for some reason whenever I load an index

CChancellor Hands LLC

for some reason whenever I load an index.json from disk into a GPTSimpleVectorStoreIndex I cannot pass the service_context:

from langchain.agents import Tool
from langchain.chains.conversation.memory import ConversationBufferMemory
from langchain import OpenAI
from langchain.agents import initialize_agent

from gpt_index import GPTSimpleVectorIndex

# define LLM
llm_predictor = LLMPredictor(llm=ChatOpenAI(temperature=0))

# define prompt helper
# set maximum input size
max_input_size = 4096
# set number of output tokens
num_output = 256
# set maximum chunk overlap
max_chunk_overlap = 20
prompt_helper = PromptHelper(max_input_size, num_output, max_chunk_overlap)

service_context = ServiceContext.from_defaults(llm_predictor=llm_predictor, prompt_helper=prompt_helper, chunk_size_limit=750)

index = GPTSimpleVectorIndex.load_from_disk(save_path='minimalist_entrepreneur_2.json',  service_context=service_context)

and get this error:

    331 """Run query.
    332 
    333 NOTE: Relies on mutual recursion between
   (...)
    344     composable graph.
...
     45     )
     46 llm_predictor = service_context.llm_predictor
     47 embed_model = service_context.embed_model

ValueError: Cannot use llm_token_counter on an instance without a service context.

11 comments

LLogan M

this is pretty weird, never had an issue with loading the index 🤔 Was the index created a long time ago?

You could try this instead
index = GPTSimpleVectorIndex.load_from_disk('minimalist_entrepreneur_2.json', service_context=service_context)

I'm assuming it works without the service context?

CChancellor Hands LLC

I just made the index and saved it to disk in the same day, with the same llama_index version as well. And yes it does work without the service_context parameter.

CChancellor Hands LLC

@Logan M Are you able to pass service_context when you load from disk? if so, what version of llama-index are you using? I'm using 0.5.13.post1 on my computer

LLogan M

I sure am, I've never had an issue doing it 😅 I'm still on 0.5.7 it looks like, I'll try upgrading

LLogan M

Upgraded to 0.5.15, works fine for me 🤔

CChancellor Hands LLC

trying 0.5.15 now

LLogan M

Maybe you need to specify the model name in ChatOpenAI? All my code is on github here https://github.com/logan-markewich/llama_index_starter_pack/blob/main/streamlit_vector/streamlit_demo.py

CChancellor Hands LLC

@Logan M it worked when I updated to 0.5.15. Also not sure if it made a difference but I imported GPTSimpleVectorIndex from gpt-index the first time and llama-index currently. Same versions

LLogan M

Oh! I must have missed that earlier. Definitely use the llama_index imports 🙏🙏

CChancellor Hands LLC

Ah I see. I assumed that they were identical but I guess not.

CChancellor Hands LLC

Thanks for your help btw! @Logan M

Add a reply

Find answers from the community

for some reason whenever I load an index