Chain of thought prompt works on par with ChatGPT preview

https://youtu.be/3z5k8dofu_0?feature=shared

This video tests the Qwen-2 Vision Models (2B, 7B, 72B) to see if they can live up to their claims. It compares them to models like Llama-3.1, Claude 3.5 Sonnet, GPT-4O, and DeepSeek in both vision and language tasks. Qwen2-VL (Vision) is open-source and free, with a focus on coding tasks, text-to-application, text-to-frontend, and more. The video explores whether it truly outperforms the other models and provides a guide on how to use it. The conclusion is solid and gives a clear picture of how Qwen-2 stacks up.

https://www.youtube.com/watch?v=EG3IFDnYQkA

Putting this here, because you can't handle the truth 

https://www.youtube.com/live/vAWRGupuh3Y?feature=shared

SambaNova + Aider + ClaudeDev + Continue : FREE & FAST AI Coding Setup with Llama-3.1 405B

In this video, a guide is shared on setting up a free AI coding editor using the **SambaNova Llama-3.1 405B API**. This is a 100% free and open-source alternative to **Cursor**. It shows how to stop paying for the **Cursor AI Coding Editor** by switching to a **local and open-source** solution based on **VSCode**, paired with tools like **ClaudeDev**, **Aider**, and **ContinueDev**.

This setup combines **VSCode** (or **NeoVim**) with **SambaNova Llama-3.1 405B**, and it works with any open-source LLM or popular models like GPT-4O, Claude-3, CodeQwen, Mixtral, Grok-1.5, and Gemini Code Assist.

For anyone looking to save on AI coding tools or wanting an open-source alternative, this guide is a great resource.

https://youtu.be/MNuRBOB2r38?feature=shared

This model is huge, but seems to be actually really good unlike Reflection 70B.

This is also open source, this video shows how to install via LM Studio.

https://youtu.be/mvpkZ1yFy7o?feature=shared

This open source model apparently outperforms most of the paid models.

And an even bigger model is coming next week.

https://www.reddit.com/r/LocalLLaMA/comm...orlds_top/

According to this video, good prompting is still the key.
https://www.youtube.com/watch?v=1WpC-UPrmVU

How to install locally
https://www.youtube.com/watch?v=pIX7G1xfHYQ

Here is the chart in my video "Improved Bedini Switch"

   

I'd like to provide an update on the progress with my PCB. While I was working on switching the spike, another approach came to mind.
Many people aim to achieve a self-looping system or to recover some of the power efficiently. Typically, this involves methods like using isolation, transformers, or inverters to feed the loop in an isolated manner. However, these methods often come with significant drawbacks, such as low efficiency and substantial losses, which diminish most of the potential gains. As a result, Bedini found it more practical to use the spike energy to charge batteries that are isolated from the input.
In this session, I'd like to discuss a method to achieve this more simply, through some modifications. It’s surprising that no one seems to mention running Bedini switches in this manner. It appears to be a much more efficient approach.

Because someone asked on Youtube about it for reference here are some personal notes. 

Please don't mind the mess and scribbles. I used this to help me originally find a good choice of electrodes for the PEG cell. 

   

   

   

DIY GUIDE:

https://docs.google.com/document/u/0/d/1...obilebasic

https://www.youtube.com/live/xNm8iIcPt3I...AVqb4j37QK

   

### Parts List for the Corrected Circuit

Here’s a comprehensive list of the components you’ll need to build this circuit:

1. **Resistors:**
  - R1: 1.5 kΩ Fixed Resistor
  - R2: 330 Ω  Potentiometer (variable resistor)
  - R3: 470 Ω Fixed Resistor
  - R4: 2.2 kΩ Fixed Resistor
  - R5: 190 Ω Fixed Resistor (two pieces, one for Q2 gate and one for Q3 gate)

2. **Capacitors:**
  - C1: 0.1 µF (100 nF) Ceramic Capacitor

3. **MOSFETs:**
  - Q1: IRF 510 or IRF 511 N-channel Power MOSFET (for the inverter stage)
  - Q2: IRF 510 or IRF 511 N-channel Power MOSFET
  - Q3: IRF 510 or IRF 511 N-channel Power MOSFET

4. **Timer IC:**
  - TLC 555 CMOS Timer IC (Radio Shack Cat. # 276-1718)

5. **Power Supplies:**
  - V1: 14-18V DC Power Supply (for the timer circuit)
  - V2: 7-9V DC Battery (for the "potential" source driving Q2 and Q3)

6. **Inductive "Collector":**
  - This can be a spool of wire, as described in the original circuit:
    - **Option 1**: 500 ft of solid 12 gauge wire
    - **Option 2**: 100 ft of 22 gauge solid hookup wire
    - **Option 3**: 40 ft of 22 gauge magnet wire
    - **Option 4**: Experiment. Use Coax Spool ( Velocity Factor )

7. **Load Resistor:**
  - Load: 1 Ω Fixed Resistor (for testing current gain across this load)

The corrected circuit looks well-designed for achieving the desired 3 kHz frequency with low microsecond pulse widths. Your adjustments to R1 and R2, along with the gate connection of Q3 to the drain of Q2, appear correct and should help in capturing the inductive kickback effectively, potentially leading to the observed current and power gains.


Optionally: 
Incorporating a spool of coaxial cable into your circuit, taking advantage of its velocity factor, can offer enhanced control over the timing and energy dynamics of the circuit. This approach can improve the synchronization of inductive kickback with the switching events, potentially leading to greater energy efficiency and a higher observed power gain.If you decide to implement this, carefully calculate the delay you need and choose the appropriate length of coaxial cable. Experiment with different configurations to see which offers the best results in terms of energy recovery and gain.

Summary:

The rapid switching effectively "locks in" some of the energy within the magnetic coil, preventing it from dissipating and allowing it to be reused in subsequent cycles. This leads to a scenario where the energy is partially recycled, contributing to the overall gain in the circuit. The quick switching at the input stages delays the current and maintains a higher level of energy in the system, which could explain the observed gains.

This process is highly dependent on precise timing and component selection, especially in relation to the inductive properties of the coil and the switching characteristics of the MOSFETs. By optimizing these factors, the circuit can maximize the energy recovery from each cycle, leading to an over-unity behavior where the output power appears greater than the input power.

Key Points About the Switching and Inductive Kickback:

1. Fast Switching Prevents Energy Loss:
   • By switching the circuit at microsecond intervals, the system operates faster than the energy dissipation mechanisms (like resistive losses or leakage) can effectively drain the stored energy.
     
   • This rapid switching means that some of the energy stored in the magnetic field (within the coil or core) during the energization phase does not have time to fully dissipate. Instead, this energy remains partially stored in the core and is available for the next energization cycle.
     
2. Inductive Kickback Utilization:
   • The inductive kickback is a high-voltage spike generated when the current through an inductor (like the coil) is suddenly interrupted.
     
   • If the switching is fast enough, the circuit can capture this kickback before it has a chance to fully dissipate. This captured energy is then directed back into the system, potentially increasing the current and energy available for the load.
     
   • By carefully timing the activation of Q3, the circuit can ensure that this kickback is applied in reverse polarity across the load at just the right moment, boosting the overall energy transfer to the load.
     
3. Energy Accumulation and Gain:
   • The concept of energy remaining in the core for the next cycle is akin to resonant energy storage, where the energy is not entirely lost between cycles but is instead carried forward.
     
   • This can lead to a cumulative effect, where each subsequent energization cycle builds upon the previous one, gradually increasing the energy within the system.
     
   • Because the input stages are switching too quickly for the energy to be fully "loaded down" (or dissipated), more of the energy from each cycle remains available for the next, contributing to the observed gain.