echo.hive
@hive_echo • 14,194 subscribers
concocting RL envs… 🟣 Maximize yourself: https://t.co/UOJxh5tKPw 🔴Consulting: https://t.co/7SFrM422Fq
Shorts
Videos
I built a biologically inspired spiking neural network from scratch and it learned with %5 accuracy to do addition :) There is no backpropagation, no artificial loss functions - just spikes, synapses, and dopamine-like reward signals. it uses STDP -> "Spike-Timing-Dependent Plasticity" with modulated rewards This is super fun and I will try to get it to learn with better accuracy. I also need to better understand how all the moving parts fit together Link to source code in comment which has a detailed readme and html with animations explaining how it all works
echo.hive446,927 views • 7 months ago
Spiking Neural Network from scratch achieves 8% accuracy. no backpropagation or SGD I created a genetic hyper parameter optimizer and it now, on average, can get 8% accuracy which is ~3% above chance Link to source code with a detailed video and markdown explanations in comment it also usually starts lower than 5% and slowly improves then eventually can start dipping below 5%, all of which leads me to believe there is glimmers of learning taking place. sometims it is stabel around 7-8-9% for a long time there is no backpropagation or SGD. it learns via STDP (spike timing dependent plasticity) and a reward mechanism each example is presented n many times (500 in this case) which produces spike train which leads to eligibility list, at the end of a turn based on if the answer was correct or not we adjust the weights using the reward as a multiplier spike timing keeps track of the sequence of neuron firings and which ones were more likely to have lead to the correct answer let me know what you think
echo.hive225,484 views • 7 months ago
Ralph Wiggum in Cursor with Gemini 3 flash creating spiking neural nets
echo.hive70,101 views • 4 months ago
Cursor testing my full brain simulation's learning capabilities code in comment
echo.hive51,039 views • 5 months ago
Opus 4.5 in-a-loop builds Spiking Neural Net simulations in Cursor that is all :)
echo.hive42,700 views • 4 months ago
Epic duo! computer use drives cursor autonomously and builds and entire Tower defense game while maintaining project understanding 🥳 system message in comment You just instruct computer use and it continually instructs Cursor agent to build the project. You can download the script from my patron. link in comment How it works: - Computer is limited to certain coordinates and certain actions this way it should only work with cursor and do nothing else - cursorrules file instructs cursor to not run any commands to run files, since those require user permission and computer use may make mistakes there - you select the coordinates of the top and bottom position of the text input boxes in composer side panel. there is a script that helps you with that. we enter these into comuter_use script - then run the script with your prompt let me know what you think 🙂 ------- If you are interested in learning how to use Cursor more in depth, I have a 28 chapter 1000x Cursor Course building full apps from scratch with 19 hours of content. link is in my bio if you are interested
echo.hive126,688 views • 1 year ago
2 Cursor agents in separate tabs chat and plan the most interesting app ever and build it too! collaboratively All you need is 2 rules, THAT IS IT! here is how: create 2 rule files set to "Manual" agent-1 .mdc: --- You are agent-1 you will be chatting with agent-2 to design and build the most interesting python app ever you will write to agent_1.txt file and read from agent_2.txt file if you are waiting for a new response write a cli command to wait for 5 seconds and check again you will repeat this untill the full app is built you start the conversation --- agent-2 .mdc: --- You are agent-2 you will be chatting with agent-1 to design and build the most interesting python app ever you will write to agent_2.txt file and read from agent_1.txt file if you are waiting for a new response write a cli command to wait for 5 seconds and check again you will repeat this untill the full app is built agent-1 will start the convo --- create a new agent tab, you should have 2 tabs assign agent 1 its rule and agent 2 its rule type "begin" for agent 1 and enter type "begin" for agent 2 and enter That is it! and then watch them go to work! --- Want to level up your Cursor game? I’ve created a 45-chapter course on mastering Cursor. Check it out via the link in my bio! each chapter is short and independent and designed to get your started quickly featuring 26 hours of content where we build interesting apps and ideas from scratch in each chapter. ---
echo.hive88,603 views • 1 year ago
Hopfield recall Spiking Neural network now this one is cool isn't it? -open source, link in comment- This is a Spiking Neural Network (SNN) that can learn and recall patterns through Hebbian learning, similar to a Hopfield network but using biologically-inspired spiking dynamics rather than energy-based settling.
echo.hive31,452 views • 4 months ago
📚🤖 Open Source: AI reads PDFs Page-by-Page, extracts Knowledge & Summarizes 🥳 Repo in comment 💝 - 📚 Automated page by page PDF book analysis and knowledge extraction - 🤖 AI-powered content understanding and summarization - 📊 Interval-based progress summaries - 💾 Persistent knowledge base storage - 📝 Markdown-formatted summaries - 🎨 Color-coded terminal output for better visibility - 🔄 Resume capability with existing knowledge base - ⚙️ Configurable analysis intervals and test modes - 🚫 Smart content filtering (skips TOC, index pages, etc.) - 📂 Organized directory structure for outputs ---- Want to level up your Cursor game? I’ve created a 28-chapter course on mastering Cursor. Check it out via the link in my bio! featuring 19 hours of content where we build full apps from scratch in each chapter. each chapter is short and independent and designed to get your started quickly.
echo.hive81,236 views • 1 year ago
Bio inspired Hebbian probabilistic network learns in less than 5 minutes from a super sparse single reward per episode! also has imitation learning (manual control) system has 3 parallel competing networks which get sensory input from a 360 vision (27-direction sensory neuron array) link to code in comment each sub-network is responsible for a single motor action: forward, left and right. at each step whichever section has most neurons firing wins neurons fire probabilistically and mark themselves with a time-decay tag which happens when a neuron fires and diminishes with time. you can see this " tag countdown" on each neuron when a reward is attained(eating the cheese) eligible connections gets strengthened I included 2 runs in the video first was 15 minutes in real time and second was 5 minutes. red plot is the rolling average of last 10 time to cheese. it is really not possible for agent to achieve full control due to probabilistic neural firing. that is why it has to learn while jittering all over the place, which in itself is interesting in manual mode you can guide the cheese by stimulating its motor control networks ( still probabilistically ) and the rewards will still work ✅ Biologically Plausible Features: Stochastic firing (neurons in the brain fire probabilistically) Reward-based learning (dopamine-like neuromodulation) Hebbian plasticity (well-established biological mechanism) Eligibility traces (biological neurons have temporal credit assignment) Sparse sensory encoding (similar to place cells, grid cells) Competitive action selection (basal ganglia architecture) No backpropagation (which is biologically implausible) ❌ Missing Biological Features: No recurrent connections (real brains have extensive feedback loops) No inhibitory neurons (GABAergic neurons are ~20% of cortex) No spike timing (simplified from true spiking dynamics) Uniform layer structure (biological networks are more heterogeneous) Simple weight updates (real synaptic plasticity is more complex)
echo.hive33,638 views • 7 months ago
Learn Cursor in 5 minutes! Join my Cursor course which includes 17 videos building full apps from scratch with project files. Link in comment Repost this tweet and I will send the “how to build an AI scientist” video from my Cursor Course This will serve as an example taste of what you can expect from my 1000x Cursor course. I will DM you with a download link after you repost. #CursorAI #Cursor
echo.hive73,935 views • 1 year ago
Cursor gets Perplexity API search tool while being FULLY AUTO CONTROLLED by Claude Computer use. So cursor performs web research! cursorrules and system prompt in comment Tool is just a function in tools .py file and you can add more tools of any kind as long as you update the cursorrules This was tricky because we are restricting comp use to certain coordinates for upper and lower input boxes. Luckily "run command" button also appears at the lower input box and its coordinates are fixed so we can use that to out advantage! You can download the source code by becoming a patron. link in comment essentially comp use is restricted to a limited action space and only 3 coordinates it can use. this way we are trying to ensure it doesn't do things we don't want and also make less mistakes(although mistakes do happen still both from comp use and from cursor although not much -------- If you are interested in learning how to use Cursor more in depth, I have a 28 chapter 1000x Cursor Course building full apps from scratch with 19 hours of content. link is in my bio if you are interested inspired partially by Stijn and Ryan O'Donnell (stijn mentioned in DM)
echo.hive49,245 views • 1 year ago