---

Screenshots

---

Details

---

---

MiniGPT Lab Main Features

• Sampling
  Changes the probability distribution slightly and picks a word based on these chances.
• Attention
  Helps the model focus on important earlier tokens when predicting the next word.
• Inspector
  Allows you to click on parts to see detailed data like tokens, vectors, and attention weights.
• Embeddings
  Converts each token number into a special vector that shows its meaning.
• Parts Tray
  You can add different parts to build your mini ChatGPT engine or use a demo to fill everything automatically.
• Tokenization
  Splits raw text into small parts called tokens that the model can understand.
• Engine Chassis
  The main part that holds all modules; you can add or fill parts to build your engine.
• Run the engine
  Starts the process to see how text is transformed at each step and understand why each step is important.
• Workbench Mode
  You can install parts, run the engine with your prompt, and watch how text changes step by step.
• Positional Encoding
  Adds information about the position of each token so the model knows the order of words.
• Autoregressive Output
  Adds the chosen word to the sequence and repeats the process for the next word.
• Logits / Probabilities
  Calculates scores for all possible next words and turns them into probabilities.
• Vocabulary / Token IDs
  Finds each token in a list and replaces it with a number that represents it.

---

Verified Badge

---

MiniGPT Lab Alternatives

---

Technologies Used

View All Technologies