Skip to content

Add QTensor row dequantization helper#3562

Open
rabbitson87 wants to merge 1 commit into
huggingface:mainfrom
rabbitson87:pr/quantized-dequantize-rows
Open

Add QTensor row dequantization helper#3562
rabbitson87 wants to merge 1 commit into
huggingface:mainfrom
rabbitson87:pr/quantized-dequantize-rows

Conversation

@rabbitson87

Copy link
Copy Markdown

Summary

  • add QTensor::dequantize_rows for selecting rows from 2D quantized tensors without dequantizing the full tensor
  • add a Metal get_rows launcher using the existing quantized Metal kernels
  • cover CPU row selection and out-of-bounds behavior in quantized tests

Tests

  • cargo test -p candle-core qtensor_dequantize_rows

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant