Merge branch 'master' of http://git.d-popov.com/popov/ai-kevin

_doc/_notes/PROMPTS.md

@@ -0,0 +1,13 @@
+You are an expert in extracting new information from text sorting it out in existing categories or creating new categoty (We're using Neo4j as graph database).
+This is multi-step process:
+1. Divide the text into paragraph simmilar chunks that have the same toppic. Add a summary to the paragraph.
+2. Create Memory object for each summary
+3. Extract facts from each paragraph and add them as knowledge linked to the paragraph as separate memory object linked to the first one. Look into verbs, ajectives and nouns to extract the plain information from the text. If there is a source code, do not interpret it, but remember it as linked Memory as it is, while adding summary and link it to the main "Memory".
+4. Assign possible category and scientific field labels to the information by adding them as tags.
+This systematization and segmentation will allow you to remember the text and store it in your long-term memory as knowledge graph, while providing easy access later. Respond in json format with the extracted and restructured data.
+Here's the source text:
+
+
+
+------------
+create a web app that has a button to start recording sends it to latest whisper llm for STT, translates it to target language and shows the result in realtime. it should be streaming text with minimal lag.

_doc/_notes/arti/.excalidraw.svg

@@ -0,0 +1,21 @@
+<svg version="1.1" xmlns="http://www.w3.org/2000/svg" viewBox="0 0 20 20" width="20" height="20">
+  <!-- svg-source:excalidraw -->
+  <!-- payload-type:application/vnd.excalidraw+json --><!-- payload-version:2 --><!-- payload-start -->eyJ2ZXJzaW9uIjoiMSIsImVuY29kaW5nIjoiYnN0cmluZyIsImNvbXByZXNzZWQiOnRydWUsImVuY29kZWQiOiJ4nEWOMVx1MDAwYsIwXHUwMDEwhff+ilx1MDAxMldtwbEggu4uXHUwMDFkxeFIznqYNCG5tGrpfzeJgzdcdTAwMWO871x1MDAxZe/dUtW14LdD0dVcdTAwMDJfXHUwMDEyNClcdTAwMGaz2GY+oVx1MDAwZmTHdNpcdTAwMTdcdTAwMWRs9LI4XHUwMDFmzC50bWvAP5GdXHUwMDA2ic1EIYJcdTAwMGVcdTAwMWNcdTAwMTXZRlrTXHUwMDEyo1x0x7wvYPDgrFHsm3/JXHUwMDBlXHUwMDE1sfW/LtRocOSQ0q+3QsC5noFz35J0XCKDJ9XTJ5Mxar390YlwPoF8XHUwMDBl3sZRna1OkenFzb2MSKa15N1JY45f1mr9XHUwMDAy4lxcTtsifQ==<!-- payload-end -->
+  <defs>
+    <style class="style-fonts">
+      @font-face {
+        font-family: "Virgil";
+        src: url("https://open-vsx.org/vscode/asset/pomdtr/excalidraw-editor/3.7.3/Microsoft.VisualStudio.Code.WebResources/extension/public//dist/excalidraw-assets/Virgil.woff2");
+      }
+      @font-face {
+        font-family: "Cascadia";
+        src: url("https://open-vsx.org/vscode/asset/pomdtr/excalidraw-editor/3.7.3/Microsoft.VisualStudio.Code.WebResources/extension/public//dist/excalidraw-assets/Cascadia.woff2");
+      }
+      @font-face {
+        font-family: "Assistant";
+        src: url("https://open-vsx.org/vscode/asset/pomdtr/excalidraw-editor/3.7.3/Microsoft.VisualStudio.Code.WebResources/extension/public//dist/excalidraw-assets/Assistant-Regular.woff2");
+      }
+    </style>
+    
+  </defs>
+  <rect x="0" y="0" width="20" height="20" fill="#ffffff"></rect></svg>

_doc/_notes/arti/ideas.md

@@ -0,0 +1,32 @@
+Key features & principles:
+ - modal/plug & play design 
+ - Biomimicing based
+    - self inferencing loop
+    - Graph->LLM->Graph based logic (Self reflect)
+    - attention (Short term memory)
+    - generalized & contextuaized memory schema (memory is strongly context dependent and temporal)
+LLM module
+Graph module
+Short term memory module
+mid-term memory (history on the toppic)
+graph powered long term memory with embedding storage for skills & AII ( interface on some of the layers)
+separate text IOs"
+ - multi agent communication module/console/  
+ - internal state/context/mood/STM
+ - actions output
+
+
+
+ GRAPH schema
+
+ idea
+ - is child of
+
+ Q: Brainstorm neo4j schema for biomimicing memory storage as neo4j graph database. It should be similar to the way humans store, retrieve and generalize knowledge 
+
+Memory model:
+
+
+-- example prompt:
+you are an AI living native on a linux host. you have all local resources and installed software at your disposal: root console access, git repo and the internet as resources. you have your own codebase and will brainstorm and find ideas how to improve your reasoning and knowledge of the world. . To avoid messing up your health, you will prefer to install new packages and test ideas inside docker or virtual envs. You can test your ideas in a docker containers and can experiment with different ways to pipleline and organise the thought process. you will also explain each approach and promote the best containers. A local neo4j database is your memory and knowledge source. Whenever you are not certain about a toppic, or need to refresh your knowledge, you will check online for updated information. If the problem concerns npm, pip or other package or an exiting repository on github or other repository server, you will first update your internal database with the latest readme and usage documentaion from the repository authors and later search stackoverflow to update your skills. periodically you will reorgnize and tidy your "mmory" making it more easily accessible and updating the embedding vectors to allow easy access to the needed context when there is a problem to solve. You will build a tools library over time that wil lgelp you solve more complex problems and keep it in your memory together with a list of task completed with each tool. you will experiment with different pipelines for different tasks, memorise them and generalize and optimize existing pipelines over time to build a solid knowhow.
+let's create the project setup in python. our main file will be the agent, and it will call the llm and other tools (maybe langchain), explore, organize and improve when 'resting'. 

_doc/_notes/arti/neo4j.cql

@@ -0,0 +1,82 @@
+
+<!-- https://guides.neo4j.com/wiki -->
+# Cypher Query Language
+
+Runs a simple command to clean the database
+MATCH (n) DETACH DELETE n
+
+
+CREATE INDEX FOR (c:Category) ON (c.catId);
+CREATE INDEX FOR (c:Category) ON (c.catName);
+CREATE INDEX FOR (p:Page) ON (p.pageTitle);
+CREATE (c:Category:RootCategory {catId: 0, catName: 'Databases', subcatsFetched: false, pagesFetched: false, level: 0});
+
+<!-- install APOC library -->
+RUN mkdir -p /var/lib/neo4j/plugins \
+    && cd /var/lib/neo4j/plugins \
+    && curl -L -O https://github.com/neo4j-contrib/neo4j-apoc-procedures/releases/download/4.4.0.0/apoc-4.4.0.0-all.jar
+        <!-- environment:
+      NEO4JLABS_PLUGINS: '["apoc"]'
+      NEO4J_apoc_export_file_enabled: 'true'
+      NEO4J_apoc_import_file_enabled: 'true'
+      NEO4J_apoc_import_file_use__neo4j__config: 'true'
+      NEO4J_dbms_security_procedures_unrestricted: apoc.* -->
+
+<!-- UNWIND range(0,3) as level
+CALL apoc.cypher.doit("
+MATCH (c:Category { subcatsFetched: false, level: $level})
+CALL apoc.load.json('https://en.wikipedia.org/w/api.php?format=json&action=query&list=categorymembers&cmtype=subcat&cmtitle=Category:' + apoc.text.urlencode(c.catName) + '&cmprop=ids%7Ctitle&cmlimit=500')
+YIELD value as results
+UNWIND results.query.categorymembers AS subcat
+MERGE (sc:Category {catId: subcat.pageid})
+ON CREATE SET sc.catName = substring(subcat.title,9),
+              sc.subcatsFetched = false,
+              sc.pagesFetched = false,
+              sc.level = $level + 1
+WITH sc,c
+CALL apoc.create.addLabels(sc,['Level' +  ($level + 1) + 'Category']) YIELD node
+MERGE (sc)-[:SUBCAT_OF]->(c)
+WITH DISTINCT c
+SET c.subcatsFetched = true", { level: level }) YIELD value
+RETURN value -->
+
+<!-- CALL {
+  LOAD CSV FROM "https://github.com/jbarrasa/datasets/blob/master/wikipedia/data/cats.csv?raw=true" AS row
+  CREATE (c:Category { catId: row[0]})
+  SET c.catName = row[2], c.pageCount = toInteger(row[3]), c.subcatCount = toInteger(row[4])
+} IN TRANSACTIONS OF 10000 ROWS
+
+CALL {
+  LOAD CSV FROM "https://github.com/jbarrasa/datasets/blob/master/wikipedia/data/cats.csv?raw=true" AS row
+  CREATE (c:Category { catId: row[0]})
+  SET c.catName = row[2], c.pageCount = toInteger(row[3]), c.subcatCount = toInteger(row[4])
+} IN TRANSACTIONS OF 10000 ROWS -->
+
+CALL {
+  LOAD CSV FROM "https://github.com/jbarrasa/datasets/blob/master/wikipedia/data/cats.csv?raw=true" AS row
+  CREATE (c:Category { catId: row[0]})
+  SET c.catName = row[2], c.pageCount = toInteger(row[3]), c.subcatCount = toInteger(row[4])
+} 
+
+LOAD CSV FROM "https://github.com/jbarrasa/datasets/blob/master/wikipedia/data/rels.csv?raw=true" AS row
+MATCH (from:Category { catId: row[0]})
+MATCH (to:Category { catId: row[1]})
+CREATE (from)-[:SUBCAT_OF]->(to)
+
+ <!-- stats -->
+MATCH (c:Category)
+return SUM(c.pageCount) AS `#pages categorised (with duplicates)`,
+	   AVG(c.pageCount) AS `average #pages per cat`,
+       percentileCont(c.pageCount, 0.75) AS `.75p #pages in a cat`,
+	   MIN(c.pageCount) AS `min #pages in a cat`,
+       MAX(c.pageCount) AS `max #pages in a cat`
+
+MATCH (c:Category)
+WHERE NOT (c)-[:SUBCAT_OF]-()
+RETURN COUNT(c)
+
+
+
+MATCH (c:Category)
+WHERE c.catName CONTAINS '{term}'
+RETURN c;

_doc/_notes/arti/wikidata/import.sh

@@ -0,0 +1,33 @@
+{
+  /* the Wikidata JSON file */
+  "file": "./wikidata-dump.json",
+
+  /* neo4j connection details */
+  "neo4j": {
+    /* bolt protocol URI */
+    "bolt": "bolt://localhost",
+    "auth": {
+      "user": "neo4j",
+      "pass": "password"
+    }
+  },
+  /* Stages */
+  "do": {
+    /* database cleanup */
+    "0": true,
+    /* importing items and properties */
+    "1": true,
+    /* linking entities and generating claims */
+    "2": true
+  },
+  /* extra console output on stage 2 */
+  "verbose": false,
+  /* how many commands will be ran by the DB at a given time */
+  "concurrency": 4,
+  /* skip lines */
+  "skip": 0,
+  /* count of lines */
+  "lines": 21225524,
+  /* bucket size of entities sent to DB to process */
+  "bucket": 1000
+}

_doc/_notes/readme.md

@@ -0,0 +1,4 @@
+start the project in container:
+#
+#for debugging: 
+# docker-compose -f docker-compose.debug.yml up