From 6eeb62b82c03361f0fa4437a1769d8f614634ef4 Mon Sep 17 00:00:00 2001 From: kingbri Date: Fri, 2 Feb 2024 22:24:44 -0500 Subject: [PATCH] Requirements: Update exllamav2, torch, and FA2 Torch to 2.2, exllamav2 to 0.0.13, FA2 to 2.4.2 on Windows and 2.5.2 on Linux. Signed-off-by: kingbri --- backends/exllamav2/utils.py | 2 +- requirements-amd.txt | 7 +++---- requirements-cu118.txt | 14 +++++++------- requirements.txt | 19 +++++++++---------- 4 files changed, 20 insertions(+), 22 deletions(-) diff --git a/backends/exllamav2/utils.py b/backends/exllamav2/utils.py index ac5ce1a..41081c8 100644 --- a/backends/exllamav2/utils.py +++ b/backends/exllamav2/utils.py @@ -9,7 +9,7 @@ logger = init_logger(__name__) def check_exllama_version(): """Verifies the exllama version""" - required_version = "0.0.12" + required_version = "0.0.13" current_version = package_version("exllamav2").split("+")[0] if version.parse(current_version) < version.parse(required_version): diff --git a/requirements-amd.txt b/requirements-amd.txt index 416e897..dcad274 100644 --- a/requirements-amd.txt +++ b/requirements-amd.txt @@ -1,11 +1,10 @@ # Torch -# Patch until torch 2.2 is supported --extra-index-url https://download.pytorch.org/whl/rocm5.6 -torch < 2.2, >= 2 +torch ~= 2.2 # Exllamav2 -https://github.com/turboderp/exllamav2/releases/download/v0.0.12/exllamav2-0.0.12+rocm5.6-cp311-cp311-linux_x86_64.whl; python_version == "3.11" -https://github.com/turboderp/exllamav2/releases/download/v0.0.12/exllamav2-0.0.12+rocm5.6-cp310-cp310-linux_x86_64.whl; python_version == "3.10" +https://github.com/turboderp/exllamav2/releases/download/v0.0.13/exllamav2-0.0.13+rocm5.6-cp311-cp311-linux_x86_64.whl; python_version == "3.11" +https://github.com/turboderp/exllamav2/releases/download/v0.0.13/exllamav2-0.0.13+rocm5.6-cp310-cp310-linux_x86_64.whl; python_version == "3.10" # Pip dependencies fastapi diff --git a/requirements-cu118.txt b/requirements-cu118.txt index ed0042b..837b2bc 100644 --- a/requirements-cu118.txt +++ b/requirements-cu118.txt @@ -1,17 +1,16 @@ # Torch -# Patch until torch 2.2 is supported --extra-index-url https://download.pytorch.org/whl/cu118 -torch < 2.2, >= 2 +torch ~= 2.2 # Exllamav2 # Windows -https://github.com/turboderp/exllamav2/releases/download/v0.0.12/exllamav2-0.0.12+cu118-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/turboderp/exllamav2/releases/download/v0.0.12/exllamav2-0.0.12+cu118-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/turboderp/exllamav2/releases/download/v0.0.13/exllamav2-0.0.13+cu118-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/turboderp/exllamav2/releases/download/v0.0.13/exllamav2-0.0.13+cu118-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # Linux -https://github.com/turboderp/exllamav2/releases/download/v0.0.12/exllamav2-0.0.12+cu118-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/turboderp/exllamav2/releases/download/v0.0.12/exllamav2-0.0.12+cu118-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/turboderp/exllamav2/releases/download/v0.0.13/exllamav2-0.0.13+cu118-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/turboderp/exllamav2/releases/download/v0.0.13/exllamav2-0.0.13+cu118-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # Pip dependencies fastapi @@ -23,4 +22,5 @@ jinja2 >= 3.0.0 colorlog # Linux FA2 from https://github.com/Dao-AILab/flash-attention/releases -https://github.com/Dao-AILab/flash-attention/releases/download/v2.4.1/flash_attn-2.4.1+cu118torch2.1cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.5.2/flash_attn-2.5.2+cu118torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.5.2/flash_attn-2.5.2+cu118torch2.2cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" diff --git a/requirements.txt b/requirements.txt index 4e529e1..04123ba 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,17 +1,16 @@ # Torch -# Patch until torch 2.2 is supported --extra-index-url https://download.pytorch.org/whl/cu121 -torch < 2.2, >= 2 +torch ~= 2.2 # Exllamav2 # Windows -https://github.com/turboderp/exllamav2/releases/download/v0.0.12/exllamav2-0.0.12+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/turboderp/exllamav2/releases/download/v0.0.12/exllamav2-0.0.12+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/turboderp/exllamav2/releases/download/v0.0.13/exllamav2-0.0.13+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/turboderp/exllamav2/releases/download/v0.0.13/exllamav2-0.0.13+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # Linux -https://github.com/turboderp/exllamav2/releases/download/v0.0.12/exllamav2-0.0.12+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/turboderp/exllamav2/releases/download/v0.0.12/exllamav2-0.0.12+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/turboderp/exllamav2/releases/download/v0.0.13/exllamav2-0.0.13+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/turboderp/exllamav2/releases/download/v0.0.13/exllamav2-0.0.13+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" # Pip dependencies fastapi @@ -25,9 +24,9 @@ colorlog # Flash attention v2 # Windows FA2 from https://github.com/bdashore3/flash-attention/releases -https://github.com/bdashore3/flash-attention/releases/download/v2.4.1/flash_attn-2.4.1+cu121torch2.1cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" -https://github.com/bdashore3/flash-attention/releases/download/v2.4.1/flash_attn-2.4.1+cu121torch2.1cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" +https://github.com/bdashore3/flash-attention/releases/download/v2.4.2/flash_attn-2.4.2+cu122torch2.2.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11" +https://github.com/bdashore3/flash-attention/releases/download/v2.4.2/flash_attn-2.4.2+cu122torch2.2.0cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10" # Linux FA2 from https://github.com/Dao-AILab/flash-attention/releases -https://github.com/Dao-AILab/flash-attention/releases/download/v2.4.1/flash_attn-2.4.1+cu122torch2.1cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" -https://github.com/Dao-AILab/flash-attention/releases/download/v2.4.1/flash_attn-2.4.1+cu122torch2.1cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.5.2/flash_attn-2.5.2+cu122torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11" +https://github.com/Dao-AILab/flash-attention/releases/download/v2.5.2/flash_attn-2.5.2+cu122torch2.2cxx11abiFALSE-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"