Giter Club home page Giter Club logo

Comments (4)

evshiron avatar evshiron commented on June 23, 2024

pip3 uninstall bitsandbytes should do the trick.

AutoGPTQ just adds support for ROCm days ago, you can do ROCM_VERSION=5.6 pip3 install auto-gptq and it should work (and quantization works too). I haven't used BitsAndBytes for a long time due to bad ROCm support and low performance.

If it doesn't compile, please report back, and try this:

Clone https://github.com/PanQiWei/AutoGPTQ, add these lines:

__host__ __forceinline__ hipblasStatus_t hipblasHgemm(hipblasHandle_t    handle,
                                                               hipblasOperation_t transA,
                                                               hipblasOperation_t transB,
                                                               int                m,
                                                               int                n,
                                                               int                k,
                                                               const half*        alpha,
                                                               const half*        AP,
                                                               int                lda,
                                                               const half*        BP,
                                                               int                ldb,
                                                               const half*        beta,
                                                               half*              CP,
                                                               int                ldc) {
    return hipblasHgemm(handle, transA, transB, m, n, k,
                        reinterpret_cast<const hipblasHalf *>(alpha),
                        reinterpret_cast<const hipblasHalf *>(AP), lda,
                        reinterpret_cast<const hipblasHalf *>(BP), ldb,
                        reinterpret_cast<const hipblasHalf *>(beta),
                        reinterpret_cast<hipblasHalf *>(CP), ldc);
}

to autogptq_cuda/exllama/hip_compat.cuh, and then cd AutoGPTQ && pip3 install -e ..

If there is a need, I might check and update BitsAndBytes builds in this repo for ROCm 5.6 compatibility, but no new feature will be supported.

Btw, I have a plan to update the article about text-generation-webui, but haven't had time to do it.

from rocm_lab.

ZhenyaPav avatar ZhenyaPav commented on June 23, 2024

Removing bitsandbytes removes the issue. ExLlama works fine. Haven't tried auto-gptq yet.

from rocm_lab.

ZhenyaPav avatar ZhenyaPav commented on June 23, 2024

Running pip3 install -e . fails with this error:

  Preparing metadata (setup.py) ... error
  error: subprocess-exited-with-error
  
  × python setup.py egg_info did not run successfully.
  │ exit code: 1
  ╰─> [6 lines of output]
      Traceback (most recent call last):
        File "<string>", line 2, in <module>
        File "<pip-setuptools-caller>", line 34, in <module>
        File "/home/zhenyapav/Projects/text-generation-webui/repositories/AutoGPTQ/setup.py", line 52, in <module>
          CUDA_VERSION = "".join(os.environ.get("CUDA_VERSION", default_cuda_version).split("."))
      AttributeError: 'NoneType' object has no attribute 'split'
      [end of output]
  
  note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

It does work with ROCM_VERSION=5.6 pip install .

from rocm_lab.

evshiron avatar evshiron commented on June 23, 2024

@ZhenyaPav

Yeah. You have to add that ROCM_VERSION=5.6 to build it for ROCm, at least for now.

from rocm_lab.

Related Issues (15)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.