https://docs.google.com/forms/d/1wtc0ttL9HLqU6aC3D8QhOBm9x5srCfndK5GXcAV0Stc/edit#responses
Mahalo
SIGNATURE:
Clifford "RAY" Hackett I founded www.adapt.org in 1980 it now has over 50 million members.
$500 of material=World’s fastest hydrofoil sailboat. http://sunrun.biz
Example: 45-second reel on the 2-minute habit rule
Hook (0–4s)
“You’re not lazy. Your habits are built to fail.”
1. Setup
If you’ve started working out, reading, or journaling and quit within two weeks, this is why it keeps happening.
2. Re-hook
The fix isn’t more willpower. It’s making the habit so small it feels stupid not to do it.
3. Body
Pick the habit. Shrink it to two minutes.
Want to run? Put on your shoes and walk to the door.
Want to write? Open the doc and type one sentence.
Want to meditate? Sit down and take three breaths.
Do only that. Stop when the two minutes are up.
4. Climax
Once showing up is automatic, the real work starts happening on its own. That’s how most people who actually keep habits begin—not with hour-long sessions.
5. CTA
Pick one habit. Do the two-minute version tonight. Comment “2MIN” if you’re trying it
Mahalo
SIGNATURE:
Clifford "RAY" Hackett I founded www.adapt.org in 1980 it now has over 50 million members.
$500 of material=World’s fastest hydrofoil sailboat. http://sunrun.biz
name: nanoGPT Training
on:
workflow_dispatch:
jobs:
train:
runs-on: ubuntu-latest
timeout-minutes: 360
steps:
– name: Checkout repository
uses: actions/checkout@v4
– name: Set up Python
uses: actions/checkout@v4
with:
python-version: ‘3.10’
– name: Cache dependencies
uses: actions/cache@v4
with:
path: ~/.cache/pip
key: ${{ runner.os }}-pip-${{ hashFiles(‘**/requirements.txt’) }}
restore-keys: |
${{ runner.os }}-pip-
– name: Install Dependencies
run: |
python -m pip install –upgrade pip
pip install torch numpy transformers datasets tiktoken tqdm
– name: Prepare Dataset
run: |
python data/shakespeare_char/prepare.py
– name: Train Model (CPU)
run: |
python train.py config/train_shakespeare_char.py –device=cpu –compile=False
– name: Generate Sample Output
run: |
python sample.py –out_dir=out-shakespeare-char –device=cpu > generated_sample.txt
cat generated_sample.txt
– name: Upload Training Artifacts
uses: actions/upload-artifact@v4
with:
name: nanogpt-checkpoint-and-sample
path: |
out-shakespeare-char/
generated_sample.txt
Mahalo
SIGNATURE:
Clifford "RAY" Hackett I founded www.adapt.org in 1980 it now has over 50 million members.
$500 of material=World’s fastest hydrofoil sailboat. http://sunrun.biz
cat << 'EOF' > setup.sh #!/bin/bash set -e echo "=== Installing Core Dependencies ===" /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)" eval "$(/opt/homebrew/bin/brew shellenv)" brew install go git ollama echo "=== Starting Ollama Service ===" brew services start ollama sleep 5 echo "=== Pulling Free Local LLM Models ===" ollama pull qwen2.5-coder ollama pull llama3.3 echo "=== Setting Up PicoClaw Repository ===" cd ~ if [ -d "picoclaw" ]; then cd picoclaw && git pull else git clone https://github.com/ram133/picoclaw.git && cd picoclaw fi echo "=== Compiling PicoClaw Binary ===" make build echo "=== Initializing PicoClaw Configuration ===" ./build/picoclaw onboard mkdir -p ~/.picoclaw cat << 'CONFIG' > ~/.picoclaw/config.json { "agents": { "defaults": { "provider": "ollama", "model": "qwen2.5-coder" } }, "providers": { "ollama": { "api_base": "http://localhost:11434/v1", "api_key": "" } } } CONFIG echo "=== Testing PicoClaw Execution ===" ./build/picoclaw agent -m "Write a single-page HTML pricing table" echo "=== Setup Complete ===" EOF chmod +x setup.sh ./setup.sh
Mahalo
SIGNATURE:
Clifford "RAY" Hackett I founded www.adapt.org in 1980 it now has over 50 million members.
$500 of material=World’s fastest hydrofoil sailboat. http://sunrun.biz
https://ram133.github.io/nanoGPT/
Mahalo
SIGNATURE:
Clifford "RAY" Hackett I founded www.adapt.org in 1980 it now has over 50 million members.
$500 of material=World’s fastest hydrofoil sailboat. http://sunrun.biz
https://github.com/karpathy/micrograd[cite: 2]
Mahalo
SIGNATURE:
Clifford "RAY" Hackett I founded www.adapt.org in 1980 it now has over 50 million members.
$500 of material=World’s fastest hydrofoil sailboat. http://sunrun.biz
Mahalo
SIGNATURE:
Clifford "RAY" Hackett I founded www.adapt.org in 1980 it now has over 50 million members.
$500 of material=World’s fastest hydrofoil sailboat. http://sunrun.biz
|
Fri, Sep 25, 12:29 AM (1 day ago) | |||
|
||||
import math
import random
# ==========================================
# 1. Engine Utilities & Network Implementation
# ==========================================
def exp(z):
return math.exp(z)
class NeuralNetwork:
def __init__(self, input_dim, hidden_dim, output_dim, act):
self.act = act
# Initialize random weights and biases
self.w1 = [[random.uniform(-1, 1) for _ in range(hidden_dim)] for _ in range(input_dim)]
self.b1 = [random.uniform(-1, 1) for _ in range(hidden_dim)]
self.w2 = [[random.uniform(-1, 1) for _ in range(output_dim)] for _ in range(hidden_dim)]
self.b2 = [random.uniform(-1, 1) for _ in range(output_dim)]
def forward(self, x):
# Hidden layer pass
self.h_raw = [
sum(x[i] * self.w1[i][j] for i in range(len(x))) + self.b1[j]
for j in range(len(self.b1))
]
self.h_out = [self.act(z) for z in self.h_raw]
# Output layer pass
self.o_raw = [
sum(self.h_out[i] * self.w2[i][j] for i in range(len(self.h_out))) + self.b2[j]
for j in range(len(self.b2))
]
# Output uses sigmoid for classification mapping (0 to 1)
self.o_out = [1 / (1 + math.exp(-z)) for z in self.o_raw]
return self.o_out
def learn(xs, ys, hidden=2, act=None, lr=0.1, epochs=10000):
input_dim = len(xs[0])
output_dim = len(ys[0])
net = NeuralNetwork(input_dim, hidden, output_dim, act)
for _ in range(epochs):
for x, y in zip(xs, ys):
pred = net.forward(x)
# Compute loss error and adjust weights numerically
for i in range(len(x)):
for j in range(hidden):
error = pred[0] – y[0]
net.w1[i][j] -= lr * error * x[i]
for j in range(hidden):
for k in range(output_dim):
error = pred[0] – y[0]
net.w2[j][k] -= lr * error * net.h_out[j]
return net
def grade(net, xs, ys):
total_loss = 0.0
correct = 0
for x, y in zip(xs, ys):
pred = net.forward(x)
# Mean Squared Error
total_loss += sum((p – t) ** 2 for p, t in zip(pred, y))
# Binary rounding check
binary_pred = 1 if pred[0] >= 0.5 else 0
if binary_pred == y[0]:
correct += 1
accuracy = correct / len(xs)
mse = total_loss / len(xs)
return f"Accuracy: {accuracy * 100:.0f}%, MSE Loss: {mse:.4f}"
# ==========================================
# 2. Activations, Dataset & Benchmark Execution
# ==========================================
def none(z):
return z
def sigmoid(z):
return 1 / (1 + math.exp(-z))
def relu(z):
return max(0, z)
def neuron(x, w, b, act):
z = sum(xi * wi for xi, wi in zip(x, w)) + b
return act(z)
xs = [[0, 0], [0, 1], [1, 0], [1, 1]]
ys = [[0], [1], [1], [0]]
for act in (none, sigmoid, relu):
net = learn(xs, ys, hidden=2, act=act)
print(act.__name__, grade(net, xs, ys))
Mahalo
SIGNATURE:
Clifford "RAY" Hackett I founded www.adapt.org in 1980 it now has over 50 million members.
$500 of material=World’s fastest hydrofoil sailboat. http://sunrun.biz
Mahalo
SIGNATURE:
Clifford "RAY" Hackett I founded www.adapt.org in 1980 it now has over 50 million members.
$500 of material=World’s fastest hydrofoil sailboat. http://sunrun.biz
Mahalo
SIGNATURE:
Clifford "RAY" Hackett I founded www.adapt.org in 1980 it now has over 50 million members.
$500 of material=World’s fastest hydrofoil sailboat. http://sunrun.biz
