← Blog
wllama + GPT-2 Q2_K — Browser Inference
local WASM · no server
Model
—
Status
Idle — pick a model and Load
⬇ Load Model
Sampling
Max tokens
64
Temperature
0.80
Top-K
40
Top-P
0.95
GPT-2 is a completion model.
It works best with list-continuation prompts ("Item 1, Item 2, ") not chat-style questions.
Playground
Blog Name Generator
Console
Prompt — GPT-2 continues from here
Popular machine learning blogs: "Distill", "The Gradient", "Import AI", "The Batch", "Interconnects", "Loss Landscapes", "Ahead of AI", "Latent Space", "
Output
Load the model, then click Generate.
These names cycle on the blog index every 3 s. Highlighted = currently active.
Load the model then click
Generate more
to let GPT-2 suggest new ones.
✦ Generate more names with GPT-2