Merge dev branch

README.md

@@ -48,7 +48,7 @@ If you ever need to install something manually in the `installer_files` environm
 * There is no need to run any of those scripts (`start_`, `update_wizard_`, or `cmd_`) as admin/root.
 * To install the requirements for extensions, you can use the `extensions_reqs` script for your OS. At the end, this script will install the main requirements for the project to make sure that they take precedence in case of version conflicts.
 * For additional instructions about AMD and WSL setup, consult [the documentation](https://github.com/oobabooga/text-generation-webui/wiki).
-* For automated installation, you can use the `GPU_CHOICE`, `USE_CUDA118`, and `LAUNCH_AFTER_INSTALL` environment variables. For instance: `GPU_CHOICE=A USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE ./start_linux.sh`.
+* For automated installation, you can use the `GPU_CHOICE`, `USE_CUDA118`, `LAUNCH_AFTER_INSTALL`, and `INSTALL_EXTENSIONS` environment variables. For instance: `GPU_CHOICE=A USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh`.
 
 ### Manual installation using Conda
 
@@ -279,6 +279,7 @@ List of command-line flags
 |`--cfg-cache`                         | ExLlamav2_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader. |
 |`--no_flash_attn`                     | Force flash-attention to not be used. |
 |`--cache_8bit`                        | Use 8-bit cache to save VRAM. |
+|`--cache_4bit`                        | Use Q4 cache to save VRAM. |
 |`--num_experts_per_token NUM_EXPERTS_PER_TOKEN` |  Number of experts to use for generation. Applies to MoE models like Mixtral. |
 
 #### AutoGPTQ

docker/amd/Dockerfile

@@ -13,7 +13,7 @@ RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
 WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
-RUN GPU_CHOICE=B USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE ./start_linux.sh --verbose
+RUN GPU_CHOICE=B USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
 COPY CMD_FLAGS.txt /home/app/text-generation-webui/
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
 WORKDIR /home/app/text-generation-webui

docker/cpu/Dockerfile

@@ -17,7 +17,7 @@ RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
 WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
-RUN GPU_CHOICE=N USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE ./start_linux.sh --verbose
+RUN GPU_CHOICE=N USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
 COPY CMD_FLAGS.txt /home/app/text-generation-webui/
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
 # set umask to ensure group read / write at runtime

docker/intel/Dockerfile

@@ -13,7 +13,7 @@ RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
 WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
-RUN GPU_CHOICE=D USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE ./start_linux.sh --verbose
+RUN GPU_CHOICE=D USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
 COPY CMD_FLAGS.txt /home/app/text-generation-webui/
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
 # set umask to ensure group read / write at runtime

docker/nvidia/Dockerfile

@@ -13,7 +13,7 @@ RUN --mount=type=cache,target=/var/cache/apt,sharing=locked,rw \
 WORKDIR /home/app/
 RUN git clone https://github.com/oobabooga/text-generation-webui.git 
 WORKDIR /home/app/text-generation-webui
-RUN GPU_CHOICE=A USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE ./start_linux.sh --verbose
+RUN GPU_CHOICE=A USE_CUDA118=FALSE LAUNCH_AFTER_INSTALL=FALSE INSTALL_EXTENSIONS=TRUE ./start_linux.sh --verbose
 COPY CMD_FLAGS.txt /home/app/text-generation-webui/
 EXPOSE ${CONTAINER_PORT:-7860} ${CONTAINER_API_PORT:-5000} ${CONTAINER_API_STREAM_PORT:-5005}
 WORKDIR /home/app/text-generation-webui

docs/04 - Model Tab.md

@@ -46,6 +46,7 @@ Examples:
 * **cfg-cache**: Creates a second cache to hold the CFG negative prompts. You need to set this if and only if you intend to use CFG in the "Parameters" > "Generation" tab. Checking this parameter doubles the cache VRAM usage.
 * **no_flash_attn**: Disables flash attention. Otherwise, it is automatically used as long as the library is installed.
 * **cache_8bit**: Create a 8-bit precision cache instead of a 16-bit one. This saves VRAM but increases perplexity (I don't know by how much).
+* **cache_4bit**: Creates a Q4 cache using grouped quantization.
 
 ### ExLlamav2
 

modules/exllamav2.py

@@ -6,6 +6,7 @@
     ExLlamaV2,
     ExLlamaV2Cache,
     ExLlamaV2Cache_8bit,
+    ExLlamaV2Cache_Q4,
     ExLlamaV2Config,
     ExLlamaV2Tokenizer
 )
@@ -60,6 +61,8 @@ def from_pretrained(self, path_to_model):
 
         if shared.args.cache_8bit:
             cache = ExLlamaV2Cache_8bit(model, lazy=shared.args.autosplit)
+        elif shared.args.cache_4bit:
+            cache = ExLlamaV2Cache_Q4(model, lazy=shared.args.autosplit)
         else:
             cache = ExLlamaV2Cache(model, lazy=shared.args.autosplit)
 

modules/exllamav2_hf.py

@@ -8,6 +8,7 @@
     ExLlamaV2,
     ExLlamaV2Cache,
     ExLlamaV2Cache_8bit,
+    ExLlamaV2Cache_Q4,
     ExLlamaV2Config
 )
 from torch.nn import CrossEntropyLoss
@@ -50,6 +51,8 @@ def __init__(self, config: ExLlamaV2Config):
 
         if shared.args.cache_8bit:
             self.ex_cache = ExLlamaV2Cache_8bit(self.ex_model, lazy=shared.args.autosplit)
+        elif shared.args.cache_4bit:
+            self.ex_cache = ExLlamaV2Cache_Q4(self.ex_model, lazy=shared.args.autosplit)
         else:
             self.ex_cache = ExLlamaV2Cache(self.ex_model, lazy=shared.args.autosplit)
 
@@ -60,6 +63,8 @@ def __init__(self, config: ExLlamaV2Config):
         if shared.args.cfg_cache:
             if shared.args.cache_8bit:
                 self.ex_cache_negative = ExLlamaV2Cache_8bit(self.ex_model)
+            elif shared.args.cache_4bit:
+                self.ex_cache_negative = ExLlamaV2Cache_Q4(self.ex_model)
             else:
                 self.ex_cache_negative = ExLlamaV2Cache(self.ex_model)
 

modules/loaders.py

@@ -78,6 +78,7 @@
         'no_flash_attn',
         'num_experts_per_token',
         'cache_8bit',
+        'cache_4bit',
         'autosplit',
         'alpha_value',
         'compress_pos_emb',
@@ -90,6 +91,7 @@
         'no_flash_attn',
         'num_experts_per_token',
         'cache_8bit',
+        'cache_4bit',
         'autosplit',
         'alpha_value',
         'compress_pos_emb',

modules/shared.py

@@ -139,6 +139,7 @@
 group.add_argument('--cfg-cache', action='store_true', help='ExLlamav2_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader.')
 group.add_argument('--no_flash_attn', action='store_true', help='Force flash-attention to not be used.')
 group.add_argument('--cache_8bit', action='store_true', help='Use 8-bit cache to save VRAM.')
+group.add_argument('--cache_4bit', action='store_true', help='Use Q4 cache to save VRAM.')
 group.add_argument('--num_experts_per_token', type=int, default=2, help='Number of experts to use for generation. Applies to MoE models like Mixtral.')
 
 # AutoGPTQ

modules/ui.py

@@ -76,6 +76,7 @@ def list_model_elements():
         'no_flash_attn',
         'num_experts_per_token',
         'cache_8bit',
+        'cache_4bit',
         'autosplit',
         'threads',
         'threads_batch',

modules/ui_model_menu.py

@@ -132,6 +132,7 @@ def create_ui():
                             shared.gradio['disk'] = gr.Checkbox(label="disk", value=shared.args.disk)
                             shared.gradio['bf16'] = gr.Checkbox(label="bf16", value=shared.args.bf16)
                             shared.gradio['cache_8bit'] = gr.Checkbox(label="cache_8bit", value=shared.args.cache_8bit, info='Use 8-bit cache to save VRAM.')
+                            shared.gradio['cache_4bit'] = gr.Checkbox(label="cache_4bit", value=shared.args.cache_4bit, info='Use Q4 cache to save VRAM.')
                             shared.gradio['autosplit'] = gr.Checkbox(label="autosplit", value=shared.args.autosplit, info='Automatically split the model tensors across the available GPUs.')
                             shared.gradio['no_flash_attn'] = gr.Checkbox(label="no_flash_attn", value=shared.args.no_flash_attn, info='Force flash-attention to not be used.')
                             shared.gradio['cfg_cache'] = gr.Checkbox(label="cfg-cache", value=shared.args.cfg_cache, info='Necessary to use CFG with this loader.')

one_click.py

@@ -342,7 +342,7 @@ def update_requirements(initial_installation=False, pull=True):
 
         files_to_check = [
             'start_linux.sh', 'start_macos.sh', 'start_windows.bat', 'start_wsl.bat',
-            'update_linux.sh', 'update_macos.sh', 'update_windows.bat', 'update_wsl.bat',
+            'update_wizard_linux.sh', 'update_wizard_macos.sh', 'update_wizard_windows.bat', 'update_wizard_wsl.bat',
             'one_click.py'
         ]
 
@@ -356,6 +356,9 @@ def update_requirements(initial_installation=False, pull=True):
                 print_big_message(f"File '{file_name}' was updated during 'git pull'. Please run the script again.")
                 exit(1)
 
+    if os.environ.get("INSTALL_EXTENSIONS", "").lower() in ("yes", "y", "true", "1", "t", "on"):
+        install_extensions_requirements()
+
     # Update PyTorch
     if not initial_installation:
         update_pytorch()
@@ -387,7 +390,7 @@ def update_requirements(initial_installation=False, pull=True):
     if is_cuda118:
         textgen_requirements = [req.replace('+cu121', '+cu118').replace('+cu122', '+cu118') for req in textgen_requirements]
     if is_windows() and is_cuda118:  # No flash-attention on Windows for CUDA 11
-        textgen_requirements = [req for req in textgen_requirements if 'jllllll/flash-attention' not in req]
+        textgen_requirements = [req for req in textgen_requirements if 'oobabooga/flash-attention' not in req]
 
     with open('temp_requirements.txt', 'w') as file:
         file.write('\n'.join(textgen_requirements))

requirements.txt

@@ -56,11 +56,11 @@ https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+cu1
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.14.1/exllamav2-0.0.14.1+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.14.1/exllamav2-0.0.14.1+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.14.1/exllamav2-0.0.14.1+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.14.1/exllamav2-0.0.14.1+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.14.1/exllamav2-0.0.14.1-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.15/exllamav2-0.0.15+cu121-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.15/exllamav2-0.0.15+cu121-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.15/exllamav2-0.0.15+cu121-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.15/exllamav2-0.0.15+cu121-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.15/exllamav2-0.0.15-py3-none-any.whl; platform_system == "Linux" and platform_machine != "x86_64"
 https://github.com/oobabooga/flash-attention/releases/download/v2.5.6/flash_attn-2.5.6+cu122torch2.2.0cxx11abiFALSE-cp311-cp311-win_amd64.whl; platform_system == "Windows" and python_version == "3.11"
 https://github.com/oobabooga/flash-attention/releases/download/v2.5.6/flash_attn-2.5.6+cu122torch2.2.0cxx11abiFALSE-cp310-cp310-win_amd64.whl; platform_system == "Windows" and python_version == "3.10"
 https://github.com/Dao-AILab/flash-attention/releases/download/v2.5.6/flash_attn-2.5.6+cu122torch2.2cxx11abiFALSE-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"

requirements_amd.txt

@@ -40,8 +40,8 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/ro
 https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/rocm/llama_cpp_python_cuda-0.2.55+rocm5.6.1-cp310-cp310-manylinux_2_31_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.14.1/exllamav2-0.0.14.1+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.14.1/exllamav2-0.0.14.1+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.14.1/exllamav2-0.0.14.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.15/exllamav2-0.0.15+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.15/exllamav2-0.0.15+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.15/exllamav2-0.0.15-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
 https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_llama-0.1.1+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_llama-0.1.1+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"

requirements_amd_noavx2.txt

@@ -38,8 +38,8 @@ https://github.com/oobabooga/llama-cpp-python-cuBLAS-wheels/releases/download/cp
 # AMD wheels
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/jllllll/AutoGPTQ/releases/download/v0.6.0/auto_gptq-0.6.0+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.14.1/exllamav2-0.0.14.1+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.14.1/exllamav2-0.0.14.1+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
-https://github.com/oobabooga/exllamav2/releases/download/v0.0.14.1/exllamav2-0.0.14.1-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.15/exllamav2-0.0.15+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.15/exllamav2-0.0.15+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"
+https://github.com/oobabooga/exllamav2/releases/download/v0.0.15/exllamav2-0.0.15-py3-none-any.whl; platform_system != "Darwin" and platform_machine != "x86_64"
 https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_llama-0.1.1+rocm5.6-cp311-cp311-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.11"
 https://github.com/jllllll/GPTQ-for-LLaMa-CUDA/releases/download/0.1.1/gptq_for_llama-0.1.1+rocm5.6-cp310-cp310-linux_x86_64.whl; platform_system == "Linux" and platform_machine == "x86_64" and python_version == "3.10"