mathieu/ai-suite-rocm-local
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases 1 Wiki Activity

improve update for textgenui

Browse Source
This commit is contained in:
Mathieu Broillet 2024-09-14 10:48:42 +02:00
parent 08e3df166a
commit eb60acc2a0
Signed by: mathieu
GPG Key ID: A08E484FE95074C1
1 changed files with 11 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
services/text_generation_webui.py
View File

@ -2,6 +2,7 @@ from core.stack import Stack
class TextGenerationWebui(Stack): class TextGenerationWebui(Stack):
def __init__(self): def __init__(self):
super().__init__( super().__init__(
'Text Generation', 'Text Generation',
@ -10,6 +11,8 @@ class TextGenerationWebui(Stack):
'https://github.com/oobabooga/text-generation-webui/' 'https://github.com/oobabooga/text-generation-webui/'
) )
self.exllama = "0.2.1"
def _install(self): def _install(self):
# Install LlamaCpp from prebuilt # Install LlamaCpp from prebuilt
self.pip_install("llama-cpp-python", env=["CMAKE_ARGS=\"-DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS\""]) # cpu self.pip_install("llama-cpp-python", env=["CMAKE_ARGS=\"-DGGML_BLAS=ON -DGGML_BLAS_VENDOR=OpenBLAS\""]) # cpu
@ -41,13 +44,19 @@ class TextGenerationWebui(Stack):
# Fix llama trying to use cuda version # Fix llama trying to use cuda version
self.remove_line_in_file("llama_cpp_cuda", "webui/modules/llama_cpp_python_hijack.py") self.remove_line_in_file("llama_cpp_cuda", "webui/modules/llama_cpp_python_hijack.py")
# Install useful packages # Install ExLlamaV2 and auto-gptq
self.pip_install( self.pip_install(
"https://github.com/turboderp/exllamav2/releases/download/v0.1.9/exllamav2-0.1.9+rocm6.1.torch2.4.0-cp310-cp310-linux_x86_64.whl") f"https://github.com/turboderp/exllamav2/releases/download/v{self.exllama}/exllamav2-{self.exllama}+rocm6.1.torch2.4.0-cp310-cp310-linux_x86_64.whl")
self.install_from_prebuilt("bitsandbytes") self.install_from_prebuilt("bitsandbytes")
self.pip_install("auto-gptq", args=["--no-build-isolation", "--extra-index-url", self.pip_install("auto-gptq", args=["--no-build-isolation", "--extra-index-url",
"https://huggingface.github.io/autogptq-index/whl/rocm573/"]) "https://huggingface.github.io/autogptq-index/whl/rocm573/"])
def _update(self):
self.pip_install(
f"https://github.com/turboderp/exllamav2/releases/download/v{self.exllama}/exllamav2-{self.exllama}+rocm6.1.torch2.4.0-cp310-cp310-linux_x86_64.whl")
self.pip_install("auto-gptq", args=["--no-build-isolation", "--extra-index-url",
"https://huggingface.github.io/autogptq-index/whl/rocm573/"])
def _start(self): def _start(self):
args = ["--listen", "--listen-port", str(self.port)] args = ["--listen", "--listen-port", str(self.port)]
self.python(f"server.py", args=args, current_dir="webui", self.python(f"server.py", args=args, current_dir="webui",