Ecosyste.ms: Timeline
Browse the timeline of events for every public repo on GitHub. Data updated hourly from GH Archive.
YangWang92 pushed 1 commit to main microsoft/VPTQ
- Update vptq_example.ipynb (#138) fix vptq install in ipynb 171af83
YangWang92 closed a pull request on microsoft/VPTQ
Update vptq_example.ipynb
fix vptq install in ipynb
YangWang92 opened a pull request on microsoft/VPTQ
Update vptq_example.ipynb
fix vptq install in ipynb
YangWang92 created a branch on microsoft/VPTQ
YangWang92-patch-1 - VPTQ, A Flexible and Extreme low-bit quantization algorithm
YangWang92 created a comment on an issue on microsoft/VPTQ
As we have updated quant_config in config.json, and you may need to update `config.json` file or VPTQ installation. Thanks for point this out, and let me update ipynb. ```bash pip install vptq -U ...
YangWang92 created a comment on an issue on microsoft/VPTQ
Hi @Duncan1115, I just noticed that you are the author of LLM-CODEBOOK. I’d like to ask if you are interested in continuing to improve VPTQ? Thank you!
YangWang92 created a comment on an issue on microsoft/VPTQ
I have some early results on FP8, and I’ll share them here this week~ Thanks for following~
YangWang92 created a comment on an issue on microsoft/VPTQ
Yes, I encountered the same issue at the time. It seemed to be related to the data type handling in NCCL. My workaround was to save the unpacked model as a .pt file and use approaches like viewing ...
YangWang92 created a comment on an issue on microsoft/VPTQ
> Firstly, thank you for your great effort to make this project. > > When will the fine-tuning code be released? If it's delayed, could you please let me know the learning rate used in the end-to-...
YangWang92 created a comment on an issue on microsoft/VPTQ
> Firstly, thank you for your great effort to make this project. > > When will the fine-tuning code be released? If it's delayed, could you please let me know the learning rate used in the end-to-...
YangWang92 pushed 12 commits to main VPTQ/compressed-tensors
- Observer Restructure: Remove Observers, `calibration`, and applying `frozen` steps from lifecycle (#189) * temporary... 2b79056
- Clean up observer defaulting logic, better error message (#200) 37df2dd
- apply style and quality (#201) a43dad2
- fix group quant bug (#203) db6ccb2
- bump version (#204) Signed-off-by: Dipika <[email protected]> ff121cc
- skip accelerate tests (#208) 9010372
- remove QuantizationScheme.default_scheme (#202) Signed-off-by: Kyle Sayers <[email protected]> 7103a27
- Allow ModelCompressor.from_pretrained to load from quantization_config, not compression config (#207) a26c03a
- Quantization Scheme Validation (#209) * add model validator to quantization scheme Signed-off-by: Kyle Sayers <ky... c6197ce
- Fix uninitialized variable in quantized compressors (#205) Both compressors have a can_quantize() check, which if ev... 525ef3a
- Implement aliasable mixin and alias activation ordering (#213) * implement aliasable mixin and alias activation orde... 724d5ce
- Revert "Implement aliasable mixin and alias activation ordering (#213)" (#217) This reverts commit 724d5cedc53097107... 8571339
YangWang92 pushed 31 commits to main VPTQ/llm-compressor
- Support Model Offloading Tied Tensors Patch (#872) * update parameter of offloaded modules Signed-off-by: Kyle Sa... c62f2e3
- add advice about dealing with non-invertable hessians (#875) Signed-off-by: Kyle Sayers <[email protected]> a268a25
- seed commit workflow (#877) * seed commit workflow Signed-off-by: andy-neuma <[email protected]> * tickle ... 08125e2
- [Observer Restructure]: Add Observers; Add `calibration` and `frozen` steps to `QuantizationModifier` (#837) * updat... 18e9a9f
- Bugfix get observer from name (#883) Signed-off-by: Rahul Tuli <[email protected]> 60c766f
- BugFix: Fix Sparsity Reload Testing (#882) * fix * fix remaining test cases * add comments * fix d7f09c1
- Use custom unique test names for e2e tests (#892) * Include `testconfig_path` in parsed config data Signed-off-by... 10facf2
- Revert "Use custom unique test names for e2e tests (#892)" (#893) This reverts commit 10facf2633e58778e82d5d53bd661d... 1c0af10
- Move config["testconfig_path"] assignment (#895) * Use custom unique test names for e2e tests (#892) * Include `t... 9c10486
- cap accelerate version to avoid bug (#897) Signed-off-by: Kyle Sayers <[email protected]> f622450
- Fix observing offloaded weight (#896) * load weight within onloading Signed-off-by: Kyle Sayers <kylesayrs@gmail.... f54989e
- Update image in README.md (#861) Co-authored-by: Dipika Sikka <[email protected]> 067c27c
- update accelerate version (#899) Signed-off-by: Kyle Sayers <[email protected]> 8bc2293
- [GPTQ] Iterative Parameter Updating (#863) * Implement iterative parameter updating Signed-off-by: Kyle Sayers <k... cd1449d
- Small fixes for release (#901) * fix device map * expose one gpu for finetune; update to use a better moodel and ... d918350
- use smaller portion of dataset (#902) 644a500
- Update example to not fail hessian inversion (#904) * update Signed-off-by: Dipika <[email protected]> * ... a173a0c
- bump version (#907) Signed-off-by: Dipika <[email protected]> 93832a6
- add default mappings (#906) Signed-off-by: Kyle Sayers <[email protected]> 86b4d56
- [SparseAutoModelForCausalLM Deprecation] Feature change (#881) * src and tests updates * save model if output_dir... 3d60221
- and 11 more ...
YangWang92 closed an issue on microsoft/VPTQ
ops.gemm
If I have two residual codebooks and indice, how do I do quantization of the network layer with two residual codebooks?How is the ops.gemm method compatible with the input and calculation of multip...
YangWang92 closed an issue on microsoft/VPTQ
Detailed code of the implementation of ops.gemm && ops.dequant
In vqliner.py,where can I see the detailed code of the implementationof ops.dequant(line 402) and ops.gemm(line:322) functions?Thanks and look forward to your answer.