mmarch/text-generation-webui
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Update to support GPTQ triton commit c90adef (#1229)

Browse source
This commit is contained in:
sgsdxzy 2023-04-17 12:11:18 +08:00 • committed by GitHub
parent 209fcd21d5
commit b57ffc2ec9
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
3 changed files with 38 additions and 23 deletions
Download patch file Download diff file Expand all files Collapse all files

2
README.md
View file

@ -236,7 +236,9 @@ Optionally, you can use the following command-line flags:
| `--model_type MODEL_TYPE` | GPTQ: Model type of pre-quantized model. Currently LLaMA, OPT, and GPT-J are supported. |
| `--groupsize GROUPSIZE` | GPTQ: 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. |
| `--no-quant_attn` | GPTQ: Disable quant attention for triton. If you encounter incoherent results try disabling this. |
| `--no-warmup_autotune` | GPTQ: Disable warmup autotune for triton. |
| `--no-fused_mlp` | GPTQ: Disable fused mlp for triton. If you encounter "Unexpected mma -> mma layout conversion" try disabling this. |
| `--monkey-patch` | GPTQ: Apply the monkey patch for using LoRAs with quantized models. |
#### FlexGen