Framework Updates and Standalone Extraction of AutoGPTQ

What's Changed

• Local autogptq fix by @achew010 in #53

Release

• framework v0.1.2: supports default argument with _check_config_and_maybe_check_values
• accelerated-peft v0.1.0.1 with extracted AutoGPTQ based on ModelCloud's rewrite. NOTE: due to some issues with the v0.1.0 release it was removed and we go directly to v0.1.0.1

Full Changelog: v0.1.1.2...v0.1.2.0

v0.1.1.1

What's Changed

• Remove the Float16 Restriction on BNB QLoRA by @fabianlim in #47
• Publish Accelerated Peft and Fused-Ops Plugins by @achew010 in #51

Released

• accelerated-peft 0.1.0: release failed here, will be redone.
• fused-ops-and-kernels 0.1.0:

Full Changelog: v0.1.1...v0.1.1.1

Release: Framework v0.1.1

This releases

• fms-acceleration v0.1.1

Update: v0.1.0 was deleted because of some README problems

What's Changed

• Add BenchMarking Script by @fabianlim in #1
• More README updates. Bench result updates and script Improvements. CLI docstring improvements. by @fabianlim in #4
• Add Configs and Arguments Listing by @fabianlim in #5
• Readme Improvements by @fabianlim in #11
• Add GitHub Workflow for Linting , Formatting and Test. Activate Workflow for Framework by @fabianlim in #7
• Improvements to Benchmark Scripts and Config Generation Workflow by @fabianlim in #13
• Added support for running official HF baseline FSDP-QLoRA benchmark by @achew010 in #16
• Fix FSDP when performing GPTQ-LoRA with Triton V2 by @fabianlim in #15
• Provide Memory Benchmarking Feature to Benchmarking Code by @achew010 in #14
• Upstream Main: Linting, Benchmarking, HF QLoRA baseline, FSDP fixes for GPTQ-LoRA by @fabianlim in #22
• Linting and Formatting for FMS-Acceleration-Peft package by @achew010 in #24
• Workaround Low-Mem-Mode Patch for GPTQ-LoRA by @achew010 in #26
• Group memory field names with prefix and minor fixes by @achew010 in #27
• Initial Addition of FusedOps and Kernels Plugin With Model Patcher by @fabianlim in #25
• Fix FSDP casting issue for Autogptq and Fused Ops by @fabianlim in #28
• Add MLP & QLoRA Fused Ops and Kernels, Mixtral by @fabianlim in #29
• Address Incorrect Ignoring of Base Layer Modules for FSDP with Kernels by @fabianlim in #31
• Shift GPU Memory Computation to End of Benchmarking Script by @achew010 in #30
• Upstream Main: Fused Ops and Kernels, FSDP and Memory Fixes by @fabianlim in #35
• Added trl installation constraint to tox command by @achew010 in #36
• Allow Benchmarks to be run with configuration pointed to in YAML env variable by @fabianlim in #40
• Bound Package Versions by @achew010 in #42
• Upstream Main: Add Lora Dropout, YAML EnvVar Configuration by @fabianlim in #41
• Add MultiRepo Build Packaging Flow and Release Upper Bounds by @fabianlim in #45

Full Changelog: https://github.com/foundation-model-stack/fms-acceleration/commits/v0.1.1