Mirrors/text-generation-webui
1
0
Fork 0
mirror of https://github.com/oobabooga/text-generation-webui.git synced 2024-11-22 16:17:57 +01:00
Code Issues Packages Projects Releases Wiki Activity

Update README

Browse Source
This commit is contained in:
oobabooga 2023-04-17 10:55:35 -03:00
parent 07de7d0426
commit 89bc540557
2 changed files with 16 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
README.md
View File

@ -232,14 +232,14 @@ Optionally, you can use the following command-line flags:
| Flag | Description | | Flag | Description |
|---------------------------|-------------| |---------------------------|-------------|
| `--wbits WBITS` | GPTQ: Load a pre-quantized model with specified precision in bits. 2, 3, 4 and 8 are supported. | | `--wbits WBITS` | Load a pre-quantized model with specified precision in bits. 2, 3, 4 and 8 are supported. |
| `--model_type MODEL_TYPE` | GPTQ: Model type of pre-quantized model. Currently LLaMA, OPT, and GPT-J are supported. | | `--model_type MODEL_TYPE` | Model type of pre-quantized model. Currently LLaMA, OPT, and GPT-J are supported. |
| `--groupsize GROUPSIZE` | GPTQ: Group size. | | `--groupsize GROUPSIZE` | Group size. |
| `--pre_layer PRE_LAYER` | GPTQ: The number of layers to allocate to the GPU. Setting this parameter enables CPU offloading for 4-bit models. | | `--pre_layer PRE_LAYER` | The number of layers to allocate to the GPU. Setting this parameter enables CPU offloading for 4-bit models. |
| `--no-quant_attn` | GPTQ: Disable quant attention for triton. If you encounter incoherent results try disabling this. | | `--monkey-patch` | Apply the monkey patch for using LoRAs with quantized models.
| `--no-warmup_autotune` | GPTQ: Disable warmup autotune for triton. | | `--no-quant_attn` | (triton) Disable quant attention. If you encounter incoherent results try disabling this.
| `--no-fused_mlp` | GPTQ: Disable fused mlp for triton. If you encounter "Unexpected mma -> mma layout conversion" try disabling this. Disabling may also help model splitting for multi-gpu setups.| | `--no-warmup_autotune` | (triton) Disable warmup autotune.
| `--monkey-patch` | GPTQ: Apply the monkey patch for using LoRAs with quantized models. | | `--no-fused_mlp` | (triton) Disable fused mlp. If you encounter "Unexpected mma -> mma layout conversion" try disabling this.
#### FlexGen #### FlexGen

16
modules/shared.py
View File

@ -119,14 +119,14 @@ parser.add_argument('--trust-remote-code', action='store_true', help="Set trust_
parser.add_argument('--threads', type=int, default=0, help='Number of threads to use in llama.cpp.') parser.add_argument('--threads', type=int, default=0, help='Number of threads to use in llama.cpp.')
# GPTQ # GPTQ
parser.add_argument('--wbits', type=int, default=0, help='GPTQ: Load a pre-quantized model with specified precision in bits. 2, 3, 4 and 8 are supported.') 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='GPTQ: Model type of pre-quantized model. Currently LLaMA, OPT, and GPT-J 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='GPTQ: Group size.') parser.add_argument('--groupsize', type=int, default=-1, help='Group size.')
parser.add_argument('--pre_layer', type=int, default=0, help='GPTQ: The number of layers to allocate to the GPU. Setting this parameter enables CPU offloading for 4-bit models.') parser.add_argument('--pre_layer', type=int, default=0, help='The number of layers to allocate to the GPU. Setting this parameter enables CPU offloading for 4-bit models.')
parser.add_argument('--no-quant_attn', action='store_true', help='GPTQ: Disable quant attention for triton. If you encounter incoherent results try disabling this.') parser.add_argument('--monkey-patch', action='store_true', help='Apply the monkey patch for using LoRAs with quantized models.')
parser.add_argument('--no-warmup_autotune', action='store_true', help='GPTQ: Disable warmup autotune for triton.') parser.add_argument('--no-quant_attn', action='store_true', help='(triton) Disable quant attention. If you encounter incoherent results try disabling this.')
parser.add_argument('--no-fused_mlp', action='store_true', help='GPTQ: Disable fused mlp for triton. If you encounter "Unexpected mma -> mma layout conversion" try disabling this.') parser.add_argument('--no-warmup_autotune', action='store_true', help='(triton) Disable warmup autotune.')
parser.add_argument('--monkey-patch', action='store_true', help='GPTQ: Apply the monkey patch for using LoRAs with quantized models.') parser.add_argument('--no-fused_mlp', action='store_true', help='(triton) Disable fused mlp. If you encounter "Unexpected mma -> mma layout conversion" try disabling this.')
# FlexGen # FlexGen
parser.add_argument('--flexgen', action='store_true', help='Enable the use of FlexGen offloading.') parser.add_argument('--flexgen', action='store_true', help='Enable the use of FlexGen offloading.')