forked from mrq/bitsandbytes-rocm
36 lines
1.2 KiB
Markdown
36 lines
1.2 KiB
Markdown
### 0.0.21
|
|
- Ampere, RTX 30 series GPUs now compatible with the library.
|
|
|
|
### 0.0.22:
|
|
|
|
- Fixed an error where a `reset_parameters()` call on the `StableEmbedding` would lead to an error in older PyTorch versions (from 1.7.0).
|
|
|
|
### 0.0.23:
|
|
|
|
Bugs:
|
|
- Unified quantization API: each quantization function now returns `Q, S` where `Q` is the quantized tensor and `S` the quantization state which may hold absolute max values, a quantization map or more. For dequantization all functions now accept the inputs `Q, S` so that `Q` is dequantized with the quantization state `S`.
|
|
- Fixed an issue where the CUDA 11.1 binary was not compiled with the right headers
|
|
|
|
API changes:
|
|
- Block-wise quantization for optimizers now enabled by default
|
|
|
|
Features:
|
|
- Block-wise quantization routines now support CPU Tensors.
|
|
|
|
|
|
### 0.0.24:
|
|
|
|
- Fixed a bug where a float/half conversion led to a compilation error for CUDA 11.1 on Turning GPUs.
|
|
- removed Apex dependency for bnb LAMB
|
|
|
|
### 0.0.25:
|
|
|
|
Features:
|
|
- Added `skip_zeros` for block-wise and 32-bit optimizers. This ensures correct updates for sparse gradients and sparse models.
|
|
- Added support for Kepler GPUs. (#4)
|
|
|
|
Bug fixes:
|
|
- fixed "undefined symbol: \_\_fatbinwrap_38" error for P100 GPUs on CUDA 10.1 (#5)
|
|
|
|
|