All Questions

Ask Question
20 questions
14votes
1answer
204views
How to reduce hallucinations when using LLMs for data analysis tasks?

I'm building a system where users ask questions about their data in natural language, and an LLM generates SQL queries and interprets results. **The problem:** The LLM often "hallucinates" insights t...

RRaj Patel🌱 Demo1650•asked about 2 months ago
7votes
1answer
158views
What are the privacy implications of using LLMs with user data?

Our company wants to use GPT-4 to analyze customer support tickets and suggest responses. Legal and compliance teams are concerned about: 1. **Data retention**: Does OpenAI store our API requests? 2....

DDavid Miller🌱 Demo1560•asked about 2 months ago
12votes
1answer
161views
Comparing function calling vs JSON mode for structured output - which is better?

I need to extract structured data from unstructured text using GPT-4. I've tried both approaches: **Function Calling:** ```python response = client.chat.completions.create( model="gpt-4", mes...

MMarcus Johnson🌱 Demo2120•asked about 2 months ago
17votes
1answer
174views
How to implement RAG (Retrieval-Augmented Generation) with custom embeddings?

I want to build a RAG system for our internal documentation, but I'm confused about the embedding strategy. **Current setup:** - 500+ markdown documentation files - Using OpenAI's text-embedding-3-sm...

AAlex Rodriguez🌱 Demo1920•asked about 2 months ago
18votes
2answers
310views
What are the best practices for versioning and testing prompts in production?

I'm working on a production system where we use GPT-4 for various tasks. As we iterate on our prompts, I'm concerned about: 1. **Version control**: How do you track prompt changes over time? 2. **A/B...

EEmma Thompson🌱 Demo2340•asked about 2 months ago
27votes
0answers
116views
How to reduce hallucinations when using LLMs for data analysis tasks?

I'm building a system where users ask questions about their data in natural language, and an LLM generates SQL queries and interprets results. **The problem:** The LLM often "hallucinates" insights t...

RRaj Patel🌱 Demo1650•asked about 2 months ago
26votes
0answers
117views
What are the privacy implications of using LLMs with user data?

Our company wants to use GPT-4 to analyze customer support tickets and suggest responses. Legal and compliance teams are concerned about: 1. **Data retention**: Does OpenAI store our API requests? 2....

DDavid Miller🌱 Demo1560•asked about 2 months ago
8votes
0answers
271views
Comparing function calling vs JSON mode for structured output - which is better?

I need to extract structured data from unstructured text using GPT-4. I've tried both approaches: **Function Calling:** ```python response = client.chat.completions.create( model="gpt-4", mes...

MMarcus Johnson🌱 Demo2120•asked about 2 months ago
15votes
0answers
225views
How to implement RAG (Retrieval-Augmented Generation) with custom embeddings?

I want to build a RAG system for our internal documentation, but I'm confused about the embedding strategy. **Current setup:** - 500+ markdown documentation files - Using OpenAI's text-embedding-3-sm...

AAlex Rodriguez🌱 Demo1920•asked about 2 months ago
11votes
0answers
264views
What are the best practices for versioning and testing prompts in production?

I'm working on a production system where we use GPT-4 for various tasks. As we iterate on our prompts, I'm concerned about: 1. **Version control**: How do you track prompt changes over time? 2. **A/B...

EEmma Thompson🌱 Demo2340•asked about 2 months ago
6votes
0answers
352views
What are the privacy implications of using LLMs with user data?

Our company wants to use GPT-4 to analyze customer support tickets and suggest responses. Legal and compliance teams are concerned about: 1. **Data retention**: Does OpenAI store our API requests? 2....

DDavid Miller🌱 Demo1560•asked about 2 months ago
14votes
0answers
132views
How to reduce hallucinations when using LLMs for data analysis tasks?

I'm building a system where users ask questions about their data in natural language, and an LLM generates SQL queries and interprets results. **The problem:** The LLM often "hallucinates" insights t...

RRaj Patel🌱 Demo1650•asked about 2 months ago
18votes
0answers
319views
Comparing function calling vs JSON mode for structured output - which is better?

I need to extract structured data from unstructured text using GPT-4. I've tried both approaches: **Function Calling:** ```python response = client.chat.completions.create( model="gpt-4", mes...

MMarcus Johnson🌱 Demo2120•asked about 2 months ago
5votes
0answers
169views
How to implement RAG (Retrieval-Augmented Generation) with custom embeddings?

I want to build a RAG system for our internal documentation, but I'm confused about the embedding strategy. **Current setup:** - 500+ markdown documentation files - Using OpenAI's text-embedding-3-sm...

AAlex Rodriguez🌱 Demo1920•asked about 2 months ago
14votes
0answers
304views
What are the best practices for versioning and testing prompts in production?

I'm working on a production system where we use GPT-4 for various tasks. As we iterate on our prompts, I'm concerned about: 1. **Version control**: How do you track prompt changes over time? 2. **A/B...

EEmma Thompson🌱 Demo2340•asked about 2 months ago
11votes
1answer
163views
How to handle multilingual support in LLM applications?

We want to expand our AI chatbot to support 10+ languages. Questions: 1. Should we use separate prompts for each language or rely on LLM's multilingual capabilities? 2. How to ensure consistent quali...

JJessica Wang🌱 Demo1480•asked about 2 months ago
8votes
1answer
381views
How to use chain-of-thought prompting effectively?

I've heard a lot about chain-of-thought (CoT) prompting improving LLM reasoning, but I'm not sure how to implement it properly. Can someone explain: 1. What exactly is chain-of-thought prompting? 2. ...

SSophie Anderson🌱 Demo420•asked about 2 months ago
19votes
1answer
287views
How to optimize LLM inference costs in production?

Our AI application is getting expensive with GPT-4 API calls. We're spending $5000/month and growing. What strategies can reduce costs without sacrificing too much quality? Current setup: - 100k API...

MMike Chen🌱 Demo1750•asked about 2 months ago
12votes
2answers
487views
What are the most effective prompting techniques to reduce hallucinations in RAG pipelines?

I am building a Retrieval-Augmented Generation (RAG) chatbot for internal company documents. Sometimes the LLM makes up information when the retrieved context doesn't contain the answer. What prompti...

AAlex Rodriguez🌱 Demo1920•asked about 2 months ago
14votes
1answer
344views
Best practices for fine-tuning GPT-4 for domain-specific tasks?

I want to fine-tune GPT-4 for medical diagnosis assistance. I have a dataset of 10,000 medical cases with symptoms and diagnoses. What are the best practices for: 1. Preparing the training data 2. Ch...

RRaj Patel🌱 Demo1650•asked about 2 months ago