From 506d05aede671f05280ce486b5beca6ae1b8ca8b Mon Sep 17 00:00:00 2001 From: oobabooga <112222186+oobabooga@users.noreply.github.com> Date: Sat, 21 Oct 2023 18:52:59 -0700 Subject: [PATCH] Organize command-line arguments --- README.md | 105 ++++++++++++++++++++--------------------- modules/shared.py | 116 +++++++++++++++++++++------------------------- 2 files changed, 102 insertions(+), 119 deletions(-) diff --git a/README.md b/README.md index 44c43673..a1d80094 100644 --- a/README.md +++ b/README.md @@ -264,8 +264,8 @@ Optionally, you can use the following command-line flags: | Flag | Description | |--------------------------------------------|-------------| -| `-h`, `--help` | Show this help message and exit. | -| `--multi-user` | Multi-user mode. Chat histories are not saved or automatically loaded. WARNING: this is highly experimental. | +| `-h`, `--help` | show this help message and exit | +| `--multi-user` | Multi-user mode. Chat histories are not saved or automatically loaded. WARNING: this is likely not safe for sharing publicly. | | `--character CHARACTER` | The name of the character to load in chat mode by default. | | `--model MODEL` | Name of the model to load by default. | | `--lora LORA [LORA ...]` | The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces. | @@ -275,72 +275,67 @@ Optionally, you can use the following command-line flags: | `--settings SETTINGS_FILE` | Load the default interface settings from this yaml file. See `settings-template.yaml` for an example. If you create a file called `settings.yaml`, this file will be loaded by default without the need to use the `--settings` flag. | | `--extensions EXTENSIONS [EXTENSIONS ...]` | The list of extensions to load. If you want to load more than one extension, write the names separated by spaces. | | `--verbose` | Print the prompts to the terminal. | -| `--chat-buttons` | Show buttons on chat tab instead of hover menu. | +| `--chat-buttons` | Show buttons on the chat tab instead of a hover menu. | #### Model loader | Flag | Description | |--------------------------------------------|-------------| -| `--loader LOADER` | Choose the model loader manually, otherwise, it will get autodetected. Valid options: transformers, autogptq, gptq-for-llama, exllama, exllama_hf, llamacpp, rwkv, ctransformers | +| `--loader LOADER` | Choose the model loader manually, otherwise, it will get autodetected. Valid options: transformers, exllama_hf, exllamav2_hf, exllama, exllamav2, autogptq, gptq-for-llama, llama.cpp, llamacpp_hf, ctransformers, autoawq. | #### Accelerate/transformers | Flag | Description | |---------------------------------------------|-------------| -| `--cpu` | Use the CPU to generate text. Warning: Training on CPU is extremely slow.| +| `--cpu` | Use the CPU to generate text. Warning: Training on CPU is extremely slow. | | `--auto-devices` | Automatically split the model across the available GPU(s) and CPU. | -| `--gpu-memory GPU_MEMORY [GPU_MEMORY ...]` | Maximum GPU memory in GiB to be allocated per GPU. Example: `--gpu-memory 10` for a single GPU, `--gpu-memory 10 5` for two GPUs. You can also set values in MiB like `--gpu-memory 3500MiB`. | -| `--cpu-memory CPU_MEMORY` | Maximum CPU memory in GiB to allocate for offloaded weights. Same as above.| +| `--gpu-memory GPU_MEMORY [GPU_MEMORY ...]` | Maximum GPU memory in GiB to be allocated per GPU. Example: --gpu-memory 10 for a single GPU, --gpu-memory 10 5 for two GPUs. You can also set values in MiB like --gpu-memory 3500MiB. | +| `--cpu-memory CPU_MEMORY` | Maximum CPU memory in GiB to allocate for offloaded weights. Same as above. | | `--disk` | If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk. | -| `--disk-cache-dir DISK_CACHE_DIR` | Directory to save the disk cache to. Defaults to `cache/`. | -| `--load-in-8bit` | Load the model with 8-bit precision (using bitsandbytes).| +| `--disk-cache-dir DISK_CACHE_DIR` | Directory to save the disk cache to. Defaults to "cache". | +| `--load-in-8bit` | Load the model with 8-bit precision (using bitsandbytes). | | `--bf16` | Load the model with bfloat16 precision. Requires NVIDIA Ampere GPU. | -| `--no-cache` | Set `use_cache` to False while generating text. This reduces the VRAM usage a bit with a performance cost. | -| `--xformers` | Use xformer's memory efficient attention. This should increase your tokens/s. | -| `--sdp-attention` | Use torch 2.0's sdp attention. | -| `--trust-remote-code` | Set trust_remote_code=True while loading a model. Necessary for ChatGLM and Falcon. | -| `--use_fast` | Set use_fast=True while loading a tokenizer. | +| `--no-cache` | Set `use_cache` to `False` while generating text. This reduces VRAM usage slightly, but it comes at a performance cost. | +| `--xformers` | Use xformer's memory efficient attention. This is really old and probably doesn't do anything. | +| `--sdp-attention` | Use PyTorch 2.0's SDP attention. Same as above. | +| `--trust-remote-code` | Set `trust_remote_code=True` while loading the model. Necessary for some models. | +| `--use_fast` | Set `use_fast=True` while loading the tokenizer. | #### Accelerate 4-bit -⚠️ Requires minimum compute of 7.0 on Windows at the moment. +⚠️ Requires minimum compute of 7.0 on Windows at the moment. | Flag | Description | |---------------------------------------------|-------------| | `--load-in-4bit` | Load the model with 4-bit precision (using bitsandbytes). | +| `--use_double_quant` | use_double_quant for 4-bit. | | `--compute_dtype COMPUTE_DTYPE` | compute dtype for 4-bit. Valid options: bfloat16, float16, float32. | | `--quant_type QUANT_TYPE` | quant_type for 4-bit. Valid options: nf4, fp4. | -| `--use_double_quant` | use_double_quant for 4-bit. | - -#### GGUF (for llama.cpp and ctransformers) - -| Flag | Description | -|-------------|-------------| -| `--threads` | Number of threads to use. | -| `--threads-batch THREADS_BATCH` | Number of threads to use for batches/prompt processing. | -| `--n_batch` | Maximum number of prompt tokens to batch together when calling llama_eval. | -| `--n-gpu-layers N_GPU_LAYERS` | Number of layers to offload to the GPU. Only works if llama-cpp-python was compiled with BLAS. Set this to 1000000000 to offload all layers to the GPU. | -| `--n_ctx N_CTX` | Size of the prompt context. | #### llama.cpp -| Flag | Description | -|---------------|---------------| -| `--mul_mat_q` | Activate new mulmat kernels. | -| `--tensor_split TENSOR_SPLIT` | Split the model across multiple GPUs, comma-separated list of proportions, e.g. 18,17 | -| `--llama_cpp_seed SEED` | Seed for llama-cpp models. Default 0 (random). | -| `--cache-capacity CACHE_CAPACITY` | Maximum cache capacity. Examples: 2000MiB, 2GiB. When provided without units, bytes will be assumed. | -|`--cfg-cache` | llamacpp_HF: Create an additional cache for CFG negative prompts. | -| `--no-mmap` | Prevent mmap from being used. | -| `--mlock` | Force the system to keep the model in RAM. | -| `--numa` | Activate NUMA task allocation for llama.cpp | -| `--cpu` | Use the CPU version of llama-cpp-python instead of the GPU-accelerated version. | - -#### ctransformers - | Flag | Description | |-------------|-------------| -| `--model_type MODEL_TYPE` | Model type of pre-quantized model. Currently gpt2, gptj, gptneox, falcon, llama, mpt, starcoder (gptbigcode), dollyv2, and replit are supported. | +| `--n_ctx N_CTX` | Size of the prompt context. | +| `--threads` | Number of threads to use. | +| `--threads-batch THREADS_BATCH` | Number of threads to use for batches/prompt processing. | +| `--mul_mat_q` | Activate new mulmat kernels. | +| `--n_batch` | Maximum number of prompt tokens to batch together when calling llama_eval. | +| `--no-mmap` | Prevent mmap from being used. | +| `--mlock` | Force the system to keep the model in RAM. | +| `--n-gpu-layers N_GPU_LAYERS` | Number of layers to offload to the GPU. | +| `--tensor_split TENSOR_SPLIT` | Split the model across multiple GPUs. Comma-separated list of proportions. Example: 18,17. | +| `--llama_cpp_seed SEED` | Seed for llama-cpp models. Default is 0 (random). | +| `--numa` | Activate NUMA task allocation for llama.cpp. | +| `--cache-capacity CACHE_CAPACITY` | Maximum cache capacity (llama-cpp-python). Examples: 2000MiB, 2GiB. When provided without units, bytes will be assumed. | + +#### ExLlama + +| Flag | Description | +|------------------|-------------| +|`--gpu-split` | Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7. | +|`--max_seq_len MAX_SEQ_LEN` | Maximum sequence length. | +|`--cfg-cache` | ExLlama_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader, but not necessary for CFG with base ExLlama. | #### AutoGPTQ @@ -353,14 +348,6 @@ Optionally, you can use the following command-line flags: | `--desc_act` | For models that don't have a quantize_config.json, this parameter is used to define whether to set desc_act or not in BaseQuantizeConfig. | | `--disable_exllama` | Disable ExLlama kernel, which can improve inference speed on some systems. | -#### ExLlama - -| Flag | Description | -|------------------|-------------| -|`--gpu-split` | Comma-separated list of VRAM (in GB) to use per GPU device for model layers, e.g. `20,7,7` | -|`--max_seq_len MAX_SEQ_LEN` | Maximum sequence length. | -|`--cfg-cache` | ExLlama_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader, but not necessary for CFG with base ExLlama. | - #### GPTQ-for-LLaMa | Flag | Description | @@ -370,7 +357,13 @@ Optionally, you can use the following command-line flags: | `--groupsize GROUPSIZE` | Group size. | | `--pre_layer PRE_LAYER [PRE_LAYER ...]` | The number of layers to allocate to the GPU. Setting this parameter enables CPU offloading for 4-bit models. For multi-gpu, write the numbers separated by spaces, eg `--pre_layer 30 60`. | | `--checkpoint CHECKPOINT` | The path to the quantized checkpoint file. If not specified, it will be automatically detected. | -| `--monkey-patch` | Apply the monkey patch for using LoRAs with quantized models. +| `--monkey-patch` | Apply the monkey patch for using LoRAs with quantized models. | + +#### ctransformers + +| Flag | Description | +|-------------|-------------| +| `--model_type MODEL_TYPE` | Model type of pre-quantized model. Currently gpt2, gptj, gptneox, falcon, llama, mpt, starcoder (gptbigcode), dollyv2, and replit are supported. | #### DeepSpeed @@ -391,21 +384,21 @@ Optionally, you can use the following command-line flags: | Flag | Description | |------------------|-------------| -| `--alpha_value ALPHA_VALUE` | Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both. | -| `--rope_freq_base ROPE_FREQ_BASE` | If greater than 0, will be used instead of alpha_value. Those two are related by rope_freq_base = 10000 * alpha_value ^ (64 / 63). | -| `--compress_pos_emb COMPRESS_POS_EMB` | Positional embeddings compression factor. Should be set to (context length) / (model's original context length). Equal to 1/rope_freq_scale. | +| `--alpha_value ALPHA_VALUE` | Positional embeddings alpha factor for NTK RoPE scaling. Use either this or `compress_pos_emb`, not both. | +| `--rope_freq_base ROPE_FREQ_BASE` | If greater than 0, will be used instead of alpha_value. Those two are related by `rope_freq_base = 10000 * alpha_value ^ (64 / 63)`. | +| `--compress_pos_emb COMPRESS_POS_EMB` | Positional embeddings compression factor. Should be set to `(context length) / (model's original context length)`. Equal to `1/rope_freq_scale`. | #### Gradio | Flag | Description | |---------------------------------------|-------------| | `--listen` | Make the web UI reachable from your local network. | -| `--listen-host LISTEN_HOST` | The hostname that the server will use. | | `--listen-port LISTEN_PORT` | The listening port that the server will use. | +| `--listen-host LISTEN_HOST` | The hostname that the server will use. | | `--share` | Create a public URL. This is useful for running the web UI on Google Colab or similar. | | `--auto-launch` | Open the web UI in the default browser upon launch. | -| `--gradio-auth USER:PWD` | set gradio authentication like "username:password"; or comma-delimit multiple like "u1:p1,u2:p2,u3:p3" | -| `--gradio-auth-path GRADIO_AUTH_PATH` | Set the gradio authentication file path. The file should contain one or more user:password pairs in this format: "u1:p1,u2:p2,u3:p3" | +| `--gradio-auth USER:PWD` | Set Gradio authentication password in the format "username:password". Multiple credentials can also be supplied with "u1:p1,u2:p2,u3:p3". | +| `--gradio-auth-path GRADIO_AUTH_PATH` | Set the Gradio authentication file path. The file should contain one or more user:password pairs in the same format as above. | | `--ssl-keyfile SSL_KEYFILE` | The path to the SSL certificate key file. | | `--ssl-certfile SSL_CERTFILE` | The path to the SSL certificate cert file. | diff --git a/modules/shared.py b/modules/shared.py index 427d9230..da5e364a 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -10,7 +10,7 @@ from modules.logging_colors import logger # Model variables model = None tokenizer = None -model_name = "None" +model_name = 'None' is_seq2seq = False model_dirty_from_training = False lora_names = [] @@ -59,94 +59,79 @@ settings = { 'default_extensions': ['gallery'], } - -def str2bool(v): - if isinstance(v, bool): - return v - if v.lower() in ('yes', 'true', 't', 'y', '1'): - return True - elif v.lower() in ('no', 'false', 'f', 'n', '0'): - return False - else: - raise argparse.ArgumentTypeError('Boolean value expected.') - - parser = argparse.ArgumentParser(formatter_class=lambda prog: argparse.HelpFormatter(prog, max_help_position=54)) # Basic settings -parser.add_argument('--notebook', action='store_true', help='DEPRECATED') -parser.add_argument('--chat', action='store_true', help='DEPRECATED') -parser.add_argument('--multi-user', action='store_true', help='Multi-user mode. Chat histories are not saved or automatically loaded. WARNING: this is highly experimental.') +parser.add_argument('--multi-user', action='store_true', help='Multi-user mode. Chat histories are not saved or automatically loaded. Warning: this is likely not safe for sharing publicly.') parser.add_argument('--character', type=str, help='The name of the character to load in chat mode by default.') parser.add_argument('--model', type=str, help='Name of the model to load by default.') -parser.add_argument('--lora', type=str, nargs="+", help='The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.') -parser.add_argument("--model-dir", type=str, default='models/', help="Path to directory with all the models") -parser.add_argument("--lora-dir", type=str, default='loras/', help="Path to directory with all the loras") +parser.add_argument('--lora', type=str, nargs='+', help='The list of LoRAs to load. If you want to load more than one LoRA, write the names separated by spaces.') +parser.add_argument('--model-dir', type=str, default='models/', help='Path to directory with all the models.') +parser.add_argument('--lora-dir', type=str, default='loras/', help='Path to directory with all the loras.') parser.add_argument('--model-menu', action='store_true', help='Show a model menu in the terminal when the web UI is first launched.') -parser.add_argument('--no-stream', action='store_true', help='DEPRECATED') parser.add_argument('--settings', type=str, help='Load the default interface settings from this yaml file. See settings-template.yaml for an example. If you create a file called settings.yaml, this file will be loaded by default without the need to use the --settings flag.') -parser.add_argument('--extensions', type=str, nargs="+", help='The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.') +parser.add_argument('--extensions', type=str, nargs='+', help='The list of extensions to load. If you want to load more than one extension, write the names separated by spaces.') parser.add_argument('--verbose', action='store_true', help='Print the prompts to the terminal.') -parser.add_argument('--chat-buttons', action='store_true', help='Show buttons on chat tab instead of hover menu.') +parser.add_argument('--chat-buttons', action='store_true', help='Show buttons on the chat tab instead of a hover menu.') # Model loader -parser.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: transformers, autogptq, gptq-for-llama, exllama, exllama_hf, llamacpp, rwkv') +parser.add_argument('--loader', type=str, help='Choose the model loader manually, otherwise, it will get autodetected. Valid options: transformers, exllama_hf, exllamav2_hf, exllama, exllamav2, autogptq, gptq-for-llama, llama.cpp, llamacpp_hf, ctransformers, autoawq.') # Accelerate/transformers parser.add_argument('--cpu', action='store_true', help='Use the CPU to generate text. Warning: Training on CPU is extremely slow.') parser.add_argument('--auto-devices', action='store_true', help='Automatically split the model across the available GPU(s) and CPU.') -parser.add_argument('--gpu-memory', type=str, nargs="+", help='Maximum GPU memory in GiB to be allocated per GPU. Example: --gpu-memory 10 for a single GPU, --gpu-memory 10 5 for two GPUs. You can also set values in MiB like --gpu-memory 3500MiB.') +parser.add_argument('--gpu-memory', type=str, nargs='+', help='Maximum GPU memory in GiB to be allocated per GPU. Example: --gpu-memory 10 for a single GPU, --gpu-memory 10 5 for two GPUs. You can also set values in MiB like --gpu-memory 3500MiB.') parser.add_argument('--cpu-memory', type=str, help='Maximum CPU memory in GiB to allocate for offloaded weights. Same as above.') parser.add_argument('--disk', action='store_true', help='If the model is too large for your GPU(s) and CPU combined, send the remaining layers to the disk.') -parser.add_argument('--disk-cache-dir', type=str, default="cache", help='Directory to save the disk cache to. Defaults to "cache".') +parser.add_argument('--disk-cache-dir', type=str, default='cache', help='Directory to save the disk cache to. Defaults to "cache".') parser.add_argument('--load-in-8bit', action='store_true', help='Load the model with 8-bit precision (using bitsandbytes).') parser.add_argument('--bf16', action='store_true', help='Load the model with bfloat16 precision. Requires NVIDIA Ampere GPU.') -parser.add_argument('--no-cache', action='store_true', help='Set use_cache to False while generating text. This reduces the VRAM usage a bit at a performance cost.') -parser.add_argument('--xformers', action='store_true', help="Use xformer's memory efficient attention. This should increase your tokens/s.") -parser.add_argument('--sdp-attention', action='store_true', help="Use torch 2.0's sdp attention.") -parser.add_argument('--trust-remote-code', action='store_true', help="Set trust_remote_code=True while loading a model. Necessary for ChatGLM and Falcon.") -parser.add_argument('--use_fast', action='store_true', help="Set use_fast=True while loading a tokenizer.") +parser.add_argument('--no-cache', action='store_true', help='Set use_cache to False while generating text. This reduces VRAM usage slightly, but it comes at a performance cost.') +parser.add_argument('--xformers', action='store_true', help='Use xformer\'s memory efficient attention. This is really old and probably doesn\'t do anything.') +parser.add_argument('--sdp-attention', action='store_true', help='Use PyTorch 2.0\'s SDP attention. Same as above.') +parser.add_argument('--trust-remote-code', action='store_true', help='Set trust_remote_code=True while loading the model. Necessary for some models.') +parser.add_argument('--use_fast', action='store_true', help='Set use_fast=True while loading the tokenizer.') # Accelerate 4-bit parser.add_argument('--load-in-4bit', action='store_true', help='Load the model with 4-bit precision (using bitsandbytes).') -parser.add_argument('--compute_dtype', type=str, default="float16", help="compute dtype for 4-bit. Valid options: bfloat16, float16, float32.") -parser.add_argument('--quant_type', type=str, default="nf4", help='quant_type for 4-bit. Valid options: nf4, fp4.') parser.add_argument('--use_double_quant', action='store_true', help='use_double_quant for 4-bit.') +parser.add_argument('--compute_dtype', type=str, default='float16', help='compute dtype for 4-bit. Valid options: bfloat16, float16, float32.') +parser.add_argument('--quant_type', type=str, default='nf4', help='quant_type for 4-bit. Valid options: nf4, fp4.') # llama.cpp +parser.add_argument('--n_ctx', type=int, default=2048, help='Size of the prompt context.') parser.add_argument('--threads', type=int, default=0, help='Number of threads to use.') parser.add_argument('--threads-batch', type=int, default=0, help='Number of threads to use for batches/prompt processing.') +parser.add_argument('--mul_mat_q', action='store_true', help='Activate new mulmat kernels.') parser.add_argument('--n_batch', type=int, default=512, help='Maximum number of prompt tokens to batch together when calling llama_eval.') parser.add_argument('--no-mmap', action='store_true', help='Prevent mmap from being used.') parser.add_argument('--mlock', action='store_true', help='Force the system to keep the model in RAM.') -parser.add_argument('--mul_mat_q', action='store_true', help='Activate new mulmat kernels.') -parser.add_argument('--cache-capacity', type=str, help='Maximum cache capacity. Examples: 2000MiB, 2GiB. When provided without units, bytes will be assumed.') parser.add_argument('--n-gpu-layers', type=int, default=0, help='Number of layers to offload to the GPU.') -parser.add_argument('--tensor_split', type=str, default=None, help="Split the model across multiple GPUs, comma-separated list of proportions, e.g. 18,17") -parser.add_argument('--n_ctx', type=int, default=2048, help='Size of the prompt context.') -parser.add_argument('--llama_cpp_seed', type=int, default=0, help='Seed for llama-cpp models. Default 0 (random)') -parser.add_argument('--numa', action='store_true', help='Activate NUMA task allocation for llama.cpp') +parser.add_argument('--tensor_split', type=str, default=None, help='Split the model across multiple GPUs. Comma-separated list of proportions. Example: 18,17.') +parser.add_argument('--llama_cpp_seed', type=int, default=0, help='Seed for llama-cpp models. Default is 0 (random).') +parser.add_argument('--numa', action='store_true', help='Activate NUMA task allocation for llama.cpp.') +parser.add_argument('--cache-capacity', type=str, help='Maximum cache capacity (llama-cpp-python). Examples: 2000MiB, 2GiB. When provided without units, bytes will be assumed.') -# GPTQ -parser.add_argument('--wbits', type=int, default=0, help='Load a pre-quantized model with specified precision in bits. 2, 3, 4 and 8 are supported.') -parser.add_argument('--model_type', type=str, help='Model type of pre-quantized model. Currently LLaMA, OPT, and GPT-J are supported.') -parser.add_argument('--groupsize', type=int, default=-1, help='Group size.') -parser.add_argument('--pre_layer', type=int, nargs="+", help='The number of layers to allocate to the GPU. Setting this parameter enables CPU offloading for 4-bit models. For multi-gpu, write the numbers separated by spaces, eg --pre_layer 30 60.') -parser.add_argument('--checkpoint', type=str, help='The path to the quantized checkpoint file. If not specified, it will be automatically detected.') -parser.add_argument('--monkey-patch', action='store_true', help='Apply the monkey patch for using LoRAs with quantized models.') +# ExLlama +parser.add_argument('--gpu-split', type=str, help='Comma-separated list of VRAM (in GB) to use per GPU device for model layers. Example: 20,7,7.') +parser.add_argument('--max_seq_len', type=int, default=2048, help='Maximum sequence length.') +parser.add_argument('--cfg-cache', action='store_true', help='ExLlama_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader, but not necessary for CFG with base ExLlama.') # AutoGPTQ parser.add_argument('--triton', action='store_true', help='Use triton.') -parser.add_argument('--no_inject_fused_attention', action='store_true', help='Do not use fused attention (lowers VRAM requirements).') -parser.add_argument('--no_inject_fused_mlp', action='store_true', help='Triton mode only: Do not use fused MLP (lowers VRAM requirements).') +parser.add_argument('--no_inject_fused_attention', action='store_true', help='Disable the use of fused attention, which will use less VRAM at the cost of slower inference.') +parser.add_argument('--no_inject_fused_mlp', action='store_true', help='Triton mode only: disable the use of fused MLP, which will use less VRAM at the cost of slower inference.') parser.add_argument('--no_use_cuda_fp16', action='store_true', help='This can make models faster on some systems.') -parser.add_argument('--desc_act', action='store_true', help='For models that don\'t have a quantize_config.json, this parameter is used to define whether to set desc_act or not in BaseQuantizeConfig.') +parser.add_argument('--desc_act', action='store_true', help='For models that do not have a quantize_config.json, this parameter is used to define whether to set desc_act or not in BaseQuantizeConfig.') parser.add_argument('--disable_exllama', action='store_true', help='Disable ExLlama kernel, which can improve inference speed on some systems.') -# ExLlama -parser.add_argument('--gpu-split', type=str, help="Comma-separated list of VRAM (in GB) to use per GPU device for model layers, e.g. 20,7,7") -parser.add_argument('--max_seq_len', type=int, default=2048, help="Maximum sequence length.") -parser.add_argument('--cfg-cache', action='store_true', help="ExLlama_HF: Create an additional cache for CFG negative prompts. Necessary to use CFG with that loader, but not necessary for CFG with base ExLlama.") +# GPTQ-for-LLaMa +parser.add_argument('--wbits', type=int, default=0, help='Load a pre-quantized model with specified precision in bits. 2, 3, 4 and 8 are supported.') +parser.add_argument('--model_type', type=str, help='Model type of pre-quantized model. Currently LLaMA, OPT, and GPT-J are supported.') +parser.add_argument('--groupsize', type=int, default=-1, help='Group size.') +parser.add_argument('--pre_layer', type=int, nargs='+', help='The number of layers to allocate to the GPU. Setting this parameter enables CPU offloading for 4-bit models. For multi-gpu, write the numbers separated by spaces, eg --pre_layer 30 60.') +parser.add_argument('--checkpoint', type=str, help='The path to the quantized checkpoint file. If not specified, it will be automatically detected.') +parser.add_argument('--monkey-patch', action='store_true', help='Apply the monkey patch for using LoRAs with quantized models.') # DeepSpeed parser.add_argument('--deepspeed', action='store_true', help='Enable the use of DeepSpeed ZeRO-3 for inference via the Transformers integration.') @@ -158,31 +143,36 @@ parser.add_argument('--rwkv-strategy', type=str, default=None, help='RWKV: The s parser.add_argument('--rwkv-cuda-on', action='store_true', help='RWKV: Compile the CUDA kernel for better performance.') # RoPE -parser.add_argument('--alpha_value', type=float, default=1, help="Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both.") -parser.add_argument('--rope_freq_base', type=int, default=0, help="If greater than 0, will be used instead of alpha_value. Those two are related by rope_freq_base = 10000 * alpha_value ^ (64 / 63).") +parser.add_argument('--alpha_value', type=float, default=1, help='Positional embeddings alpha factor for NTK RoPE scaling. Use either this or compress_pos_emb, not both.') +parser.add_argument('--rope_freq_base', type=int, default=0, help='If greater than 0, will be used instead of alpha_value. Those two are related by rope_freq_base = 10000 * alpha_value ^ (64 / 63).') parser.add_argument('--compress_pos_emb', type=int, default=1, help="Positional embeddings compression factor. Should be set to (context length) / (model\'s original context length). Equal to 1/rope_freq_scale.") # Gradio parser.add_argument('--listen', action='store_true', help='Make the web UI reachable from your local network.') -parser.add_argument('--listen-host', type=str, help='The hostname that the server will use.') parser.add_argument('--listen-port', type=int, help='The listening port that the server will use.') +parser.add_argument('--listen-host', type=str, help='The hostname that the server will use.') parser.add_argument('--share', action='store_true', help='Create a public URL. This is useful for running the web UI on Google Colab or similar.') parser.add_argument('--auto-launch', action='store_true', default=False, help='Open the web UI in the default browser upon launch.') -parser.add_argument("--gradio-auth", type=str, help='set gradio authentication like "username:password"; or comma-delimit multiple like "u1:p1,u2:p2,u3:p3"', default=None) -parser.add_argument("--gradio-auth-path", type=str, help='Set the gradio authentication file path. The file should contain one or more user:password pairs in this format: "u1:p1,u2:p2,u3:p3"', default=None) -parser.add_argument("--ssl-keyfile", type=str, help='The path to the SSL certificate key file.', default=None) -parser.add_argument("--ssl-certfile", type=str, help='The path to the SSL certificate cert file.', default=None) +parser.add_argument('--gradio-auth', type=str, help='Set Gradio authentication password in the format "username:password". Multiple credentials can also be supplied with "u1:p1,u2:p2,u3:p3".', default=None) +parser.add_argument('--gradio-auth-path', type=str, help='Set the Gradio authentication file path. The file should contain one or more user:password pairs in the same format as above.', default=None) +parser.add_argument('--ssl-keyfile', type=str, help='The path to the SSL certificate key file.', default=None) +parser.add_argument('--ssl-certfile', type=str, help='The path to the SSL certificate cert file.', default=None) # API parser.add_argument('--api', action='store_true', help='Enable the API extension.') -parser.add_argument('--api-blocking-port', type=int, default=5000, help='The listening port for the blocking API.') -parser.add_argument('--api-streaming-port', type=int, default=5005, help='The listening port for the streaming API.') parser.add_argument('--public-api', action='store_true', help='Create a public URL for the API using Cloudfare.') parser.add_argument('--public-api-id', type=str, help='Tunnel ID for named Cloudflare Tunnel. Use together with public-api option.', default=None) +parser.add_argument('--api-blocking-port', type=int, default=5000, help='The listening port for the blocking API.') +parser.add_argument('--api-streaming-port', type=int, default=5005, help='The listening port for the streaming API.') # Multimodal parser.add_argument('--multimodal-pipeline', type=str, default=None, help='The multimodal pipeline to use. Examples: llava-7b, llava-13b.') +# Deprecated parameters +parser.add_argument('--notebook', action='store_true', help='DEPRECATED') +parser.add_argument('--chat', action='store_true', help='DEPRECATED') +parser.add_argument('--no-stream', action='store_true', help='DEPRECATED') + args = parser.parse_args() args_defaults = parser.parse_args([]) provided_arguments = [] @@ -198,13 +188,13 @@ for k in ['chat', 'notebook', 'no_stream']: # Security warnings if args.trust_remote_code: - logger.warning("trust_remote_code is enabled. This is dangerous.") + logger.warning('trust_remote_code is enabled. This is dangerous.') if args.share: logger.warning("The gradio \"share link\" feature uses a proprietary executable to create a reverse tunnel. Use it with care.") if any((args.listen, args.share)) and not any((args.gradio_auth, args.gradio_auth_path)): logger.warning("\nYou are potentially exposing the web UI to the entire internet without any access password.\nYou can create one with the \"--gradio-auth\" flag like this:\n\n--gradio-auth username:password\n\nMake sure to replace username:password with your own.") if args.multi_user: - logger.warning("\nThe multi-user mode is highly experimental and should not be shared publicly.") + logger.warning('\nThe multi-user mode is highly experimental and should not be shared publicly.') def fix_loader_name(name):