I am trying the Raptor pack with custom LLM and Embedding model import using langchain. and getting this error . Any ideas? File ~/Documents/ai/.venv/lib/python3.10/site-packages/llama_index/core/indices/prompt_helper.py:117, in PromptHelper.from_llm_metadata(cls, llm_metadata, chunk_overlap_ratio, chunk_size_limit, tokenizer, separator)
103 @classmethod
104 def from_llm_metadata(
105 cls,
(...)
110 separator: str = " ",
111 ) -> "PromptHelper":
112 """Create from llm predictor.
113
114 This will autofill values like context_window and num_output.
115
116 """
--> 117 context_window = llm_metadata.context_window
119 if llm_metadata.num_output == -1:
120 num_output = DEFAULT_NUM_OUTPUTS

AttributeError: 'NoneType' object has no attribute 'context_window'

Hello All, I am trying to build a demo with "SubDocSummaryPack" . All the examples i see, they run the query directly on the llama_index.packs.subdoc_summary.base.SubDocSummaryPack object after running the pack to chunk and add summary. Can someone please share an example of how i can use it with a vector store. Ideally Chroma db?