KG and RAG
Tools
NER
https://github.com/huggingface/neuralcoref
Tavily
Connect Your Agent to the Web
Powering the Internet of Agents with fast, secure
and reliable web access APIs.
SpaCy
https://explosion.ai/blog/coref
A Tricky Problem: Split Antecedents
More precisely the problem is treated as a clustering problem over non-overlapping spans in a document. The non-overlapping constraint renders the system incapable of handling the “split antecedent” problem. For example in “Alice and Bob said they like cheese, but he prefers sushi.” The pronoun “they” refers to “Alice and Bob” and “he” refers to “Bob”. However, the span “Bob” is inside “Alice and Bob” so we have to choose to either resolve “they” to “Alice and Bob” or “he” to “Bob”. The lack of split antecedent handling is a limitation of many coreference resolutions systems including ours.
Relik
ReLiK: The Reader-Retriever Paradigm
- Dense Representations:
- Similarity Matching: it figures out how similar these dense vectors are to each other. This helps it find the best candidates from a huge list (think of a huge library of possibilities).
- Multi-label Noise Contrastive Estimation (NCE) for Efficiency:
A method called multi-label NCE is used to train the Retriever. This lets the Retriever learn quickly by comparing the right candidates with a lot of wrong ones during each training round.
https://www.freedium.cfd/https://pub.towardsai.net/is-relik-the-future-of-nlp-01998766ae55
Kotaemon
An open-source clean & customizable RAG UI for chatting with your documents. Built with both end users and developers in mind.
https://github.com/Cinnamon/kotaemon
RAGAS
https://docs.ragas.io/en/stable/concepts/index.html
https://huggingface.co/spaces/relik-ie/Information-Extraction
AweSOME AGENTS + MEMORY
https://www.youtube.com/watch?v=7LWTZqksmSg
"I want Llama3.1 to perform 10x with my private knowledge"
Self learning Local Llama3.1 405B
https://www.youtube.com/watch?v=2PKCOVqhngY&t=2s
Atomic agents
The Atomic Agents framework is designed around the concept of atomicity to be an extremely lightweight and modular framework for building Agentic AI pipelines and applications without sacrificing developer experience and maintainability.
Think of it like building AI applications with LEGO blocks - each component (agent, tool, context provider) is:
- Single-purpose: Does one thing well
- Reusable: Can be used in multiple pipelines
- Composable: Easily combines with other components
- Predictable: Produces consistent, reliable outputs
Anatomy of an Agent
In Atomic Agents, an agent is composed of several key components:
- System Prompt: Defines the agent's behavior and purpose.
- Input Schema: Specifies the structure and validation rules for the agent's input.
- Output Schema: Specifies the structure and validation rules for the agent's output.
- History: Stores conversation history or other relevant data.
- Context Providers: Inject dynamic context into the agent's system prompt at runtime.
Here's a high-level architecture diagram:
https://www.youtube.com/watch?v=Sp30YsjGUW0
https://github.com/BrainBlend-AI/atomic-agents/