LLM MODEL – LANCHIN -- GENERATIVE DATA AI
VECTOR DB – FOR STORAGE – NEITHER SEQUAL
EMBEDDING
CRUD – CREATE , UPDATE DELETE
SEVERLESS
LLM MODEL – LANCHIN -- GENERATIVE DATA AI ----pre trained ----- TEXT TO TEXT ---PRETRAIN DATA MEI
LOGICAL IMPLEMENTATION
VECTOR DB – FOR STORAGE – NEITHER SEQUAL
EMBEDDING
CRUD – CREATE , UPDATE DELETE
SEVERLESS
Llm – ai
Vector database ---ARRAY
DEGREE OF DIFFERENCE
Tokenaiser ----- the cat sat on a mat
The – 10
Cat 12
Transform , human
Ing – computational value
limE , lemon
rotational vector embedding
word of sequence
schementic meanings
RAG --- Retrieval Augmented Generation
external knowledge sources---
ENHANCE LLM CUSTOM DATA ----UPDATED --
efeciency enchance –increases the efficiency of LLm , correct data
WHY RAG?
HALLUCINATION ----LOOP
OUTDATED KNOWLEDGE ---ALWAYS UPADATED
SPECIFIC DOMAIN KNOWLEDGE----
PRODCUT – DOMAIN ---CONTEXT IS DIFFEENT (ZOMATAO , AMAZON
MSL
AI TOOL ---- RANDOMNESS , GENERATIVE
MID JOURNEY
NLPSP
⚡: Schematic , contextual meaning , Long term
[20:31, 5/14/2024] Palak
memory retrieval
- vector database
[20:33, 5/14/2024] Palak⚡: Image similarly based on their attributes
[20:34, 5/14/2024] Palak⚡: Similar products based on the rating
[20:36, 5/14/2024] Palak⚡: Retrieving storing information
[20:38, 5/14/2024] Palak⚡: Large language model is bad at producing factual data , it can show irrelevant
information
[20:38, 5/14/2024] Palak⚡: Long term memory retrieval
CUSTOM DATA – PRIVATE DATA (DATA NOT AVAILABLE IN MARKET) DATA SHOULD NOT BE AVAILABLE TO
ANYONE , PERSONAL DATA
Graph ql
GENERALISED INFO
TRAIN
TRANSFORMERS
AUTO MERGING RETREIVAL
CHUNK ------
ADAPTIVE CHUNKING – DYNAMIC
