Oooo, 0.6.9 is quite old. The library has changed quite a bit since then 😅
There will be a few key changes
I can help migrate the code of its not too big
Just few lines. very basic.
Couple of lines
import os
import openai
from flask import Flask, jsonify, request, make_response
from llama_index import StorageContext, load_index_from_storage
from flask_cors import CORS, cross_origin
os.environ["OPENAI_API_KEY"] = 'sk-**'
storage_context = StorageContext.from_defaults(persist_dir='./storage')
indexMASTER = load_index_from_storage(storage_context)
query_engine = indexMASTER.as_query_engine()
app = Flask(name)
CORS(app)
@app.route('/', methods=['GET', 'POST'])
@cross_origin()
def test():
if request.method == 'POST':
data = request.json
statement = data[ 'statement' ]
context = data['context']
print(statement)
index_results = query_engine.query(f"{context} /n /n {statement}")
print(index_results)
index_results = str(index_results)
return jsonify({'response':{'a':index_results}})
else:
return jsonify({'response':'Nothing here'})
if name == 'main':
app.run()
I had a similar issue, are you using a virtual environment? I had to restore the venv from zero and also I'm using python 3.10. Hope you get it.
Ah tbh this looks good. But you can't use the same saved index I think? Or I guess what kind of error are you getting?
So I have uploaded this flask app to pythonanywhere.com
When I updated the llama-index to latest version I started getting the 500 internal server error.
Then I have GoDaddy server, namecheap server, and I also have my own reseller hosting server. I have tried installing the llama-index on all of these servers but it is showing the following error I have attahced in this image.
These serves support python 3.9
That seems like an issue with the server support for numpy libraries?
Are you deploying with a docker image?
What you mean by same inedx Bro?
If you are referencing to saved vecotor store files, so I am using it like that and it works well.
So I just vectored my document, saved it in the storage folder, then I just use it
No, I am just creating a standard python app using cpannel. Then we get a virtual enviourment as well. So I just activate that and then using the terminal I tried installing the libraries.
Not sure what to tell you, just seems like a general incompatibility with numpy. Probably best to follow the link the error message is giving 🙂
On pythonanywhere.com, when I upgrade the llama-index to latest one
I see something like below
__
Traceback (most recent call last):
File "/home/Ahtasham/mysite/app.py", line 4, in <module>
from llama_index import StorageContext, load_index_from_storage
File "/home/Ahtasham/.local/lib/python3.10/site-packages/llama_index/init.py", line 12, in <module>
from llama_index.data_structs.struct_type import IndexStructType
File "/home/Ahtasham/.local/lib/python3.10/site-packages/llama_index/data_structs/init.py", line 3, in <module>
from llama_index.data_structs.data_structs import (
File "/home/Ahtasham/.local/lib/python3.10/site-packages/llama_index/data_structs/data_structs.py", line 14, in <module>
from llama_index.schema import BaseNode, TextNode
File "/home/Ahtasham/.local/lib/python3.10/site-packages/llama_index/schema.py", line 15, in <module>
from llama_index.bridge.langchain import Document as LCDocument
File "/home/Ahtasham/.local/lib/python3.10/site-packages/llama_index/bridge/langchain.py", line 1, in <module>
import langchain
File "/home/Ahtasham/.local/lib/python3.10/site-packages/langchain/init.py", line 6, in <module>
from langchain.agents import MRKLChain, ReActChain, SelfAskWithSearchChain
File "/home/Ahtasham/.local/lib/python3.10/site-packages/langchain/agents/init.py", line 40, in <module>
from langchain.agents.agent_toolkits import (
File "/home/Ahtasham/.local/lib/python3.10/site-packages/langchain/agents/agent_toolkits/init.py", line 13, in
<module>
from langchain.agents.agent_toolkits.csv.base import create_csv_agent
File "/home/Ahtasham/.local/lib/python3.10/site-packages/langchain/agents/agent_toolkits/csv/base.py", line 4, in <module>
from langchain.agents.agent_toolkits.pandas.base import create_pandas_dataframe_agent
File "/home/Ahtasham/.local/lib/python3.10/site-packages/langchain/agents/agent_toolkits/pandas/base.py", line 18, in <module>
from langchain.agents.types import AgentType
File "/home/Ahtasham/.local/lib/python3.10/site-packages/langchain/agents/types.py", line 5, in <module>
from langchain.agents.chat.base import ChatAgent
File "/home/Ahtasham/.local/lib/python3.10/site-packages/langchain/agents/chat/base.py", line 4, in <module>
from langchain.agents.chat.output_parser import ChatOutputParser
File "/home/Ahtasham/.local/lib/python3.10/site-packages/langchain/agents/chat/output_parser.py", line 12, in <module>
class ChatOutputParser(AgentOutputParser):
File "pydantic/main.py", line 228, in pydantic.main.ModelMetaclass.new
File "pydantic/fields.py", line 488, in pydantic.fields.ModelField.infer
File "pydantic/fields.py", line 419, in pydantic.fields.ModelField.init
File "pydantic/fields.py", line 539, in pydantic.fields.ModelField.prepare
File "pydantic/fields.py", line 801, in pydantic.fields.ModelField.populate_validators
File "pydantic/validators.py", line 723, in find_validators
RuntimeError: no validator found for <class 're.Pattern'>, see arbitrary_types_allowed in Config
I understand, The above error is from python 3.10/ server is pyhtonanywhere.com
and it happens when I update to latest llama-index
But the numpy incompatibilty error occurs on my godaddy, namecheap or personal hosting servers
This error seems related to maybe your langchain version or pydantic version? For reference, locally I have
llama-index==0.8.13
pydantic==1.10.12
langchain==0.0.277
My local machine has
llama-index 0.8.12
pydantic 1.10.7
langchain 0.0.275
This is on my pythonanyehere server
langchain 0.0.174
llama-index 0.6.9
pydantic 1.9.0
The code on pythonanyehere only work this combination of lanchain and llama-index.
If i upadte these two libraries it breaks
well, idk man, I'm lost lol
I would recommend deploying somewhere that supports a docker container, less issues with this kind of stuff, easier to debug
No worries I will try with you recommended solution
Can you suggest me somthing which service I should that supports docker container
I think the most general choice is AWS? But its also a little complex. I'm sure some other providers out there may have easier solutions (likely at the cost of flexibility)
Well great help. I sorted out the issues. It was all becasue of the compatibility of different libraries. But yes, finally i made it working
