Yes, we can.

It doesn't have that feature out of the box, but the source code can be modified to save page number info as metadata for the nodes and can be returned to the user. Its not an easy task though...

GraphRAG builds its own custom Knowledge graph with some new concepts like communities and summaries. We can't use RDF KGs GraphRAG unless you modify the source code substantially. I think, the best approach would be to build a RAG on top of your RDF KG. With all these RAG's the main thing, extracting the information relevant to the query and providing that to an LLM to create an answer.

@@SridharKumarKannam thanks for the answer. :) Yes, as it generates KG internally, its not a good idea to modify source code right now as it is already have many open issues for GRAPHRAG. I didn't find any good implementation source to build RAG on top of my RDF KG.

there are function calling issues with ollama models. Try the solution suggested here, I've not tested it though.. github.com/langchain-ai/langchainjs/issues/6051

What is Germlin? Instead of Neo4j you connect to neptune. Bothe are graph DB's and should work the same way.

@@SridharKumarKannam can i get any references or explanation for that

@@dineshmani1846 docs.aws.amazon.com/neptune/latest/userguide/access-graph-gremlin-python.html

Whats the exact error message? The variable name is "horizon_len". model = timesfm.TimesFm( context_len=512, horizon_len=128, input_patch_len=32, output_patch_len=128, num_layers=20, model_dims=1280, backend=timesfm_backend, )

you can follow the instructions here - github.com/facebookresearch/nougat

If you found this content helpful, please consider liking, subscribing, and sharing it with others who might benefit. Your support is greatly appreciated :)

thanks, I'll try but i'm occupied with quite a few things at the moment.

yes, the source code is available, just make sure there are no security issues related to API call to the servers...

If you found this content helpful, please consider liking, subscribing, and sharing it with others who might benefit. Your support is greatly appreciated :)

yes, you can when instantiating the RAG rag = LightRAG( working_dir=WORKING_DIR, chunk_token_size = N )

yes, the source code is available, just make sure there are no security issues related to API call to the servers...

If you found this content helpful, please consider liking, subscribing, and sharing it with others who might benefit. Your support is greatly appreciated :)

You can change the vector DB from nano-vectordb in the source code.

ya, there are a lot of issues with it as you can see from git repo issues, but it works. All the best.

If you found this content helpful, please consider liking, subscribing, and sharing it with others who might benefit. Your support is greatly appreciated :)

The codebase is tricky. A couple of suggestions. 1. Write docstrings for all the functions and important and complex code lines. 2. Use LLMs to create functions decryption. 3. Use the description and docstrings info to create embeddings and add them to the graph entity nodes. 4. Use such info in the retrieval. All the best :)

If you are new to GenAI, I suggest (1) start with OpenAI, Ollama Python SDKs and API to chat with online and local LLMs. (2) Build simple LLM and RAG applications using UI tolls like LangFlow. (3) Move to agents only after you become comfortable doing 1 and 2. All the best...

no this is not creating lineage graphs. You can use something like this... lineage.xml ---------- <lineage> <nodes> <node id="1" name="Data Source 1"/> <node id="2" name="Transformation 1"/> <node id="3" name="Data Source 2"/> <node id="4" name="Transformation 2"/> <node id="5" name="Final Output"/> </nodes> <edges> <edge source="1" target="2"/> <edge source="2" target="3"/> <edge source="3" target="4"/> <edge source="4" target="5"/> </edges> </lineage> ------------------ Python code: import xml.etree.ElementTree as ET import networkx as nx import matplotlib.pyplot as plt # Step 1: Parse the XML File def parse_xml(file): tree = ET.parse(file) root = tree.getroot() return root # Step 2: Extract nodes and edges from the XML def extract_nodes_edges(root): nodes = [] edges = [] for node in root.find('nodes').findall('node'): node_id = node.get('id') node_name = node.get('name') nodes.append((node_id, node_name)) for edge in root.find('edges').findall('edge'): source = edge.get('source') target = edge.get('target') edges.append((source, target)) return nodes, edges # Step 3: Create and visualize the lineage graph def create_lineage_graph(nodes, edges): G = nx.DiGraph() # Create a directed graph # Add nodes to the graph for node_id, node_name in nodes: G.add_node(node_id, label=node_name) # Add edges to the graph for source, target in edges: G.add_edge(source, target) # Step 4: Customize graph layout and labels pos = nx.spring_layout(G) labels = nx.get_node_attributes(G, 'label') # Draw the graph with labels nx.draw(G, pos, labels=labels, with_labels=True, node_color='lightblue', font_size=10, font_weight='bold') # Step 5: Show the graph plt.show() # Main function to execute the script if __name__ == "__main__": # Load and parse the XML file root = parse_xml('/Users/sridharkannam/Downloads/lineage.xml') # Extract nodes and edges nodes, edges = extract_nodes_edges(root) # Create and visualize the lineage graph create_lineage_graph(nodes, edges)

@@SridharKumarKannam Thanks

Thank you. If you found this content helpful, please consider liking, subscribing, and sharing it with others who might benefit. Your support is greatly appreciated :)

Thank you. If you found this content helpful, please consider liking, subscribing, and sharing it with others who might benefit. Your support is greatly appreciated :)

Thank you. If you found this content helpful, please consider sharing it with others who might benefit. Your support is greatly appreciated :)

thats a very good idea and yes you can that. Add the connections to the graph and prompt the agent to use connections for additional info or neighbourhood search...

Thank you. If you found this content helpful, please consider sharing it with others who might benefit. Your support is greatly appreciated :)

Sure, pls contact me at [email protected]

Thank you. If you found this content helpful, please consider sharing it with others who might benefit. Your support is greatly appreciated :)

It worked for me. What the error. Did you check vanna github issues ...

​@@SridharKumarKannam I have raise issue in details please check your github repository

Thank you. If you found this content helpful, please consider sharing it with others who might benefit. Your support is greatly appreciated :)

For example, I have a handbook of cosmetic science with various subjects or entities and I want to collect it's relevant information scattered along the document. After that, the agent might structure and arrange the collected information into well-structured document about the mentioned subjects/entities.

i would use traditional knowledge graph for that as there are many graph based algos which can run on top of traditional knowledge graphs. The new method also has some advantages, try both with a short text and see which one works better for your usecase.

for this one, you can use any agentic system including langgraph, you don't have to use this 4 layers graph structure.

1. kzbin.info/www/bejne/aXfMp5yDbLWeoKM 2. kzbin.info/www/bejne/d5fThKWHqbl7n9k

thank you. If you found this content helpful, please consider sharing it with others who might benefit. Your support is greatly appreciated :)

start with RAG and these strategies medium.com/@anuragmishra_27746/five-levels-of-chunking-strategies-in-rag-notes-from-gregs-video-7b735895694d

@@SridharKumarKannam Thanks for the reply but I want to know where should i make changes in the microsoft rag, to implement my own semantic chunking. And its local and global methods are eating up a lot of tokens so i want to implement my own searching strategy of searching in nearby node and all the top hierarchical communities related to the given entity but this community concept is not very clear to me(i tried searching their github files also but still). So in case you have some resource or something it would be really helpful. Again thanks a lot

@@Anonymous-bu9ch I don't have any additional resources other than my youtube. GraphRAG is quite complex as you said and its very slow and costly. I'm going to do a couple of videos on simplified GraphRAG in the next few days. I suggest you start with RAG and there are many advanced RAGs. Only if none of those work as expected, go for GraphRAG as changing the source code is quite complex and takes a lot of time.

@@SridharKumarKannam Okay thanks !!

no you can not. The GraphRAG query engine expects the graph to be in a specific format, for example, communities and summaries, etc. You need to rebuild the graph.

@@SridharKumarKannam COuld u plz bit explain rebuilding process how to do it ? I have a KG with RDF (.ttl) format

GraphRAG builds its own custom Knowledge graph with some new concepts like communities and summaries. We can't use RDF KGs GraphRAG unless you modify the source code substantially. I think, the best approach would be to build a RAG on top of your RDF KG. With all these RAG's the main thing, extracting the information relevant to the query and providing that to an LLM to create an answer.

?

Once we define an approach, such things are computed in the backend in real time. We don't need to write separate queries for each user.

Yes, you can ingest your data into a KG DB similar to how you ingest data into a sql db.