Hacker News Reader

joeyxiongon Show HN: Open-source real-time talk-to-AI wearable device for few $

You are right, “always listening" they mean it's sitting in a low power state listening for a very specific trigger like "Hey Siri" or "OK Google" and literally nothing else.

joeyxiongon Show HN: Open-source real-time talk-to-AI wearable device for few $

Sorry for the confusion, we are still discussing the paid subscription pricing, but I can be sure that the price of premium subscription will not be higher than $9 per month.

joeyxiongon Show HN: I build an editor-style RAG app

sorry, the issue is fixed now

joeyxiongon Show HN: I build an editor-style RAG app

BTW, it's free.

joeyxiongon Ask HN: Have you seen any useful GPTs yet?

Not really

joeyxiongon Show HN: IncarnaMind-Chat with your multiple docs using LLMs

Thanks for pointing it out. It's not your fault at all. I should write the pre-requisite resources downloader in the code as well. Will merge the new code into the main branch.

joeyxiongon Show HN: IncarnaMind-Chat with your multiple docs using LLMs

Yeah, the sliding window chunking mechanism appears in stage 2.

joeyxiongon Show HN: IncarnaMind-Chat with your multiple docs using LLMs

Yeah, This can definatly be used for local models, but the problem is that most personal computers cannot host large LLMs and the cost is not cheaper than closed LLMs. But for organisations, local LLMs are a better choice.

joeyxiongon Show HN: IncarnaMind-Chat with your multiple docs using LLMs

Hi, I've pushed the new commit to the main branch. Could you please test it out? If it still has this error, you can check if your doc has relevant metadata.

```` for d in doc: print("metadata:", d.metadata) ```

before file_name = doc[0].metadata["source"].split("/")[-1].split(".")[0]

joeyxiongon Show HN: IncarnaMind-Chat with your multiple docs using LLMs

Hi, I created the diagrams using Figma.

The retrieval process consists of three stages. The first stage retrieves small chunks from multiple documents to create a document filter using their metadat. This filter is then applied in the second stage to extract relevant large chunks, essentially sections of documents, which further refines our search parameters. Finally, using both the document and large chunk filters, the third stage retrieves the most pertinent medium-sized chunks of information to be passed to the Language Model, ensuring a focused and relevant response to your query.

joeyxiongon Show HN: IncarnaMind-Chat with your multiple docs using LLMs

Thank you for your comment. The sliding window approach allows me to dynamically identify relevant "large chunks," which can be thought of as sections in a document. Often, your questions may pertain to multiple such sections. Using only medium chunks for retrieval could result in sparse or fragmented information.

The third retrieval focuses on "medium chunks" within these identified large chunks. This ensures that only the most relevant information is passed to the Language Model, enhancing both time efficiency and focus. For example, if you're asking for a paper summary, I can zero in on medium chunks within the Abstract, Introduction, and Conclusion sections, eliminating noise from other irrelevant sections. Additionally, this strategy helps manage token limitations, like GPT-3.5's 4000-token cap, by selectively retrieving information

joeyxiongon IncarnaMind: Dynamic chunking for LLM Retrieval-augmented generation(RAG)

IncarnaMind (https://github.com/junruxiong/IncarnaMind) can chat with your private documents (PDF, TXT) using Large Language Models (LLMs) like GPT. While OpenAI has recently launched a fine-tuning API for GPT models, it doesn't enable the base pretrained models to learn new data, and the responses can be prone to factual hallucinations. The dynamic Sliding Window Chunking mechanism and Emsemble Retriever enable efficient querying of both fine-grained and coarse-grained information within ground truth documents to augment the LLMs.