Spaces:
Running
Running
Merge pull request #3 from almutareb/reference_parser
Browse files- mixtral_agent.py +116 -51
- requirements.txt +183 -0
mixtral_agent.py
CHANGED
@@ -12,9 +12,15 @@ from langchain.agents.format_scratchpad import format_log_to_str
|
|
12 |
from langchain.agents.output_parsers import (
|
13 |
ReActJsonSingleInputOutputParser,
|
14 |
)
|
|
|
|
|
|
|
|
|
|
|
15 |
from langchain.tools.render import render_text_description
|
16 |
import os
|
17 |
|
|
|
18 |
import dotenv
|
19 |
|
20 |
dotenv.load_dotenv()
|
@@ -26,39 +32,92 @@ OLLMA_BASE_URL = os.getenv("OLLMA_BASE_URL")
|
|
26 |
# supports many more optional parameters. Hover on your `ChatOllama(...)`
|
27 |
# class to view the latest available supported parameters
|
28 |
llm = ChatOllama(
|
29 |
-
model="mistral",
|
30 |
base_url= OLLMA_BASE_URL
|
31 |
)
|
32 |
prompt = ChatPromptTemplate.from_template("Tell me a short joke about {topic}")
|
33 |
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
59 |
|
60 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
61 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
62 |
|
63 |
|
64 |
prompt = hub.pull("hwchase17/react-json")
|
@@ -85,33 +144,39 @@ agent_executor = AgentExecutor(
|
|
85 |
agent=agent,
|
86 |
tools=tools,
|
87 |
verbose=True,
|
88 |
-
handle_parsing_errors=True #prevents error
|
89 |
)
|
90 |
|
91 |
-
|
92 |
-
# {
|
93 |
-
# "input": "Who is the current holder of the speed skating world record on 500 meters? What is her current age raised to the 0.43 power?"
|
94 |
-
# }
|
95 |
-
# )
|
96 |
-
|
97 |
-
# agent_executor.invoke(
|
98 |
-
# {
|
99 |
-
# "input": "what are large language models and why are they so expensive to run?"
|
100 |
-
# }
|
101 |
-
# )
|
102 |
|
103 |
-
|
104 |
-
|
105 |
-
#
|
106 |
-
|
107 |
-
# )
|
108 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
109 |
|
110 |
-
agent_executor.invoke(
|
111 |
-
{
|
112 |
-
|
113 |
-
|
114 |
-
|
115 |
-
}
|
116 |
-
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
117 |
|
|
|
12 |
from langchain.agents.output_parsers import (
|
13 |
ReActJsonSingleInputOutputParser,
|
14 |
)
|
15 |
+
# Import things that are needed generically
|
16 |
+
from langchain.pydantic_v1 import BaseModel, Field
|
17 |
+
from langchain.tools import BaseTool, StructuredTool, tool
|
18 |
+
from typing import List, Dict
|
19 |
+
from datetime import datetime
|
20 |
from langchain.tools.render import render_text_description
|
21 |
import os
|
22 |
|
23 |
+
|
24 |
import dotenv
|
25 |
|
26 |
dotenv.load_dotenv()
|
|
|
32 |
# supports many more optional parameters. Hover on your `ChatOllama(...)`
|
33 |
# class to view the latest available supported parameters
|
34 |
llm = ChatOllama(
|
35 |
+
model="mistral:instruct",
|
36 |
base_url= OLLMA_BASE_URL
|
37 |
)
|
38 |
prompt = ChatPromptTemplate.from_template("Tell me a short joke about {topic}")
|
39 |
|
40 |
+
arxiv_retriever = ArxivRetriever(load_max_docs=2)
|
41 |
+
|
42 |
+
|
43 |
+
|
44 |
+
def format_info_list(info_list: List[Dict[str, str]]) -> str:
|
45 |
+
"""
|
46 |
+
Format a list of dictionaries containing information into a single string.
|
47 |
+
|
48 |
+
Args:
|
49 |
+
info_list (List[Dict[str, str]]): A list of dictionaries containing information.
|
50 |
+
|
51 |
+
Returns:
|
52 |
+
str: A formatted string containing the information from the list.
|
53 |
+
"""
|
54 |
+
formatted_strings = []
|
55 |
+
for info_dict in info_list:
|
56 |
+
formatted_string = "|"
|
57 |
+
for key, value in info_dict.items():
|
58 |
+
if isinstance(value, datetime.date):
|
59 |
+
value = value.strftime('%Y-%m-%d')
|
60 |
+
formatted_string += f"'{key}': '{value}', "
|
61 |
+
formatted_string = formatted_string.rstrip(', ') + "|"
|
62 |
+
formatted_strings.append(formatted_string)
|
63 |
+
return '\n'.join(formatted_strings)
|
64 |
+
|
65 |
+
@tool
|
66 |
+
def arxiv_search(query: str) -> str:
|
67 |
+
"""Using the arxiv search and collects metadata."""
|
68 |
+
# return "LangChain"
|
69 |
+
global all_sources
|
70 |
+
data = arxiv_retriever.invoke(query)
|
71 |
+
meta_data = [i.metadata for i in data]
|
72 |
+
# meta_data += all_sources
|
73 |
+
# all_sources += meta_data
|
74 |
+
all_sources += meta_data
|
75 |
+
|
76 |
+
# formatted_info = format_info(entry_id, published, title, authors)
|
77 |
+
|
78 |
+
# formatted_info = format_info_list(all_sources)
|
79 |
+
|
80 |
+
return meta_data.__str__()
|
81 |
+
|
82 |
+
@tool
|
83 |
+
def google_search(query: str) -> str:
|
84 |
+
"""Using the google search and collects metadata."""
|
85 |
+
# return "LangChain"
|
86 |
+
global all_sources
|
87 |
+
|
88 |
+
x = SerpAPIWrapper()
|
89 |
+
search_results:dict = x.results(query)
|
90 |
+
|
91 |
+
|
92 |
+
organic_source = search_results['organic_results']
|
93 |
+
# formatted_string = "Title: {title}, link: {link}, snippet: {snippet}".format(**organic_source)
|
94 |
+
cleaner_sources = ["Title: {title}, link: {link}, snippet: {snippet}".format(**i) for i in organic_source]
|
95 |
+
|
96 |
+
all_sources += cleaner_sources
|
97 |
+
|
98 |
+
return cleaner_sources.__str__()
|
99 |
+
# return organic_source
|
100 |
+
|
101 |
+
|
102 |
+
|
103 |
+
|
104 |
+
tools = [arxiv_search,google_search]
|
105 |
|
106 |
+
# tools = [
|
107 |
+
# create_retriever_tool(
|
108 |
+
# retriever,
|
109 |
+
# "search arxiv's database for",
|
110 |
+
# "Use this to recomend the user a paper to read Unless stated please choose the most recent models",
|
111 |
+
# # "Searches and returns excerpts from the 2022 State of the Union.",
|
112 |
+
# ),
|
113 |
|
114 |
+
# Tool(
|
115 |
+
# name="SerpAPI",
|
116 |
+
# description="A low-cost Google Search API. Useful for when you need to answer questions about current events. Input should be a search query.",
|
117 |
+
# func=SerpAPIWrapper().run,
|
118 |
+
# )
|
119 |
+
|
120 |
+
# ]
|
121 |
|
122 |
|
123 |
prompt = hub.pull("hwchase17/react-json")
|
|
|
144 |
agent=agent,
|
145 |
tools=tools,
|
146 |
verbose=True,
|
147 |
+
# handle_parsing_errors=True #prevents error
|
148 |
)
|
149 |
|
150 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
151 |
|
152 |
+
if __name__ == "__main__":
|
153 |
+
|
154 |
+
# global variable for collecting sources
|
155 |
+
all_sources = []
|
|
|
156 |
|
157 |
+
input = agent_executor.invoke(
|
158 |
+
{
|
159 |
+
"input": "How to generate videos from images using state of the art macchine learning models; Using the axriv retriever " +
|
160 |
+
"add the urls of the papers used in the final answer using the metadata from the retriever please do not use '`' "
|
161 |
+
# f"Please prioritize the newest papers this is the current data {get_current_date()}"
|
162 |
+
}
|
163 |
+
)
|
164 |
|
165 |
+
# input_1 = agent_executor.invoke(
|
166 |
+
# {
|
167 |
+
# "input": "I am looking for a text to 3d model; Using the axriv retriever " +
|
168 |
+
# "add the urls of the papers used in the final answer using the metadata from the retriever"
|
169 |
+
# # f"Please prioritize the newest papers this is the current data {get_current_date()}"
|
170 |
+
# }
|
171 |
+
# )
|
172 |
+
|
173 |
+
# input_1 = agent_executor.invoke(
|
174 |
+
# {
|
175 |
+
# "input": "I am looking for a text to 3d model; Using the google search tool " +
|
176 |
+
# "add the urls in the final answer using the metadata from the retriever, also provid a summary of the searches"
|
177 |
+
# # f"Please prioritize the newest papers this is the current data {get_current_date()}"
|
178 |
+
# }
|
179 |
+
# )
|
180 |
+
|
181 |
+
x = 0
|
182 |
|
requirements.txt
CHANGED
@@ -189,3 +189,186 @@ websockets==11.0.3
|
|
189 |
wrapt==1.16.0
|
190 |
yarl==1.9.4
|
191 |
zipp==3.17.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
189 |
wrapt==1.16.0
|
190 |
yarl==1.9.4
|
191 |
zipp==3.17.0
|
192 |
+
aiofiles==23.2.1
|
193 |
+
aiohttp==3.9.3
|
194 |
+
aiosignal==1.3.1
|
195 |
+
altair==5.2.0
|
196 |
+
annotated-types==0.6.0
|
197 |
+
anyio==4.2.0
|
198 |
+
arxiv==2.1.0
|
199 |
+
asgiref==3.7.2
|
200 |
+
async-timeout==4.0.3
|
201 |
+
attrs==23.2.0
|
202 |
+
backoff==2.2.1
|
203 |
+
bcrypt==4.1.2
|
204 |
+
beautifulsoup4==4.12.3
|
205 |
+
boto3==1.34.42
|
206 |
+
botocore==1.34.42
|
207 |
+
build==1.0.3
|
208 |
+
cachetools==5.3.2
|
209 |
+
certifi==2024.2.2
|
210 |
+
chardet==5.2.0
|
211 |
+
charset-normalizer==3.3.2
|
212 |
+
chroma-hnswlib==0.7.3
|
213 |
+
chromadb==0.4.22
|
214 |
+
click==8.1.7
|
215 |
+
coloredlogs==15.0.1
|
216 |
+
contourpy==1.2.0
|
217 |
+
cycler==0.12.1
|
218 |
+
dataclasses-json==0.6.4
|
219 |
+
dataclasses-json-speakeasy==0.5.11
|
220 |
+
Deprecated==1.2.14
|
221 |
+
emoji==2.10.1
|
222 |
+
exceptiongroup==1.2.0
|
223 |
+
faiss-cpu==1.7.4
|
224 |
+
fastapi==0.109.2
|
225 |
+
feedparser==6.0.10
|
226 |
+
ffmpy==0.3.2
|
227 |
+
filelock==3.13.1
|
228 |
+
filetype==1.2.0
|
229 |
+
flatbuffers==23.5.26
|
230 |
+
fonttools==4.48.1
|
231 |
+
frozenlist==1.4.1
|
232 |
+
fsspec==2024.2.0
|
233 |
+
gitdb==4.0.11
|
234 |
+
GitPython==3.1.41
|
235 |
+
google-auth==2.27.0
|
236 |
+
google_search_results==2.4.2
|
237 |
+
googleapis-common-protos==1.62.0
|
238 |
+
gradio==3.48.0
|
239 |
+
gradio_client==0.6.1
|
240 |
+
greenlet==3.0.3
|
241 |
+
grpcio==1.60.1
|
242 |
+
h11==0.14.0
|
243 |
+
httpcore==1.0.3
|
244 |
+
httptools==0.6.1
|
245 |
+
httpx==0.26.0
|
246 |
+
huggingface-hub==0.20.3
|
247 |
+
humanfriendly==10.0
|
248 |
+
idna==3.6
|
249 |
+
importlib-metadata==6.11.0
|
250 |
+
importlib-resources==6.1.1
|
251 |
+
Jinja2==3.1.3
|
252 |
+
jmespath==1.0.1
|
253 |
+
joblib==1.3.2
|
254 |
+
jsonpatch==1.33
|
255 |
+
jsonpath-python==1.0.6
|
256 |
+
jsonpointer==2.4
|
257 |
+
jsonschema==4.21.1
|
258 |
+
jsonschema-specifications==2023.12.1
|
259 |
+
kiwisolver==1.4.5
|
260 |
+
kubernetes==29.0.0
|
261 |
+
langchain==0.1.7
|
262 |
+
langchain-community==0.0.20
|
263 |
+
langchain-core==0.1.23
|
264 |
+
langchainhub==0.1.14
|
265 |
+
langdetect==1.0.9
|
266 |
+
langsmith==0.0.87
|
267 |
+
lxml==5.1.0
|
268 |
+
MarkupSafe==2.1.5
|
269 |
+
marshmallow==3.20.2
|
270 |
+
matplotlib==3.8.3
|
271 |
+
mmh3==4.1.0
|
272 |
+
monotonic==1.6
|
273 |
+
mpmath==1.3.0
|
274 |
+
multidict==6.0.5
|
275 |
+
mypy-extensions==1.0.0
|
276 |
+
networkx==3.2.1
|
277 |
+
nltk==3.8.1
|
278 |
+
numpy==1.26.4
|
279 |
+
nvidia-cublas-cu12==12.1.3.1
|
280 |
+
nvidia-cuda-cupti-cu12==12.1.105
|
281 |
+
nvidia-cuda-nvrtc-cu12==12.1.105
|
282 |
+
nvidia-cuda-runtime-cu12==12.1.105
|
283 |
+
nvidia-cudnn-cu12==8.9.2.26
|
284 |
+
nvidia-cufft-cu12==11.0.2.54
|
285 |
+
nvidia-curand-cu12==10.3.2.106
|
286 |
+
nvidia-cusolver-cu12==11.4.5.107
|
287 |
+
nvidia-cusparse-cu12==12.1.0.106
|
288 |
+
nvidia-nccl-cu12==2.19.3
|
289 |
+
nvidia-nvjitlink-cu12==12.3.101
|
290 |
+
nvidia-nvtx-cu12==12.1.105
|
291 |
+
oauthlib==3.2.2
|
292 |
+
onnxruntime==1.17.0
|
293 |
+
opentelemetry-api==1.22.0
|
294 |
+
opentelemetry-exporter-otlp-proto-common==1.22.0
|
295 |
+
opentelemetry-exporter-otlp-proto-grpc==1.22.0
|
296 |
+
opentelemetry-instrumentation==0.43b0
|
297 |
+
opentelemetry-instrumentation-asgi==0.43b0
|
298 |
+
opentelemetry-instrumentation-fastapi==0.43b0
|
299 |
+
opentelemetry-proto==1.22.0
|
300 |
+
opentelemetry-sdk==1.22.0
|
301 |
+
opentelemetry-semantic-conventions==0.43b0
|
302 |
+
opentelemetry-util-http==0.43b0
|
303 |
+
orjson==3.9.14
|
304 |
+
overrides==7.7.0
|
305 |
+
packaging==23.2
|
306 |
+
pandas==2.2.0
|
307 |
+
pillow==10.2.0
|
308 |
+
posthog==3.4.1
|
309 |
+
protobuf==4.25.2
|
310 |
+
pulsar-client==3.4.0
|
311 |
+
pyasn1==0.5.1
|
312 |
+
pyasn1-modules==0.3.0
|
313 |
+
pydantic==2.6.1
|
314 |
+
pydantic_core==2.16.2
|
315 |
+
pydub==0.25.1
|
316 |
+
pyparsing==3.1.1
|
317 |
+
PyPika==0.48.9
|
318 |
+
pyproject_hooks==1.0.0
|
319 |
+
python-dateutil==2.8.2
|
320 |
+
python-dotenv==1.0.1
|
321 |
+
python-iso639==2024.2.7
|
322 |
+
python-magic==0.4.27
|
323 |
+
python-multipart==0.0.9
|
324 |
+
pytz==2024.1
|
325 |
+
PyYAML==6.0.1
|
326 |
+
rapidfuzz==3.6.1
|
327 |
+
referencing==0.33.0
|
328 |
+
regex==2023.12.25
|
329 |
+
requests==2.31.0
|
330 |
+
requests-oauthlib==1.3.1
|
331 |
+
rpds-py==0.18.0
|
332 |
+
rsa==4.9
|
333 |
+
s3transfer==0.10.0
|
334 |
+
safetensors==0.4.2
|
335 |
+
scikit-learn==1.4.0
|
336 |
+
scipy==1.12.0
|
337 |
+
semantic-version==2.10.0
|
338 |
+
sentence-transformers==2.3.1
|
339 |
+
sentencepiece==0.1.99
|
340 |
+
sgmllib3k==1.0.0
|
341 |
+
six==1.16.0
|
342 |
+
smmap==5.0.1
|
343 |
+
sniffio==1.3.0
|
344 |
+
soupsieve==2.5
|
345 |
+
SQLAlchemy==2.0.27
|
346 |
+
starlette==0.36.3
|
347 |
+
sympy==1.12
|
348 |
+
tabulate==0.9.0
|
349 |
+
tenacity==8.2.3
|
350 |
+
threadpoolctl==3.3.0
|
351 |
+
tokenizers==0.15.2
|
352 |
+
tomli==2.0.1
|
353 |
+
toolz==0.12.1
|
354 |
+
torch==2.2.0
|
355 |
+
tqdm==4.66.2
|
356 |
+
transformers==4.37.2
|
357 |
+
triton==2.2.0
|
358 |
+
typer==0.9.0
|
359 |
+
types-requests==2.31.0.20240125
|
360 |
+
typing-inspect==0.9.0
|
361 |
+
typing_extensions==4.8.0
|
362 |
+
tzdata==2024.1
|
363 |
+
unstructured==0.12.4
|
364 |
+
unstructured-client==0.18.0
|
365 |
+
urllib3==2.0.7
|
366 |
+
uvicorn==0.27.1
|
367 |
+
uvloop==0.19.0
|
368 |
+
validators==0.22.0
|
369 |
+
watchfiles==0.21.0
|
370 |
+
websocket-client==1.7.0
|
371 |
+
websockets==11.0.3
|
372 |
+
wrapt==1.16.0
|
373 |
+
yarl==1.9.4
|
374 |
+
zipp==3.17.0
|