Stéphane G commited on
Commit
da00ae1
·
1 Parent(s): ef46e30

refactor + add instruct embedding

Browse files
Dockerfile CHANGED
@@ -8,7 +8,7 @@ RUN python3 -m pip install --no-cache-dir --upgrade -r /code/requirements.txt
8
 
9
  COPY . .
10
 
11
- CMD ["panel", "serve", "/code/code_qa_app.py", "--address", "0.0.0.0", "--port", "7860", "--allow-websocket-origin", "spqw-org-code-qa.hf.space", "--allow-websocket-origin", "0.0.0.0:7860"]
12
 
13
  RUN mkdir /.cache
14
  RUN chmod 777 /.cache
 
8
 
9
  COPY . .
10
 
11
+ CMD ["panel", "serve", "/code/app.py", "--address", "0.0.0.0", "--port", "7860", "--allow-websocket-origin", "spqw-org-code-qa.hf.space", "--allow-websocket-origin", "0.0.0.0:7860"]
12
 
13
  RUN mkdir /.cache
14
  RUN chmod 777 /.cache
app.py ADDED
@@ -0,0 +1,325 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # This app allows users to ask questions about a codebase.
2
+ # It uses:
3
+ # - Panel for the UI
4
+ # - Langchain for the chain
5
+ # - Chroma for the vectorstore
6
+
7
+ # For local dev env : install the following packages
8
+ #!pip install langchain openai chromadb panel gitpython
9
+
10
+ import os
11
+ import panel as pn
12
+ from typing import List
13
+ from urllib.parse import urlparse
14
+
15
+ from langchain.chains import RetrievalQA
16
+ from langchain.llms import OpenAI
17
+ from langchain.document_loaders import GitLoader
18
+ from langchain.indexes import VectorstoreIndexCreator
19
+ from langchain.text_splitter import RecursiveCharacterTextSplitter
20
+ from langchain.embeddings import OpenAIEmbeddings
21
+ from langchain.embeddings import HuggingFaceInstructEmbeddings
22
+ from langchain.vectorstores import Chroma
23
+ from langchain.chat_models import ChatOpenAI
24
+ from langchain.vectorstores import Chroma
25
+ from langchain.docstore.document import Document
26
+ from langchain.schema import BaseRetriever
27
+
28
+ #------------------------------------------------------------------------------
29
+ # Utils
30
+
31
+ def get_repo_name_from_url(url):
32
+ url_parsed = urlparse(url)
33
+ return os.path.basename(url_parsed.path).replace(".git", "")
34
+
35
+ def get_names_subdirs(path_dir):
36
+ return [f.name for f in os.scandir(path_dir) if f.is_dir()]
37
+
38
+ def remove_empty_subdirs(path_dir):
39
+ for root, dirs, files in os.walk(path_dir, topdown=False):
40
+ for name in dirs:
41
+ dir_path = os.path.join(root, name)
42
+ if len(os.listdir(dir_path)) == 0:
43
+ os.rmdir(dir_path)
44
+
45
+ def get_names_repos_already_checkout():
46
+ path_checkout = "checkout"
47
+ remove_empty_subdirs(path_checkout)
48
+ return get_names_subdirs(path_checkout)
49
+
50
+ #------------------------------------------------------------------------------
51
+ # Panel setup
52
+ pn.extension(
53
+ 'texteditor',
54
+ template = "bootstrap",
55
+ sizing_mode = 'stretch_width'
56
+ )
57
+
58
+ pn.state.template.param.update(
59
+ main_max_width = "690px",
60
+ header_background = "#F08080",
61
+ title = "Squirrel : Semantic code search and question answering"
62
+ )
63
+
64
+ #------------------------------------------------------------------------------
65
+ # User inputs
66
+
67
+ input_key_openai = pn.widgets.PasswordInput(
68
+ value = "",
69
+ placeholder = "Enter your OpenAI API Key here...",
70
+ width = 300
71
+ )
72
+
73
+ select_repo = pn.widgets.RadioBoxGroup(
74
+ options = get_names_repos_already_checkout() + ["Custom"],
75
+ inline = False
76
+ )
77
+
78
+ input_url_repo = pn.widgets.TextInput(
79
+ value = "",
80
+ placeholder = "Enter your GitHub repo URL here...",
81
+ width = 300
82
+ )
83
+
84
+ input_branch_repo = pn.widgets.TextInput(
85
+ value="",
86
+ placeholder="Enter the GitHub repo branch here...",
87
+ width=300
88
+ )
89
+
90
+ select_embedding = pn.widgets.RadioBoxGroup(
91
+ options=['openai', 'instruct'],
92
+ inline=False
93
+ )
94
+
95
+ input_question = pn.widgets.TextEditor(
96
+ value="",
97
+ placeholder="Enter your questions here...",
98
+ height=200,
99
+ toolbar=False
100
+ )
101
+
102
+ run_button = pn.widgets.Button(
103
+ name="Run",
104
+ button_type="primary"
105
+ )
106
+
107
+ select_k = pn.widgets.IntSlider(
108
+ name="Number of relevant chunks",
109
+ start=1,
110
+ end=10,
111
+ step=1,
112
+ value=3
113
+ )
114
+
115
+ select_chain_type = pn.widgets.RadioButtonGroup(
116
+ name='Chain type',
117
+ options=['stuff', 'map_reduce', "refine", "map_rerank"]
118
+ )
119
+
120
+ def toggle_github_inputs(event):
121
+ if event.new == "Custom":
122
+ input_url_repo.visible = True
123
+ input_branch_repo.visible = True
124
+ else:
125
+ input_url_repo.visible = False
126
+ input_branch_repo.visible = False
127
+
128
+ select_repo.param.watch(toggle_github_inputs, 'value')
129
+
130
+
131
+ #------------------------------------------------------------------------------
132
+ def init_embedding_openai():
133
+ return OpenAIEmbeddings(
134
+ model = "text-embedding-ada-002",
135
+ openai_api_key=input_key_openai.value
136
+ )
137
+
138
+ def init_embedding_instruct():
139
+ # Default model: instructor-large
140
+ return HuggingFaceInstructEmbeddings(
141
+ query_instruction="Represent the query for retrival:"
142
+ )
143
+
144
+ embedding_functions = {
145
+ "openai": init_embedding_openai,
146
+ "instruct": init_embedding_instruct
147
+ }
148
+
149
+ result_panel_objects = []
150
+
151
+ def load_repo(name_repo):
152
+ print(f"Name repo: {name_repo}")
153
+ path_checkout = os.path.join("checkout", name_repo)
154
+ print(f"Path checkout: {path_checkout}")
155
+ is_repo_already_checkout = os.path.exists(path_checkout) and len(os.listdir(path_checkout)) > 0
156
+ print(f"Is repo already checkout: {is_repo_already_checkout}")
157
+ print(f"Repo URL: {input_url_repo.value}")
158
+ print(f"Branch: {input_branch_repo.value}")
159
+
160
+ file_filter = lambda file_path: file_path.endswith(".swift") or file_path.endswith(".h") or file_path.endswith(".m")
161
+
162
+ if is_repo_already_checkout:
163
+ print("Repo already checkout")
164
+ print("Loading documents...")
165
+ loader = GitLoader(
166
+ repo_path=path_checkout,
167
+ branch=input_branch_repo.value,
168
+ file_filter=file_filter
169
+ )
170
+ else:
171
+ print("Repo not checkout")
172
+ print("Cloning repo and loading documents...")
173
+ loader = GitLoader(
174
+ repo_path=path_checkout,
175
+ clone_url=input_url_repo.value,
176
+ branch=input_branch_repo.value,
177
+ file_filter=file_filter
178
+ )
179
+
180
+ documents = loader.load()
181
+ return documents
182
+
183
+ def get_result_qa_chain():
184
+ os.environ["OPENAI_API_KEY"] = input_key_openai.value
185
+ if select_repo.value == "Custom":
186
+ name_repo = get_repo_name_from_url(input_url_repo.value)
187
+ else:
188
+ name_repo = select_repo.value
189
+ name_embedding = select_embedding.value
190
+ embedding_function = embedding_functions[name_embedding]()
191
+ path_dir_db = f"dbs/{name_repo}/{name_embedding}"
192
+ is_repo_already_indexed = os.path.exists(path_dir_db)
193
+
194
+ result_panel_objects.clear()
195
+
196
+ if not is_repo_already_indexed:
197
+ print("Repo not indexed. Starting indexing...")
198
+ documents = load_repo(name_repo)
199
+ texts = RecursiveCharacterTextSplitter(chunk_size=3000, chunk_overlap=0).split_documents(documents)
200
+ db = Chroma.from_documents(
201
+ documents=texts,
202
+ embedding=embedding_function,
203
+ persist_directory=path_dir_db
204
+ )
205
+ else:
206
+ print("Repo already indexed")
207
+ db = Chroma(
208
+ embedding_function=embedding_function,
209
+ persist_directory=path_dir_db,
210
+ )
211
+ db.persist()
212
+
213
+ retriever = db.as_retriever(
214
+ search_type = "similarity",
215
+ search_kwargs = { "k": select_k.value }
216
+ )
217
+
218
+ # llm=OpenAI(temperature=0, model_name="text-davinci-002")
219
+ llm = ChatOpenAI(
220
+ model_name = "gpt-3.5-turbo",
221
+ temperature=0.7,
222
+ max_tokens=1000)
223
+
224
+ # create a chain to answer questions
225
+ qa = RetrievalQA.from_chain_type(
226
+ llm=llm,
227
+ chain_type=select_chain_type.value,
228
+ retriever=retriever,
229
+ return_source_documents=True
230
+ )
231
+
232
+ result = qa({"query": input_question.value})
233
+ return result
234
+
235
+ def result_box(_):
236
+ if input_question.value:
237
+ result = get_result_qa_chain()
238
+
239
+ result_panel_objects.extend([
240
+ pn.Row(
241
+ pn.panel("\U0001F60A", width=10),
242
+ input_question.value
243
+ ),
244
+ pn.Row(
245
+ pn.panel("\U0001F916", width=10),
246
+ pn.Column(
247
+ result["result"],
248
+ "<hr>",
249
+ pn.pane.Markdown(f"### Retrieved sources\n"),
250
+ pn.Column(
251
+ *[pn.pane.Markdown(f"#####{i+1}\n```\n{doc.page_content}\n```", width=600) for i, doc in enumerate(result["source_documents"])]
252
+ )
253
+ )
254
+ )
255
+ ])
256
+
257
+
258
+ return pn.Column(
259
+ *result_panel_objects,
260
+ min_height=400,
261
+ min_width=600,
262
+ )
263
+
264
+ # -----------------------------------------------------------------------------
265
+ # Layout
266
+ config_section = pn.Column(
267
+ pn.Column(
268
+ input_key_openai
269
+ ),
270
+ pn.Card(
271
+ pn.Column(
272
+ select_repo,
273
+ input_url_repo,
274
+ input_branch_repo
275
+ ),
276
+ title="Load codebase"
277
+ ),
278
+ pn.Card(
279
+ pn.Column(
280
+ select_embedding
281
+ ),
282
+ title="Select embedding"
283
+ ),
284
+ pn.Card(
285
+ "Chain type:",
286
+ pn.Column(
287
+ select_chain_type,
288
+ select_k
289
+ ),
290
+ title="Advanced settings",
291
+ collapsed=True
292
+ )
293
+ )
294
+
295
+ input_section = pn.Card(
296
+ pn.Row(input_question),
297
+ run_button,
298
+ title="Ask a question"
299
+ )
300
+
301
+ # show time elapsed
302
+ output_section = pn.Card(
303
+ pn.panel(
304
+ pn.bind(
305
+ result_box,
306
+ run_button
307
+ ),
308
+ loading_indicator=True,
309
+ ),
310
+ title="Output"
311
+ )
312
+
313
+ pn.Column(
314
+ pn.pane.Markdown("""
315
+ ### \U0001F60A Ask Squirrel a question about your codebase
316
+
317
+ 1. Enter the url of a github repo and a branch (swift / objc)
318
+ 2. Enter OpenAI API key. This costs $. Set up billing at [OpenAI](https://platform.openai.com/account).
319
+ 3. Type a question and click "Run".
320
+
321
+ """),
322
+ config_section,
323
+ input_section,
324
+ output_section
325
+ ).servable()
checkout/BusyNavigationBar DELETED
@@ -1 +0,0 @@
1
- Subproject commit 48059ab3bbf4159d71e8ce69c62c51ae284303c7
 
 
checkout/tuist DELETED
@@ -1 +0,0 @@
1
- Subproject commit 099e3b79347a70822dcb3d10138f50787c4fd0e9
 
 
code_qa_app.py DELETED
@@ -1,272 +0,0 @@
1
- #!pip install langchain openai chromadb tiktoken panel gitpython
2
-
3
- import os
4
- import panel as pn
5
- from typing import List
6
- from urllib.parse import urlparse
7
-
8
- from langchain.chains import RetrievalQA
9
- from langchain.llms import OpenAI
10
- from langchain.document_loaders import GitLoader
11
- from langchain.indexes import VectorstoreIndexCreator
12
- from langchain.text_splitter import RecursiveCharacterTextSplitter
13
- from langchain.embeddings import OpenAIEmbeddings
14
- from langchain.vectorstores import Chroma
15
- from langchain.chat_models import ChatOpenAI
16
- from langchain.vectorstores import Chroma
17
- from langchain.docstore.document import Document
18
- from langchain.schema import BaseRetriever
19
-
20
- # UI configuration
21
- pn.extension(
22
- 'texteditor',
23
- template="bootstrap",
24
- sizing_mode='stretch_width'
25
- )
26
-
27
- pn.state.template.param.update(
28
- main_max_width="690px",
29
- header_background="#F08080",
30
- )
31
-
32
- # User inputs
33
- use_cached_repo = pn.widgets.Checkbox(name="Use cached Alamofire repo instead")
34
-
35
- url_repo = pn.widgets.TextInput(
36
- value="",
37
- placeholder="Enter your GitHub repo URL here...",
38
- width=300
39
- )
40
-
41
- branch_repo = pn.widgets.TextInput(
42
- value="",
43
- placeholder="Enter the GitHub repo branch here...",
44
- width=300
45
- )
46
-
47
- openaikey = pn.widgets.PasswordInput(
48
- value="",
49
- placeholder="Enter your OpenAI API Key here...",
50
- width=300
51
- )
52
-
53
- context_texteditor = pn.widgets.TextEditor(
54
- value="",
55
- placeholder="(Optional) Enter your index query here...",
56
- height=200,
57
- toolbar=False
58
- )
59
-
60
- prompt_texteditor = pn.widgets.TextEditor(
61
- value="",
62
- placeholder="Enter your questions here...",
63
- height=200,
64
- toolbar=False
65
- )
66
-
67
- run_button = pn.widgets.Button(name="Run!")
68
-
69
- select_k = pn.widgets.IntSlider(
70
- name="Number of relevant chunks",
71
- start=1,
72
- end=5,
73
- step=1,
74
- value=2
75
- )
76
-
77
- select_chain_type = pn.widgets.RadioButtonGroup(
78
- name='Chain type',
79
- options=['stuff', 'map_reduce', "refine", "map_rerank"]
80
- )
81
-
82
- class FixedRetriever(BaseRetriever):
83
- def __init__(self, documents: List[Document]):
84
- self.documents = documents
85
-
86
- def get_relevant_documents(self, query: str):
87
- # Ignoring query
88
- return self.documents
89
-
90
- async def aget_relevant_documents(self, query: str) -> List[Document]:
91
- # Ignoring query
92
- return self.documents
93
-
94
-
95
- # Question answering function
96
- def qa_fetching_repo(url_repo, branch, context_query, prompt_query, chain_type, k):
97
- print("* QA fetching repo")
98
- # Extract the repo name from the URL
99
- url_parsed = urlparse(url_repo)
100
- name_repo = os.path.basename(url_parsed.path).replace(".git", "")
101
-
102
- # Check if the repo checkout folder exists and create it if not
103
- path_repo = os.path.join("checkout", name_repo)
104
- is_repo_already_checkout = False
105
- if os.path.exists(path_repo):
106
- is_repo_already_checkout = True
107
- else:
108
- os.makedirs(path_repo)
109
-
110
- if is_repo_already_checkout:
111
- loader = GitLoader(
112
- repo_path=path_repo,
113
- branch=branch,
114
- file_filter=lambda file_path: file_path.endswith(".swift") or file_path.endswith(".h") or file_path.endswith(".m")
115
- )
116
- else:
117
- loader = GitLoader(
118
- clone_url=url_repo,
119
- repo_path=path_repo,
120
- branch=branch,
121
- file_filter=lambda file_path: file_path.endswith(".swift") or file_path.endswith(".h") or file_path.endswith(".m")
122
- )
123
-
124
- documents = loader.load()
125
-
126
- # split documents into chunks
127
- text_splitter = RecursiveCharacterTextSplitter(chunk_size=3000, chunk_overlap=0)
128
- texts = text_splitter.split_documents(documents)
129
-
130
- # select which embeddings to use
131
- embeddings = OpenAIEmbeddings()
132
-
133
- persist_directory = os.path.join("dbs", name_repo)
134
-
135
- # create the vectorstore to use as the index
136
- db = Chroma.from_documents(texts, embeddings, persist_directory=persist_directory)
137
- db.persist()
138
-
139
- return qa(db, context_query, prompt_query, chain_type, k)
140
-
141
- # Question answering function
142
- def qa_cached_tuist(context_query, prompt_query, chain_type, k):
143
-
144
- print("* QA cached Tuist")
145
- db = Chroma(persist_directory="dbs/Alamofire/", embedding_function=OpenAIEmbeddings())
146
- # db = Chroma(persist_directory="data/tuist_cached/", embedding_function=OpenAIEmbeddings())
147
-
148
- return qa(db, context_query, prompt_query, chain_type, k)
149
-
150
- def qa(db, context_query, prompt_query, chain_type, k):
151
-
152
- if context_query.strip():
153
- print("* Using FixedRetriver")
154
- docs = db.similarity_search(context_query, k=k)
155
- retriever = FixedRetriever(docs)
156
- # You can test how an empty retriever makes our LLM ignorant about the context
157
- # retriever = FixedRetriever([])
158
-
159
- else:
160
- print("* Using db as retriever")
161
- # expose this index in a retriever interface
162
- retriever = db.as_retriever(search_type="similarity", search_kwargs={"k": k})
163
-
164
- # llm=OpenAI(temperature=0, model_name="text-davinci-002")
165
- llm = ChatOpenAI(
166
- model_name = "gpt-3.5-turbo",
167
- temperature=0.7,
168
- max_tokens=1000)
169
-
170
- # create a chain to answer questions
171
- qa = RetrievalQA.from_chain_type(
172
- llm=llm,
173
- chain_type=chain_type,
174
- retriever=retriever,
175
- return_source_documents=True
176
- )
177
-
178
- result = qa({"query": prompt_query})
179
- print(result['result'])
180
- return result
181
-
182
- convos = [] # store all panel objects in a list
183
-
184
- def qa_result(_):
185
- os.environ["OPENAI_API_KEY"] = openaikey.value
186
-
187
- prompt_text = prompt_texteditor.value
188
- context_text = context_texteditor.value
189
- if prompt_text:
190
-
191
- convos.clear()
192
-
193
- if use_cached_repo.value:
194
- result = qa_cached_tuist(context_query=context_text, prompt_query=prompt_text, chain_type=select_chain_type.value, k=select_k.value)
195
- else:
196
- result = qa_fetching_repo(url_repo=url_repo.value, branch=branch_repo.value, context_query=context_text, prompt_query=prompt_text, chain_type=select_chain_type.value, k=select_k.value)
197
-
198
- convos.extend([
199
- pn.Row(
200
- pn.panel("\U0001F60A", width=10),
201
- prompt_text,
202
- width=600
203
- ),
204
- pn.Row(
205
- pn.panel("\U0001F916", width=10),
206
- pn.Column(
207
- result["result"],
208
- "From the following retrieved sources:",
209
- pn.pane.Markdown(
210
- '\n<hr>\n'.join(doc.page_content for doc in result["source_documents"])
211
- )
212
- )
213
- )
214
- ])
215
- #return convos
216
- return pn.Column(*convos, margin=15, width=575, min_height=400)
217
-
218
- qa_interactive = pn.panel(
219
- pn.bind(qa_result, run_button),
220
- loading_indicator=True,
221
- )
222
-
223
- # Output
224
- output = pn.WidgetBox('*Output will show up here:*', qa_interactive, width=1000, scroll=True)
225
-
226
- # Layout
227
- github_inputs = pn.Column(use_cached_repo, url_repo, branch_repo, width=600)
228
-
229
- def toggle_github_inputs(event):
230
- if event.new:
231
- url_repo.disabled = True
232
- branch_repo.disabled = True
233
- else:
234
- url_repo.disabled = False
235
- branch_repo.disabled = False
236
-
237
- use_cached_repo.param.watch(toggle_github_inputs, 'value')
238
-
239
- setup = pn.Column(
240
- pn.Column(github_inputs, openaikey),
241
- pn.Card(
242
- "Chain type:",
243
- pn.Column(
244
- select_chain_type,
245
- select_k
246
- ),
247
- title="Advanced settings",
248
- margin=10,
249
- collapsed=True
250
- )
251
- )
252
-
253
- queries = pn.Column(
254
- pn.Row(prompt_texteditor),
255
- # pn.Row(context_texteditor, prompt_texteditor),
256
- run_button,
257
- margin=5)
258
-
259
- pn.Column(
260
- pn.pane.Markdown("""
261
- ### \U0001F60A Ask Squirrel a question about your codebase
262
-
263
- 1. Enter the url of a github repo and a branch (swift / objc)
264
- 2. Enter OpenAI API key. This costs $. Set up billing at [OpenAI](https://platform.openai.com/account).
265
- 3. Type a question and click "Run".
266
-
267
- """),
268
- setup,
269
- queries,
270
- output,
271
-
272
- ).servable()
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
dbs/Alamofire/chroma-collections.parquet DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:ecbe4963d18e41866d6702e52613e682bb31f25ebb4033b8fc9dcc4b2a6f63cb
3
- size 557
 
 
 
 
dbs/{tuist_cached → Alamofire/openai}/chroma-collections.parquet RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:5e8b4e9da5fb81ea3bd50fd92295de6849d7fea71cd4a75886e02c048900f271
3
  size 557
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d0589b2a7968096a032b7df9b36d957d867fa8fd54131d632a99fd3ac295608f
3
  size 557
dbs/Alamofire/{chroma-embeddings.parquet → openai/chroma-embeddings.parquet} RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:49b4d111a064634d749d0932a5c8930c09df439179d6eb16a65b5bd57658c991
3
- size 7730171
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0c3c8b34bd40f26227017233845625c47864e4b3740e1750bddb74849e5749a5
3
+ size 7722000
dbs/Alamofire/{index/id_to_uuid_82e51ffe-b003-42db-a060-7ee52745f81d.pkl → openai/index/id_to_uuid_0fc9eee1-a3d5-4c9c-9ed1-059a474e69f0.pkl} RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ea43b34fb9f087fd7e508b4e9ebfe6f93aa9ad59b9d2173e6e746043dc86245f
3
- size 19458
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe5335409f03267d3f9edf0f1c94ff49f697dd9ca04832c056cd991ef4a77a3c
3
+ size 19442
dbs/Alamofire/{index/index_82e51ffe-b003-42db-a060-7ee52745f81d.bin → openai/index/index_0fc9eee1-a3d5-4c9c-9ed1-059a474e69f0.bin} RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:63c71563dd9aa8dc6443bf20edcd657bcefd8a4aecdea4613a91d8866308d653
3
  size 3807056
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0e196b2b00392a1f9fa59943e7cc8cff332abb5690026225e16575895010c929
3
  size 3807056
dbs/Alamofire/{index/index_metadata_82e51ffe-b003-42db-a060-7ee52745f81d.pkl → openai/index/index_metadata_0fc9eee1-a3d5-4c9c-9ed1-059a474e69f0.pkl} RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9df5de400e56d4b0c6219e4d17c13166ef51dd2ca574f48cb161a4c7f8d6b30b
3
  size 74
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:555e4f40a0c1b6b3fdc644d2bf2084f17510b62165d6d1a4c3b1a2dce60effc2
3
  size 74
dbs/Alamofire/{index/uuid_to_id_82e51ffe-b003-42db-a060-7ee52745f81d.pkl → openai/index/uuid_to_id_0fc9eee1-a3d5-4c9c-9ed1-059a474e69f0.pkl} RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ec54074411da3ad2c668c0362ad30c7cca1710832ea3c9423687fc176d68be66
3
  size 22751
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c790ac7824f6efade0078a5b7b0902ced62be81de74fe32f21b458a212773400
3
  size 22751
dbs/tuist_cached/chroma-embeddings.parquet DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:db53078400ba87eba743c80845b7fae4b53b583a2619b950569d0260c8ac3cb1
3
- size 88813345
 
 
 
 
dbs/tuist_cached/index/id_to_uuid_dbb9dc64-2123-4c5d-a69d-a08a9d0b3485.pkl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:cfb2122fac683f50360e99bda40bd42f7da8783c3dc1e7191b847b1514a12131
3
- size 234446
 
 
 
 
dbs/tuist_cached/index/index_dbb9dc64-2123-4c5d-a69d-a08a9d0b3485.bin DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:79e9ad6271540c9e91126aaffb2941ca5e29749eb79df26f9fc11fce7e5d1602
3
- size 45434068
 
 
 
 
dbs/tuist_cached/index/index_metadata_dbb9dc64-2123-4c5d-a69d-a08a9d0b3485.pkl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:22ff307c21335897bc15d472a59a7f6c77f977d66eda2a6271c9f0e0f5857c5d
3
- size 74
 
 
 
 
dbs/tuist_cached/index/uuid_to_id_dbb9dc64-2123-4c5d-a69d-a08a9d0b3485.pkl DELETED
@@ -1,3 +0,0 @@
1
- version https://git-lfs.github.com/spec/v1
2
- oid sha256:e1e04090d384ae09a2c1f6d954363767aafd2a6d36d07dacf9fd93b274836642
3
- size 274171
 
 
 
 
requirements.txt CHANGED
@@ -5,4 +5,5 @@ pypdf
5
  tiktoken
6
  panel
7
  notebook
8
- gitpython
 
 
5
  tiktoken
6
  panel
7
  notebook
8
+ gitpython
9
+ InstructorEmbedding
scripts/dev_up.sh CHANGED
@@ -5,4 +5,4 @@ set -o pipefail
5
  set -o nounset
6
  set -o xtrace
7
 
8
- panel serve code_qa_app.py --address 0.0.0.0 --port 7860 --allow-websocket-origin 0.0.0.0:7860
 
5
  set -o nounset
6
  set -o xtrace
7
 
8
+ panel serve app.py --address 0.0.0.0 --port 7860 --allow-websocket-origin 0.0.0.0:7860