AI/text-generation-webui
1
0
Fork 0
mirror of https://github.com/oobabooga/text-generation-webui.git synced 2024-10-01 01:26:03 -04:00
Code Issues Packages Projects Releases Wiki Activity

Fix for MacOS users encountering model load errors (#6227)

Browse Source 
---------

Co-authored-by: oobabooga <112222186+oobabooga@users.noreply.github.com>
Co-authored-by: Invectorgator <Kudzu12gaming@outlook.com>
This commit is contained in:
InvectorGator 2024-07-12 23:04:19 -04:00 committed by GitHub
parent 05676caf70
commit 4148a9201f
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 29 additions and 47 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

76
modules/llama_cpp_python_hijack.py
View File

@ -1,4 +1,5 @@
import importlib import importlib
import platform
from typing import Sequence from typing import Sequence
from tqdm import tqdm from tqdm import tqdm
@ -13,58 +14,39 @@ imported_module = None
def llama_cpp_lib(): def llama_cpp_lib():
global imported_module global imported_module
def module_to_purpose(module_name): # Determine the platform
if module_name == 'llama_cpp': is_macos = platform.system() == 'Darwin'
return 'CPU'
elif module_name == 'llama_cpp_cuda_tensorcores':
return 'tensorcores'
elif module_name == 'llama_cpp_cuda':
return 'default'
return 'unknown' # Define the library names based on the platform
if is_macos:
lib_names = [
(None, 'llama_cpp')
]
else:
lib_names = [
('cpu', 'llama_cpp'),
('tensorcores', 'llama_cpp_cuda_tensorcores'),
(None, 'llama_cpp_cuda'),
(None, 'llama_cpp')
]
return_lib = None for arg, lib_name in lib_names:
should_import = (arg is None or getattr(shared.args, arg))
if shared.args.cpu: if should_import:
if imported_module and imported_module != 'llama_cpp': if imported_module and imported_module != lib_name:
raise Exception(f"The {module_to_purpose(imported_module)} version of llama-cpp-python is already loaded. Switching to the CPU version currently requires a server restart.") # Conflict detected, raise an exception
try: raise Exception(f"Cannot import `{lib_name}` because `{imported_module}` is already imported. Switching to a different version of llama-cpp-python currently requires a server restart.")
return_lib = importlib.import_module('llama_cpp')
imported_module = 'llama_cpp'
except:
pass
if shared.args.tensorcores and return_lib is None: try:
if imported_module and imported_module != 'llama_cpp_cuda_tensorcores': return_lib = importlib.import_module(lib_name)
raise Exception(f"The {module_to_purpose(imported_module)} version of llama-cpp-python is already loaded. Switching to the tensorcores version currently requires a server restart.") imported_module = lib_name
try: monkey_patch_llama_cpp_python(return_lib)
return_lib = importlib.import_module('llama_cpp_cuda_tensorcores') return return_lib
imported_module = 'llama_cpp_cuda_tensorcores' except ImportError:
except: continue
pass
if return_lib is None: return None
if imported_module and imported_module != 'llama_cpp_cuda':
raise Exception(f"The {module_to_purpose(imported_module)} version of llama-cpp-python is already loaded. Switching to the default version currently requires a server restart.")
try:
return_lib = importlib.import_module('llama_cpp_cuda')
imported_module = 'llama_cpp_cuda'
except:
pass
if return_lib is None and not shared.args.cpu:
if imported_module and imported_module != 'llama_cpp':
raise Exception(f"The {module_to_purpose(imported_module)} version of llama-cpp-python is already loaded. Switching to the CPU version currently requires a server restart.")
try:
return_lib = importlib.import_module('llama_cpp')
imported_module = 'llama_cpp'
except:
pass
if return_lib is not None:
monkey_patch_llama_cpp_python(return_lib)
return return_lib
def eval_with_progress(self, tokens: Sequence[int]): def eval_with_progress(self, tokens: Sequence[int]):