Chat memory

Traceback (most recent call last):
langchainapi-langchain-1  |   File "/srv/src/gpt.py", line 201, in chat
langchainapi-langchain-1  |     print(self.memory.json())
langchainapi-langchain-1  |   File "pydantic/main.py", line 504, in pydantic.main.BaseModel.json
langchainapi-langchain-1  |   File "/usr/local/lib/python3.9/json/__init__.py", line 234, in dumps
langchainapi-langchain-1  |     return cls(
langchainapi-langchain-1  |   File "/usr/local/lib/python3.9/json/encoder.py", line 199, in encode
langchainapi-langchain-1  |     chunks = self.iterencode(o, _one_shot=True)
langchainapi-langchain-1  |   File "/usr/local/lib/python3.9/json/encoder.py", line 257, in iterencode
langchainapi-langchain-1  |     return _iterencode(o, 0)
langchainapi-langchain-1  |   File "pydantic/json.py", line 90, in pydantic.json.pydantic_encoder
langchainapi-langchain-1  | TypeError: Object of type 'partial' is not JSON serializable

LLogan M

pydantic noooo

LLogan M

hmmm

bbmax

i can try implementing my own jsonencoder for chatmemory buffer

LLogan M

There's a more verbose workaround, with will encode each message as json. Would be nice to have the entire thing encoded though

LLogan M

I think it just needs to skip encoding the tokenizer

LLogan M

https://github.com/jerryjliu/llama_index/blob/a92dd1fce21d7207c64475ab678aabcfecf6e86a/llama_index/memory/chat_memory_buffer.py#L16