126/spider-skeleton-context-instruct"
|
35 |
+
|
36 |
+
# Firebase code
|
37 |
+
# Initialize Firebase
|
38 |
+
base64_string = os.getenv('FIREBASE')
|
39 |
+
base64_bytes = base64_string.encode('utf-8')
|
40 |
+
json_bytes = base64.b64decode(base64_bytes)
|
41 |
+
json_data = json_bytes.decode('utf-8')
|
42 |
+
|
43 |
+
firebase_auth = json.loads(json_data)
|
44 |
+
|
45 |
+
# Load credentials and initialize Firestore
|
46 |
+
cred = credentials.Certificate(firebase_auth)
|
47 |
+
firebase_admin.initialize_app(cred)
|
48 |
+
db = firestore.client()
|
49 |
+
|
50 |
+
def log_message_to_firestore(input_message, db_info, temperature, response_text):
|
51 |
+
doc_ref = db.collection('codellama-logs').document()
|
52 |
+
log_data = {
|
53 |
+
'timestamp': firestore.SERVER_TIMESTAMP,
|
54 |
+
'temperature': temperature,
|
55 |
+
'db_info': db_info,
|
56 |
+
'input': input_message,
|
57 |
+
'output': response_text,
|
58 |
+
}
|
59 |
+
doc_ref.set(log_data)
|
60 |
+
|
61 |
+
rated_outputs = set() # set to store already rated outputs
|
62 |
+
|
63 |
+
def log_rating_to_firestore(input_message, db_info, temperature, response_text, rating):
|
64 |
+
global rated_outputs
|
65 |
+
output_id = f"{input_message} {db_info} {response_text} {temperature}"
|
66 |
+
|
67 |
+
if output_id in rated_outputs:
|
68 |
+
gr.Warning("You've already rated this output!")
|
69 |
+
return
|
70 |
+
if not input_message or not response_text or not rating:
|
71 |
+
gr.Info("You haven't asked a question yet!")
|
72 |
+
return
|
73 |
+
|
74 |
+
rated_outputs.add(output_id)
|
75 |
+
|
76 |
+
doc_ref = db.collection('codellama-ratings').document()
|
77 |
+
log_data = {
|
78 |
+
'timestamp': firestore.SERVER_TIMESTAMP,
|
79 |
+
'temperature': temperature,
|
80 |
+
'db_info': db_info,
|
81 |
+
'input': input_message,
|
82 |
+
'output': response_text,
|
83 |
+
'rating': rating,
|
84 |
+
}
|
85 |
+
doc_ref.set(log_data)
|
86 |
+
gr.Info("Thanks for your feedback!")
|
87 |
+
# End Firebase code
|
88 |
+
|
89 |
+
def format(text):
|
90 |
+
# Split the text by "|", and get the last element in the list which should be the final query
|
91 |
+
try:
|
92 |
+
final_query = text.split("|")[1].strip()
|
93 |
+
except Exception:
|
94 |
+
final_query = text
|
95 |
+
|
96 |
+
try:
|
97 |
+
# Attempt to format SQL query using sqlparse
|
98 |
+
formatted_query = sqlparse.format(final_query, reindent=True, keyword_case='upper')
|
99 |
+
except Exception:
|
100 |
+
# If formatting fails, use the original, unformatted query
|
101 |
+
formatted_query = final_query
|
102 |
+
|
103 |
+
# Convert SQL to markdown (not required, but just to show how to use the markdown module)
|
104 |
+
final_query_markdown = f"{formatted_query}"
|
105 |
+
|
106 |
+
return final_query_markdown
|
107 |
|
108 |
model_name = os.getenv("HF_MODEL_NAME", None)
|
109 |
tok = AutoTokenizer.from_pretrained(model_name)
|
|
|
118 |
#load_in_8bit=True,
|
119 |
)
|
120 |
|
121 |
+
# m.config.pad_token_id = m.config.eos_token_id
|
122 |
+
# m.generation_config.pad_token_id = m.config.eos_token_id
|
|
|
|
|
|
|
123 |
|
124 |
print(f"Successfully loaded the model {model_name} into memory")
|
125 |
|
|
|
|
|
|
|
|
|
|
|
|
|
126 |
|
127 |
+
def generate(input_message: str, db_info="", temperature=0.2, top_p=0.9, top_k=0, repetition_penalty=1.08, format_sql=True, log=False, num_return_sequences=1, num_beams=1, do_sample=False):
|
128 |
+
stop_token_ids = tok.convert_tokens_to_ids(["###"])
|
129 |
+
class StopOnTokens(StoppingCriteria):
|
130 |
+
def __call__(self, input_ids: torch.LongTensor, scores: torch.FloatTensor, **kwargs) -> bool:
|
131 |
+
for stop_id in stop_token_ids:
|
132 |
+
if input_ids[0][-1] == stop_id:
|
133 |
+
return True
|
134 |
+
return False
|
135 |
stop = StopOnTokens()
|
136 |
|
137 |
# Format the user's input message
|
|
|
144 |
input_ids=input_ids,
|
145 |
max_new_tokens=max_new_tokens,
|
146 |
temperature=temperature,
|
|
|
147 |
top_p=top_p,
|
148 |
top_k=top_k,
|
149 |
repetition_penalty=repetition_penalty,
|
150 |
streamer=streamer,
|
151 |
stopping_criteria=StoppingCriteriaList([stop]),
|
152 |
+
num_return_sequences=num_return_sequences,
|
153 |
+
num_beams=num_beams,
|
154 |
+
do_sample=do_sample,
|
155 |
)
|
156 |
|
157 |
stream_complete = Event()
|
|
|
167 |
for new_text in streamer:
|
168 |
partial_text += new_text
|
169 |
|
170 |
+
output = format(partial_text) if format_sql else partial_text
|
|
|
|
|
|
|
|
|
171 |
|
172 |
+
if log:
|
173 |
+
# Log the request to Firestore
|
174 |
+
log_message_to_firestore(input_message, db_info, temperature, output)
|
|
|
|
|
|
|
175 |
|
176 |
+
return output
|
|
|
|
|
177 |
|
178 |
+
# Gradio UI Code
|
179 |
with gr.Blocks(theme='gradio/soft') as demo:
|
180 |
+
# Elements stack vertically by default just define elements in order you want them to stack
|
181 |
header = gr.HTML("""
|
182 |
+
<h1 style="text-align: center">SQL CodeLlama Demo</h1>
|
183 |
+
<h3 style="text-align: center">π·οΈβ οΈπ¦ Generate SQL queries from Natural Language π·οΈβ οΈπ§π¦</h3>
|
184 |
+
<div style="max-width: 450px; margin: auto; text-align: center">
|
185 |
+
<p style="font-size: 12px; text-align: center">β οΈ Should take 30-60s to generate. Please rate the response, it helps a lot. If you get a blank output, the model server is currently down, please try again another time.</p>
|
186 |
+
</div>
|
187 |
""")
|
188 |
|
189 |
+
output_box = gr.Code(label="Generated SQL", lines=2, interactive=False)
|
190 |
+
|
191 |
+
with gr.Row():
|
192 |
+
rate_up = gr.Button("π", variant="secondary")
|
193 |
+
rate_down = gr.Button("π", variant="secondary")
|
194 |
+
|
195 |
input_text = gr.Textbox(lines=3, placeholder='Write your question here...', label='NL Input')
|
196 |
+
db_info = gr.Textbox(lines=4, placeholder='Make sure to place your tables information inside || for better results. Example: | table_01 : column_01 , column_02 | table_02 : column_01 , column_02 | ...', label='Database Info')
|
197 |
+
format_sql = gr.Checkbox(label="Format SQL + Remove Skeleton", value=True, interactive=True)
|
198 |
+
|
199 |
+
with gr.Row():
|
200 |
+
run_button = gr.Button("Generate SQL", variant="primary")
|
201 |
+
clear_button = gr.ClearButton(variant="secondary")
|
202 |
|
203 |
+
with gr.Accordion("Options", open=False):
|
204 |
+
temperature = gr.Slider(label="Temperature", minimum=0.0, maximum=1.0, value=0.2, step=0.1)
|
205 |
top_p = gr.Slider(label="Top-p (nucleus sampling)", minimum=0.0, maximum=1.0, value=0.9, step=0.01)
|
206 |
top_k = gr.Slider(label="Top-k", minimum=0, maximum=200, value=0, step=1)
|
207 |
repetition_penalty = gr.Slider(label="Repetition Penalty", minimum=1.0, maximum=2.0, value=1.08, step=0.01)
|
208 |
+
|
209 |
+
with gr.Accordion("Generation strategies", open=False):
|
210 |
+
num_return_sequences = gr.Slider(label="Num Return Sequences", minimum=1, maximum=5, value=1, step=1)
|
211 |
+
num_beams = gr.Slider(label="Num Beams", minimum=1, maximum=5, value=1, step=1)
|
212 |
+
do_sample = gr.Checkbox(label="Do Sample", value=False, interactive=True)
|
213 |
|
214 |
+
info = gr.HTML(f"""
|
215 |
+
<p>π Leveraging the <a href='https://huggingface.co/{quantized_model}'><strong>bitsandbytes 8-bit version</strong></a> of <a href='https://huggingface.co/{merged_model}'><strong>{merged_model}</strong></a> model.</p>
|
216 |
+
<p>π How it's made: <a href='https://huggingface.co/{initial_model}'><strong>{initial_model}</strong></a> was finetuned to create <a href='https://huggingface.co/{lora_model}'><strong>{lora_model}</strong></a>, then merged together to create <a href='https://huggingface.co/{merged_model}'><strong>{merged_model}</strong></a>.</p>
|
217 |
+
<p>π Fine-tuning was performed using QLoRA techniques on the <a href='https://huggingface.co/datasets/{dataset}'><strong>{dataset}</strong></a> dataset. You can view training metrics on the <a href='https://huggingface.co/{lora_model}'><strong>QLoRa adapter HF Repo</strong></a>.</p>
|
218 |
+
<p>π All inputs/outputs are logged to Firebase to see how the model is doing. You can also leave a rating for each generated SQL the model produces, which gets sent to the database as well.</a></p>
|
219 |
+
""")
|
220 |
+
|
221 |
+
examples = gr.Examples([
|
222 |
+
["What is the average, minimum, and maximum age of all singers from France?", "| stadium : stadium_id , location , name , capacity , highest , lowest , average | singer : singer_id , name , country , song_name , song_release_year , age , is_male | concert : concert_id , concert_name , theme , stadium_id , year | singer_in_concert : concert_id , singer_id | concert.stadium_id = stadium.stadium_id | singer_in_concert.singer_id = singer.singer_id | singer_in_concert.concert_id = concert.concert_id |"],
|
223 |
+
["How many students have dogs?", "| student : stuid , lname , fname , age , sex , major , advisor , city_code | has_pet : stuid , petid | pets : petid , pettype , pet_age , weight | has_pet.stuid = student.stuid | has_pet.petid = pets.petid | pets.pettype = 'Dog' |"],
|
224 |
+
], inputs=[input_text, db_info, temperature, top_p, top_k, repetition_penalty, format_sql], fn=generate, cache_examples=False if platform.system() == "Windows" or platform.system() == "Darwin" else True, outputs=output_box)
|
225 |
+
|
226 |
+
with gr.Accordion("More Examples", open=False):
|
227 |
examples = gr.Examples([
|
228 |
+
["What is the average weight of pets of all students?", "| student : stuid , lname , fname , age , sex , major , advisor , city_code | has_pet : stuid , petid | pets : petid , pettype , pet_age , weight | has_pet.stuid = student.stuid | has_pet.petid = pets.petid |"],
|
229 |
+
["How many male singers performed in concerts in the year 2023?", "| stadium : stadium_id , location , name , capacity , highest , lowest , average | singer : singer_id , name , country , song_name , song_release_year , age , is_male | concert : concert_id , concert_name , theme , stadium_id , year | singer_in_concert : concert_id , singer_id | concert.stadium_id = stadium.stadium_id | singer_in_concert.singer_id = singer.singer_id | singer_in_concert.concert_id = concert.concert_id |"],
|
230 |
+
["For students who have pets, how many pets does each student have? List their ids instead of names.", "| student : stuid , lname , fname , age , sex , major , advisor , city_code | has_pet : stuid , petid | pets : petid , pettype , pet_age , weight | has_pet.stuid = student.stuid | has_pet.petid = pets.petid |"],
|
231 |
["Show location and name for all stadiums with a capacity between 5000 and 10000.", "| stadium : stadium_id , location , name , capacity , highest , lowest , average | singer : singer_id , name , country , song_name , song_release_year , age , is_male | concert : concert_id , concert_name , theme , stadium_id , year | singer_in_concert : concert_id , singer_id | concert.stadium_id = stadium.stadium_id | singer_in_concert.singer_id = singer.singer_id | singer_in_concert.concert_id = concert.concert_id |"],
|
232 |
["What are the number of concerts that occurred in the stadium with the largest capacity ?", "| stadium : stadium_id , location , name , capacity , highest , lowest , average | singer : singer_id , name , country , song_name , song_release_year , age , is_male | concert : concert_id , concert_name , theme , stadium_id , year | singer_in_concert : concert_id , singer_id | concert.stadium_id = stadium.stadium_id | singer_in_concert.singer_id = singer.singer_id | singer_in_concert.concert_id = concert.concert_id |"],
|
233 |
+
["Which student has the oldest pet?", "| student : stuid , lname , fname , age , sex , major , advisor , city_code | has_pet : stuid , petid | pets : petid , pettype , pet_age , weight | has_pet.stuid = student.stuid | has_pet.petid = pets.petid |"],
|
234 |
+
["List the names of all singers who performed in a concert with the theme 'Rock'", "| stadium : stadium_id , location , name , capacity , highest , lowest , average | singer : singer_id , name , country , song_name , song_release_year , age , is_male | concert : concert_id , concert_name , theme , stadium_id , year | singer_in_concert : concert_id , singer_id | concert.stadium_id = stadium.stadium_id | singer_in_concert.singer_id = singer.singer_id | singer_in_concert.concert_id = concert.concert_id |"],
|
235 |
+
["List all students who don't have pets.", "| student : stuid , lname , fname , age , sex , major , advisor , city_code | has_pet : stuid , petid | pets : petid , pettype , pet_age , weight | has_pet.stuid = student.stuid | has_pet.petid = pets.petid |"],
|
236 |
+
], inputs=[input_text, db_info, temperature, top_p, top_k, repetition_penalty, format_sql], fn=generate, cache_examples=False, outputs=output_box)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
237 |
|
238 |
|
239 |
+
readme_content = requests.get(f"https://huggingface.co/{merged_model}/raw/main/README.md").text
|
240 |
+
readme_content = re.sub('---.*?---', '', readme_content, flags=re.DOTALL) #Remove YAML front matter
|
241 |
+
|
242 |
+
with gr.Accordion("π Model Readme", open=True):
|
243 |
+
readme = gr.Markdown(
|
244 |
+
readme_content,
|
245 |
+
)
|
246 |
+
|
247 |
+
with gr.Accordion("Disabled Options:", open=False):
|
248 |
+
log = gr.Checkbox(label="Log to Firebase", value=True, interactive=False)
|
249 |
+
|
250 |
+
# When the button is clicked, call the generate function, inputs are taken from the UI elements, outputs are sent to outputs elements
|
251 |
+
run_button.click(fn=generate, inputs=[input_text, db_info, temperature, top_p, top_k, repetition_penalty, format_sql, log, num_return_sequences, num_beams, do_sample], outputs=output_box, api_name="txt2sql")
|
252 |
+
clear_button.add([input_text, db_info, output_box])
|
253 |
+
|
254 |
+
# Firebase code - for rating the generated SQL (remove if you don't want to use Firebase)
|
255 |
+
rate_up.click(fn=log_rating_to_firestore, inputs=[input_text, db_info, temperature, output_box, rate_up])
|
256 |
+
rate_down.click(fn=log_rating_to_firestore, inputs=[input_text, db_info, temperature, output_box, rate_down])
|
257 |
|
258 |
+
demo.queue(concurrency_count=1, max_size=20).launch(debug=True)
|