Fixes after initial release

- Rename test qmpi -> cgx
- Uncomment dequantization
- Increment package release version

setup.py

@@ -66,15 +66,15 @@
 
 setup(name='pytorch_cgx',
       packages=['cgx_utils'],
-      version='0.1.0',
+      version='0.1.1',
       description='pytorch extension adding a backend '
                   'supporting allreduce of quantized buffers.',
       long_description=long_description,
       long_description_content_type='text/markdown',
       author='Ilia Markov',
       author_email='[email protected]',
       url='https://github.com/IST-DASLab/torch_cgx/',
-      download_url="https://github.com/IST-DASLab/torch_cgx/archive/refs/tags/v0.1.0.tar.gz",
+      download_url="https://github.com/IST-DASLab/torch_cgx/archive/refs/tags/v0.1.1.tar.gz",
       ext_modules=[cpp_extension.CUDAExtension('torch_cgx', sources=src,
                                               include_dirs=include_dirs,
                                               extra_compile_args={'cxx': cxx_compile_args, 'nvcc': nvcc_compile_args},

src/common/compression/cuda_compression_operations.cu

@@ -794,8 +794,8 @@ void CUDA_dequantize_maxmin(const unsigned char *input_data,
   int num_threads = THREADS_PER_BLOCK_DECOMPRESS;
   int num_blocks =
       BLOCKS_PER_GRID((num_elems + PACK_SIZE - 1) / PACK_SIZE, num_threads);
-//  DEQUANTIZE1<T, ADD>(input, meta_info, output, num_elems, bucket_size, bits,
-//                      stream, num_blocks, num_threads);
+  DEQUANTIZE1<T, ADD>(input, meta_info, output, num_elems, bucket_size, bits,
+                      stream, num_blocks, num_threads);
 }
 
 /* Functions declarations */