Browser-local hardware lab
FP Quantization Visualizer
A bit-level visualizer for hardware validation and quantization debugging. Switch FP32, FP16, BFloat16, or INT8, then flip any bit to recompute hex, fields, and real value.
Local
Format And Value
FPClick any bit cell to flip and recompute instantly.
Interactive Bit Map
32 bitsBinary
Field Decode
- Sign
- Exponent
- Mantissa
- Decimal
Practical guide
FP Quantization Visualizer
Inspect how sign, exponent, mantissa, and integer quantization bits affect numeric values without sending any data out of the browser.
How to use
- Choose a numeric format such as FP32, FP16, BFloat16, or INT8.
- Enter a decimal value or edit the bit pattern directly to see the recomputed representation.
- Flip bits to observe sign changes, exponent movement, mantissa precision, saturation, and rounding behavior.
- Copy the hexadecimal or binary representation for documentation, debugging notes, or low-level tests.
Common use cases
- Teaching how floating-point formats divide a value into sign, exponent, and fraction fields.
- Debugging quantization edge cases around zero, denormals, saturation, or small magnitude changes.
- Preparing examples for NPU, DSP, shader, or inference documentation.
Privacy note
Entered numbers and bit patterns stay inside browser memory. The page does not upload values, logs, or examples.