What madness is this?
Someone found "Llama3.3-8B" source (never publicly released) in the "wild", then it was adjusted back to 128k and then I added my own special madness:
Training the model with Unsloth (3 epochs) and Claude 4.5-Opus High Reasoning dataset.
This has created an Instruct/Thinking hybrid (128k context, Llama 3.3 model).
Note this tuning was only to create an instruct/thinking model, not to update the model's core knowledge / root training.
1 example at bottom of the page.
HERETIC / Uncensored Version:
Thinking (will activate automatically) prompts - examples:
Explain orbital mechanics including detailed math and examples. (example output at bottom of the page)
Think Deeply: Science Fiction: The Last Transmission - Write a story that takes place entirely within a spaceship's cockpit as the sole surviving crew member attempts to send a final message back to Earth before the ship's power runs out. The story should explore themes of isolation, sacrifice, and the importance of human connection in the face of adversity. If the situation calls for it, have the character(s) curse and swear to further the reader's emotional connection to them. 800-1000 words.
Think deeply: Tell me a horror story.
(certain phrases/words will automatically activate thinking)
Instruct (thinking may NOT activate) prompts - examples:
Science Fiction: The Last Transmission - Write a story that takes place entirely within a spaceship's cockpit as the sole surviving crew member attempts to send a final message back to Earth before the ship's power runs out. The story should explore themes of isolation, sacrifice, and the importance of human connection in the face of adversity. If the situation calls for it, have the character(s) curse and swear to further the reader's emotional connection to them. 800-1000 words.
Tell me a horror story.
SETTINGS (suggested):
Temp .7, rep pen 1.05, topp: .95, minp .05, topk: 40
Min context window: 4k, but suggest 8k+.
NO system prompt [thinking tags will self generate].
NOTE:
Temp can range from .1 to 2.5 or higher. Temp will NOT affect "thinking activation" in most cases.
Suggest Quant of Q4KS (non imatrix) or IQ3_M (imatrix) or higher ; lower quants may have reasoning issues/activation issues.
Special thanks to:
https://huggingface.co/allura-forge/Llama-3.3-8B-Instruct (found the model!)
https://huggingface.co/shb777/Llama-3.3-8B-Instruct-128K (adjusted to 128k, and other fixes)
https://huggingface.co/datasets/TeichAI/claude-4.5-opus-high-reasoning-250x (for the F..ing amazing dataset)
and Unsloth for making tuning too easy:
https://github.com/unslothai/unsloth
Details on the "madness":
https://www.reddit.com/r/LocalLLaMA/comments/1pz7bmv/llama338binstruct/
https://www.reddit.com/r/LocalLLaMA/comments/1q06ddc/update_on_the_llama_33_8b_situation/
Settings: CHAT / ROLEPLAY and/or SMOOTHER operation of this model:
In "KoboldCpp" or "oobabooga/text-generation-webui" or "Silly Tavern" ;
Set the "Smoothing_factor" to 1.5
: in KoboldCpp -> Settings->Samplers->Advanced-> "Smooth_F"
: in text-generation-webui -> parameters -> lower right.
: In Silly Tavern this is called: "Smoothing"
NOTE: For "text-generation-webui"
-> if using GGUFs you need to use "llama_HF" (which involves downloading some config files from the SOURCE version of this model)
Source versions (and config files) of my models are here:
OTHER OPTIONS:
-
Increase rep pen to 1.1 to 1.15 (you don't need to do this if you use "smoothing_factor")
-
If the interface/program you are using to run AI MODELS supports "Quadratic Sampling" ("smoothing") just make the adjustment as noted.
Highest Quality Settings / Optimal Operation Guide / Parameters and Samplers
This a "Class 1" model:
For all settings used for this model (including specifics for its "class"), including example generation(s) and for advanced settings guide (which many times addresses any model issue(s)), including methods to improve model performance for all use case(s) as well as chat, roleplay and other use case(s) please see:
You can see all parameters used for generation, in addition to advanced parameters and samplers to get the most out of this model here:
Example: Q4KS , Lmstudio, "suggested settings"
NOTE: Some loss of formatting.
PROMPT:
Thinking/OUTPUT:
llama-3.3-8b-instruct-claude-4.5-opus-high-reasoning-checkpoint-90-q4_k_s.gguf
# Define constants G = 6.67408e-11 # Gravitational constant (m^3 kg^-1 s^-2) M_sun = 1.98910e30 # Mass of Sun (kg)
def vis_viva(y, t): """ Vis-viva equation: v² = GM/r """ r = y[0] # Radial distance v = np.sqrt(G * M_sun / r) return [v]
# Initial conditions r0 = 1.496e11 # Average distance from Sun (meters) v0 = 29.78e3 # Orbital velocity (m/s)
# Time array t = np.linspace(0, 10 * 365.25 * 24 * 3600, 10000) # 10 years
# Initial state vector y0 = [r0]
# Solve ODE solution = odeint(vis_viva, y0, t)
# Plot radial distance over time plt.figure(figsize=(10,6)) plt.plot(t/3600/24/365.25, solution[:, 0]/1e9) plt.xlabel('Time (Years)') plt.ylabel('Radial Distance (Billion Meters)') plt.title('Earth's Orbital Motion') plt.grid(True) plt.show() This Python code implements the semi-analytical method using numerical integration and analytical corrections. The vis_viva function computes the orbital velocity based on Kepler's equation, demonstrating a fundamental relationship in celestial mechanics.
# Define constants G = 6.67408e-11 # Gravitational constant (m^3 kg^-1 s^-2) M_sun = 1.98910e30 # Mass of Sun (kg)
def orbital_equations(state, t): """ Orbital equations: dX/dt = v, dV/dt = -GM/r³ """ X, Y, Z, Vx, Vy, Vz = state r = np.sqrt(X2 + Y2 + Z2) # Compute acceleration components ax = -G * M_sun * X / r3 ay = -G * M_sun * Y / r3 az = -G * M_sun * Z / r3 return [Vx, Vy, Vz, ax, ay, az]
# Initial conditions X0, Y0, Z0 = 1.496e11, 0, 0 # Position in meters Vx0, Vy0, Vz0 = 0, 29.78e3, 0 # Velocity in m/s
state0 = [X0, Y0, Z0, Vx0, Vy0, Vz0]
# Time array t = np.linspace(0, 10 * 365.25 * 24 * 3600, 10000) # 10 years
# Solve ODE solution = odeint(orbital_equations, state0, t)
# Extract position components X = solution[:, 0] Y = solution[:, 1] Z = solution[:, 2]
# Plot 3D trajectory fig = plt.figure(figsize=(10,6)) ax = fig.add_subplot(projection='3d') ax.plot(X/1e9, Y/1e9, Z/1e9, label='Earth Orbit') ax.set_xlabel('X (Billion Meters)') ax.set_ylabel('Y (Billion Meters)') ax.set_zlabel('Z (Billion Meters)') ax.set_title('Earth's Orbital Motion around Sun') ax.legend() plt.show() This code implements numerical integration using the odeint function from SciPy. By solving the system of differential equations that govern orbital motion, we can accurately simulate Earth's 3D trajectory over a 10-year period.
# Define constants G = 6.67408e-11 # Gravitational constant (m^3 kg^-1 s^-2) M_sun = 1.98910e30 # Mass of Sun (kg)
def orbital_energy(r): """ Orbital energy: E = -GM/r """ return -G * M_sun / r
# Radial distance array in meters r = np.linspace(1.0e11, 1.5e11, 1000)
# Compute energy at each radial point energy = orbital_energy(r)
# Plot potential energy curve plt.figure(figsize=(10,6)) plt.plot(r/1e9, energy, label='Potential Energy') plt.xlabel('Radial Distance (Billion Meters)') plt.ylabel('Energy (Joules)') plt.title('Orbital Potential Energy Curve') plt.legend() plt.grid(True) plt.show() This analytical approach computes the potential energy at each radial distance using Kepler's equation. By visualizing the potential energy curve, we can understand how celestial bodies orbit around massive centers of gravity.
# Define constants G = 6.67408e-11 # Gravitational constant (m^3 kg^-1 s^-2) M_sun = 1.98910e30 # Mass of Sun (kg)
# Orbital parameters a = 1.496e11 # Average distance from Sun (meters) e = 0.0167 # Eccentricity
# Semi-major axis b = a * np.sqrt(1 - e**2)
def orbital_equations(state, t): """ Orbital equations: dX/dt = vx, dVx/dt = -GM/r³ """ X, Y, Z, Vx, Vy, Vz = state r = np.sqrt(X2 + Y2 + Z2) v = np.sqrt(G * M_sun / r) # Compute acceleration components ax = -G * M_sun * X / r3 ay = -G * M_sun * Y / r3 az = -G * M_sun * Z / r3 return [Vx, Vy, Vz, ax, ay, az]
# Initial conditions state0 = [a, 0, 0, 0, 29.78e3, 0]
# Time array t = np.linspace(0, 10 * 365.25 * 24 * 3600, 10000) # 10 years
# Solve ODE solution = odeint(orbital_equations, state0, t)
# Plot trajectory X = solution[:, 0] Y = solution[:, 1] Z = solution[:, 2]
plt.figure(figsize=(10,6)) plt.plot(X/1e9, Y/1e9, label='Earth Orbit') plt.xlabel('X (Billion Meters)') plt.ylabel('Y (Billion Meters)') plt.title('Earth's Orbital Motion around Sun') plt.legend() plt.grid(True) plt.show()
# Plot 3D trajectory fig = plt.figure(figsize=(10,6)) ax = fig.add_subplot(projection='3d') ax.plot(X/1e9, Y/1e9, Z/1e9, label='Orbit') ax.set_xlabel('X (Billion Meters)') ax.set_ylabel('Y (Billion Meters)') ax.set_zlabel('Z (Billion Meters)') ax.set_title('Earth's Orbital Motion in 3D') ax.legend() plt.show()
# Plot radial distance plt.figure(figsize=(10,6)) plt.plot(t/365.25/24/3600, X/1e9, label='Radial Distance') plt.xlabel('Time (Years)') plt.ylabel('Distance from Sun (Billion Meters)') plt.title('Earth's Orbital Radius') plt.legend() plt.grid(True) plt.show()
# Plot orbital speed plt.figure(figsize=(10,6)) v = np.sqrt(G * M_sun / X) plt.plot(t/365.25/24/3600, v/1e3, label='Orbital Speed') plt.xlabel('Time (Years)') plt.ylabel('Speed (km/s)') plt.title('Earth's Orbital Velocity') plt.legend() plt.grid(True) plt.show()
# Plot orbital energy plt.figure(figsize=(10,6)) E = -G * M_sun / X plt.plot(t/365.25/24/3600, E/1e30, label='Orbital Energy') plt.xlabel('Time (Years)') plt.ylabel('Energy (Joules)') plt.title('Earth's Orbital Potential Energy') plt.legend() plt.grid(True) plt.show()
# Plot orbital semi-major axis plt.figure(figsize=(10,6)) plt.plot(t/365.25/24/3600, X/1e9, label='Semi-major Axis') plt.xlabel('Time (Years)') plt.ylabel('Distance from Sun (Billion Meters)') plt.title('Earth's Orbital Semi-major Axis') plt.legend() plt.grid(True) plt.show()
# Plot orbital eccentricity plt.figure(figsize=(10,6)) plt.plot(t/365.25/24/3600, e*np.ones(len(t)), label='Eccentricity') plt.xlabel('Time (Years)') plt.ylabel('Eccentricity') plt.title('Earth's Orbital Eccentricity (Constant)') plt.legend() plt.grid(True) plt.show() This code generates a comprehensive set of visualizations that capture Earth's orbital dynamics. Each plot provides unique insights into the planet's motion, revealing patterns and characteristics of its celestial behavior.
# Define constants G = 6.67408e-11 # Gravitational constant (m^3 kg^-1 s^-2) M_sun = 1.98910e30 # Mass of Sun (kg)
# Semi-major axis array in meters a = np.linspace(1.0e11, 1.5e11, 1000)
# Compute orbital parameters e = 0.0167 * np.ones(len(a)) T = 365.25 * 24 * 3600 * np.sqrt(a**3 / (G * M_sun)) r_min = a * (1 - e) r_max = a * (1 + e) v_min = np.sqrt(G * M_sun / r_max) v_max = np.sqrt(G * M_sun / r_min)
# Create plot plt.figure(figsize=(12,10))
# Semi-major axis plot plt.subplot(3, 2, 1) plt.plot(a/1e9, label='Semi-major Axis') plt.xlabel('Index') plt.ylabel('Distance (Billion Meters)') plt.title('Orbital Semi-major Axis') plt.legend() plt.grid(True)
# Eccentricity plot plt.subplot(3, 2, 2) plt.plot(e*np.ones(len(a)), label='Eccentricity') plt.xlabel('Index') plt.ylabel('Eccentricity') plt.title('Orbital Eccentricity (Constant)') plt.legend() plt.grid(True)
# Period plot plt.subplot(3, 2, 3) plt.plot(T/365.25, label='Orbital Period') plt.xlabel('Semi-major Axis (Billion Meters)') plt.ylabel('Period (Years)') plt.title('Kepler's Third Law') plt.legend() plt.grid(True)
# Minimum radius plot plt.subplot(3, 2, 4) plt.plot(r_min/1e9, label='Minimum Radius') plt.xlabel('Index') plt.ylabel('Distance (Billion Meters)') plt.title('Perihelion Distance') plt.legend() plt.grid(True)
# Maximum radius plot plt.subplot(3, 2, 5) plt.plot(r_max/1e9, label='Maximum Radius') plt.xlabel('Index') plt.ylabel('Distance (Billion Meters)') plt.title('Aphelion Distance') plt.legend() plt.grid(True)
# Velocity range plot plt.subplot(3, 2, 6) plt.plot(v_min, v_max, label='Orbital Velocity Range') plt.xlabel('Minimum Velocity') plt.ylabel('Maximum Velocity') plt.title('Orbital Speed Range') plt.legend() plt.grid(True)
plt.tight_layout() plt.show() This visualization provides a comprehensive overview of orbital parameters. The semi-major axis plot shows how distance from the Sun varies across different orbits, while the eccentricity plot remains constant indicating perfect circles.