Improve quick start guide to avoid dependence

[Freed-Wu]

During trying quick start guide, I find:

>>> import torchao
[11/14/25 21:06:27] WARNING  Skipping import of cpp extensions due to incompatible torch version 2.9.1 for torchao version 0.15.0.dev20251114+cu130             Please see            __init__.py:94
                             https://github.com/pytorch/ao/issues/2919 for more info
/home/wzy/.local/lib/python3.13/site-packages/torchao/quantization/quant_api.py:2439: SyntaxWarning: invalid escape sequence '\.'
  * regex for parameter names, must start with `re:`, e.g. `re:language\.layers\..+\.q_proj.weight`.
>>> from fbgemm_gpu.experimental.gen_ai.quantize import int4_row_quantize_zp, pack_int4
[11/14/25 21:08:17] ERROR    Could not load the library 'experimental/gen_ai/fbgemm_gpu_experimental_gen_ai.so'!                                                                     __init__.py:103


                             Could not load this library: /home/wzy/.local/lib/python3.13/site-packages/fbgemm_gpu/experimental/gen_ai/fbgemm_gpu_experimental_gen_ai.so
>>> import torch
>>> torch.ops.load_library('/home/wzy/.local/lib/python3.13/site-packages/fbgemm_gpu/experimental/gen_ai/fbgemm_gpu_experimental_gen_ai.so')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python3.13/site-packages/torch/_ops.py", line 1490, in load_library
    raise OSError(f"Could not load this library: {path}") from e
OSError: Could not load this library: /home/wzy/.local/lib/python3.13/site-packages/fbgemm_gpu/experimental/gen_ai/fbgemm_gpu_experimental_gen_ai.so

uv pip install --prefix ~/.local torchao fbgemm-gpu-genai --index-url https://download.pytorch.org/whl/nightly/cu130

I recommend to avoid extra dependence (fbgemm) in quick start guide, at least to quickly let new user have something which can run.

Looks like only PLAIN and PRESHUFFLED of Int4PackingFormat use fbgemm. perhaps can change other format.

[jerryzh168]

yeah makes sense, thanks for the feedback, also sorry I just saw this, it's probably better to open an issue instead