Skip to content
Permalink

Comparing changes

Choose two branches to see what’s changed or to start a new pull request. If you need to, you can also or learn more about diff comparisons.

Open a pull request

Create a new pull request by comparing changes across two branches. If you need to, you can also . Learn more about diff comparisons here.
base repository: ModelCloud/GPTQModel
Failed to load repositories. Confirm that selected base ref is valid, then try again.
Loading
base: v1.5.0
Choose a base ref
...
head repository: ModelCloud/GPTQModel
Failed to load repositories. Confirm that selected head ref is valid, then try again.
Loading
compare: v1.5.1
Choose a head ref
  • 19 commits
  • 22 files changed
  • 6 contributors

Commits on Dec 24, 2024

  1. start v1.5.1-dev cycle (#964)

    @Qubitium
    Qubitium authored Dec 24, 2024
    Configuration menu
    Copy the full SHA
    5712c49 View commit details
    Browse the repository at this point in the history

  2. [CI] upload release logic (#965)

    @CSY-ModelCloud
    CSY-ModelCloud authored Dec 24, 2024
    Configuration menu
    Copy the full SHA
    a235687 View commit details
    Browse the repository at this point in the history

  3. Add QuantizeConfig.device and use. (#950) 

    * normalize device + device_map

* normalize device+device+map+dtype in from_pretrained()

* disallow passing of device/device_map in pretrained(). add `device` to QuantizeConfig.

* if user pass device+device_map and quantizeconfig.device is none, use...else quantizeconfig.device, fall back is auto select

* auto-device logic should not be here

* reduce reliance on accelerate

* remove bad device override

* fix dev not define

* cleanup

* already check device when select_quant_linear

* fix marlin post_init

---------

Co-authored-by: LRL-ModelCloud <[email protected]>
    @Qubitium
    Qubitium and LRL-ModelCloud authored Dec 24, 2024
    Configuration menu
    Copy the full SHA
    55f9d72 View commit details
    Browse the repository at this point in the history

  4. fix hf_select_quant_linear (#966)

    LRL-ModelCloud authored Dec 24, 2024
    Configuration menu
    Copy the full SHA
    da26575 View commit details
    Browse the repository at this point in the history

  5. fix gptq_marlin.py (#967)

    @ZX-ModelCloud
    ZX-ModelCloud authored Dec 24, 2024
    Configuration menu
    Copy the full SHA
    00fde4b View commit details
    Browse the repository at this point in the history

Commits on Dec 25, 2024

  1. fix cuda:0 not a enum device (#968) 

    * fix cuda:0 not a enum device

* use normalize_device
    @CSY-ModelCloud
    CSY-ModelCloud authored Dec 25, 2024
    Configuration menu
    Copy the full SHA
    ebf169a View commit details
    Browse the repository at this point in the history

Commits on Dec 26, 2024

  1. fix marlin info for non-cuda device (#972)

    @Qubitium
    Qubitium authored Dec 26, 2024
    Configuration menu
    Copy the full SHA
    762311e View commit details
    Browse the repository at this point in the history

  2. fix backend str bug (#973) 

    * fix backend str bug

* code review
    @CL-ModelCloud
    CL-ModelCloud authored Dec 26, 2024
    Configuration menu
    Copy the full SHA
    394324a View commit details
    Browse the repository at this point in the history

  3. hf select quant_linear with pack (#969) 

    * hf select quant_linear with pack

* mark pack Optional

* pack default True
    LRL-ModelCloud authored Dec 26, 2024
    Configuration menu
    Copy the full SHA
    c597668 View commit details
    Browse the repository at this point in the history

Commits on Dec 27, 2024

  1. [CI] use new ci docker images (#974) 

    * [CI] all use new v4 docker

* [CI] disable init_unit_tests

* [CI] disable init_unit_tests

* [CI] disable init_unit_tests

* [CI] remove login arg

* [CI] print logs

* [CI] source /opt/pyenv.sh

* [CI] add login arg

* [CI] remove source

* [CI] show list

* [CI] remove torch 2.4

* [CI] max 10

* [CI] clean cache

* [CI] add models/ tests

* update logs

* add cache clean

* [CI] show vram

* [CI] clean cache at earlier step

* [CI] fix env

* [CI] fix env

* [CI] show pip list

* [CI] use v5

* [CI] fix xpu env

* [CI] use 10.0.13.31

* Update release.yml

* fix runs on

* decrease delta to -20%

* install transformers for test_cohere2

* decrease delta to -20%
    @CSY-ModelCloud
    CSY-ModelCloud authored Dec 27, 2024
    Configuration menu
    Copy the full SHA
    bfb75f0 View commit details
    Browse the repository at this point in the history

  2. remove auto select BACKEND.IPEX (#975)

    @CSY-ModelCloud
    CSY-ModelCloud authored Dec 27, 2024
    Configuration menu
    Copy the full SHA
    c296123 View commit details
    Browse the repository at this point in the history

  3. fix autoround device (#976)

    @CSY-ModelCloud
    CSY-ModelCloud authored Dec 27, 2024
    Configuration menu
    Copy the full SHA
    98540eb View commit details
    Browse the repository at this point in the history

Commits on Dec 30, 2024

  1. use enum instead of magic number (#979)

    @CSY-ModelCloud
    CSY-ModelCloud authored Dec 30, 2024
    Configuration menu
    Copy the full SHA
    0bd8417 View commit details
    Browse the repository at this point in the history

Commits on Dec 31, 2024

  1. use new ci docker images (#980) 

    * [CI] all use new v4 docker

* [CI] disable init_unit_tests

* [CI] disable init_unit_tests

* [CI] disable init_unit_tests

* [CI] remove login arg

* [CI] print logs

* [CI] source /opt/pyenv.sh

* [CI] add login arg

* [CI] remove source

* [CI] show list

* [CI] remove torch 2.4

* [CI] max 10

* [CI] clean cache

* [CI] add models/ tests

* update logs

* add cache clean

* [CI] show vram

* [CI] clean cache at earlier step

* [CI] fix env

* [CI] fix env

* [CI] show pip list

* [CI] use v5

* [CI] fix xpu env

* [CI] use 10.0.13.31

* Update release.yml

* fix runs on

* decrease delta to -20%

* install transformers for test_cohere2

* decrease delta to -20%
    @CSY-ModelCloud
    CSY-ModelCloud authored Dec 31, 2024
    Configuration menu
    Copy the full SHA
    ac040bd View commit details
    Browse the repository at this point in the history

  2. fix flash attntion was auto loaded on cpu for pretrained model (#981) 

    * set _attn_implementation_autoset to fix auto loading flash attention on CPU

* add comments
    @CSY-ModelCloud
    CSY-ModelCloud authored Dec 31, 2024
    Configuration menu
    Copy the full SHA
    ce2a96e View commit details
    Browse the repository at this point in the history

  3. fix old transformer doesn't have _attn_implementation_autoset (#982) 

    * fix old transformer doesn't have _attn_implementation_autoset

* use another func to parse Version
    @CSY-ModelCloud
    CSY-ModelCloud authored Dec 31, 2024
    Configuration menu
    Copy the full SHA
    ace914b View commit details
    Browse the repository at this point in the history

  4. fix gptbigcode test (#983)

    @CSY-ModelCloud
    CSY-ModelCloud authored Dec 31, 2024
    Configuration menu
    Copy the full SHA
    5052421 View commit details
    Browse the repository at this point in the history

Commits on Jan 1, 2025

  1. prepare for 2025 1.5.1 release (#984) 

    * prepare for 2025 1.5.1 release

* Update README.md
    @Qubitium
    Qubitium authored Jan 1, 2025
    Configuration menu
    Copy the full SHA
    244f17b View commit details
    Browse the repository at this point in the history

  2. fix version parsing (#985)

    @CSY-ModelCloud
    CSY-ModelCloud authored Jan 1, 2025
    Configuration menu
    Copy the full SHA
    4f18747 View commit details
    Browse the repository at this point in the history
Loading