themissingCRAM commited on
Commit
3dd7783
·
1 Parent(s): 3d07ffa
Files changed (2) hide show
  1. app.py +20 -43
  2. employment.zip +3 -0
app.py CHANGED
@@ -1,15 +1,16 @@
1
  import datetime
2
  import json
3
- import os
4
- import polars as pl
5
  import chromadb
6
  import gradio as gr
 
7
  import spaces
8
  from chromadb.utils import embedding_functions
9
  from dotenv import load_dotenv
10
  from langchain.docstore.document import Document
11
  from langchain.text_splitter import RecursiveCharacterTextSplitter
12
- from smolagents import Tool, CodeAgent, TransformersModel, stream_to_gradio, tool, ToolCallingAgent, HfApiModel
13
  from sqlalchemy import (
14
  create_engine,
15
  MetaData,
@@ -20,15 +21,14 @@ from sqlalchemy import (
20
  insert,
21
  text, Numeric, DateTime, func
22
  )
23
- from huggingface_hub import login
24
- from Constants import BAKERY_ORDERS_DATA, BAKING_RECIPES, RAG_QUESTION, SQL_QUERY
25
  from transformers import pipeline
26
- import time
 
 
27
  load_dotenv()
28
  # os.system("python -m phoenix.server.main serve")
29
  import numpy as np
30
  import os
31
- import base64
32
 
33
 
34
  # # Get your own keys from https://cloud.langfuse.com
@@ -278,26 +278,10 @@ if __name__ == "__main__":
278
  return "", []
279
 
280
 
281
- transcriber = pipeline("automatic-speech-recognition", model="openai/whisper-large-v3-turbo",device_map = 'cuda')
282
-
283
-
284
- #
285
- # def transcribe(audio):
286
- # sr, y = audio
287
- #
288
- # # Convert to mono if stereo
289
- # if y.ndim > 1:
290
- # y = y.mean(axis=1)
291
- #
292
- # y = y.astype(np.float32)
293
- # y /= np.max(np.abs(y))
294
- #
295
- # return transcriber({"sampling_rate": sr, "raw": y})["text"]
296
  @spaces.GPU
297
- def transcribe(audio_chunk):
298
  if audio_chunk is None:
299
  return ""
300
- print('audio_chunk',audio_chunk)
301
  sr, y = audio_chunk
302
 
303
  # Convert to mono if stereo
@@ -308,28 +292,18 @@ if __name__ == "__main__":
308
  y /= np.max(np.abs(y))
309
 
310
  time.sleep(2)
311
- text2=transcriber({"sampling_rate": sr, "raw": y})["text"]
312
- return text2
313
- # sr, y = audio_chunk
314
- # # Convert to mono if stereo
315
- # if y.ndim > 1:
316
- # y = y.mean(axis=1)
317
- #
318
- # y = y.astype(np.float32)
319
- # y /= np.max(np.abs(y))
320
- #
321
- # if stream is not None or stream.shape != []:
322
- # stream = np.concatenate([stream, y])
323
- # else:
324
- # stream = y
325
- # return stream, transcriber({"sampling_rate": sr, "raw": stream})["text"]
326
 
 
 
 
 
 
327
 
328
  with gr.Blocks() as b:
329
  # GUI
330
  gr.Markdown("# Bakery shope ordering llm multi agent system")
331
  with gr.Accordion('''
332
- open for more description of this space ''',open=False):
333
  gr.Markdown('''
334
  with self correcting text2sql agent for orders and RAG agent for recipes.
335
  using smolagents, gradio, HF Spaces, sqlalchemy,langchain for sematic search, chromadb\n
@@ -342,7 +316,8 @@ if __name__ == "__main__":
342
  with gr.Accordion("Bakery orders data", open=False):
343
  gr.DataFrame(pl.DataFrame(BAKERY_ORDERS_DATA))
344
  with gr.Accordion("Baking recipes data", open=False):
345
- gr.DataFrame(pl.DataFrame(BAKING_RECIPES,schema=['baking recipes']))
 
346
  chatbot = gr.Chatbot(type="messages", height=900)
347
  message_box = gr.Textbox(lines=1, label="chat message:")
348
  with gr.Row():
@@ -363,7 +338,9 @@ if __name__ == "__main__":
363
  mbox_submit_event = message_box.submit(enter_message,
364
  [message_box, chatbot],
365
  [message_box, chatbot])
366
- audio_stream = audio_interface.change(transcribe, inputs=[audio_interface],
 
 
367
  outputs=[message_box
368
  ])
369
  rag_q_click_event = rag_q_button.click(enter_message,
@@ -383,4 +360,4 @@ if __name__ == "__main__":
383
  cancels=[reply_button_click_event, rag_q_click_event, sql_q_click_event, combi_click_event,
384
  mbox_submit_event, audio_stream])
385
 
386
- b.launch()
 
1
  import datetime
2
  import json
3
+ import time
4
+
5
  import chromadb
6
  import gradio as gr
7
+ import polars as pl
8
  import spaces
9
  from chromadb.utils import embedding_functions
10
  from dotenv import load_dotenv
11
  from langchain.docstore.document import Document
12
  from langchain.text_splitter import RecursiveCharacterTextSplitter
13
+ from smolagents import Tool, CodeAgent, stream_to_gradio, tool, ToolCallingAgent, HfApiModel
14
  from sqlalchemy import (
15
  create_engine,
16
  MetaData,
 
21
  insert,
22
  text, Numeric, DateTime, func
23
  )
 
 
24
  from transformers import pipeline
25
+
26
+ from Constants import BAKERY_ORDERS_DATA, BAKING_RECIPES, RAG_QUESTION, SQL_QUERY
27
+
28
  load_dotenv()
29
  # os.system("python -m phoenix.server.main serve")
30
  import numpy as np
31
  import os
 
32
 
33
 
34
  # # Get your own keys from https://cloud.langfuse.com
 
278
  return "", []
279
 
280
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
281
  @spaces.GPU
282
+ def transcribe(audio_chunk, message):
283
  if audio_chunk is None:
284
  return ""
 
285
  sr, y = audio_chunk
286
 
287
  # Convert to mono if stereo
 
292
  y /= np.max(np.abs(y))
293
 
294
  time.sleep(2)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
295
 
296
+ message = message + " " + transcriber({"sampling_rate": sr, "raw": y})["text"]
297
+ return message
298
+
299
+
300
+ transcriber = pipeline("automatic-speech-recognition", model="openai/whisper-large-v3-turbo", device_map='cuda')
301
 
302
  with gr.Blocks() as b:
303
  # GUI
304
  gr.Markdown("# Bakery shope ordering llm multi agent system")
305
  with gr.Accordion('''
306
+ open for more description of this space ''', open=False):
307
  gr.Markdown('''
308
  with self correcting text2sql agent for orders and RAG agent for recipes.
309
  using smolagents, gradio, HF Spaces, sqlalchemy,langchain for sematic search, chromadb\n
 
316
  with gr.Accordion("Bakery orders data", open=False):
317
  gr.DataFrame(pl.DataFrame(BAKERY_ORDERS_DATA))
318
  with gr.Accordion("Baking recipes data", open=False):
319
+ gr.DataFrame(pl.DataFrame(BAKING_RECIPES, schema=['baking recipes']))
320
+
321
  chatbot = gr.Chatbot(type="messages", height=900)
322
  message_box = gr.Textbox(lines=1, label="chat message:")
323
  with gr.Row():
 
338
  mbox_submit_event = message_box.submit(enter_message,
339
  [message_box, chatbot],
340
  [message_box, chatbot])
341
+
342
+ audio_stream = audio_interface.change(transcribe, inputs=[audio_interface, message_box
343
+ ],
344
  outputs=[message_box
345
  ])
346
  rag_q_click_event = rag_q_button.click(enter_message,
 
360
  cancels=[reply_button_click_event, rag_q_click_event, sql_q_click_event, combi_click_event,
361
  mbox_submit_event, audio_stream])
362
 
363
+ b.launch()
employment.zip ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:adc3e8999af8e4c83332f62a68b949b75135e4a3c0eb482722d306416bae45f5
3
+ size 478921