metadata
language:
- en
- zh
pipeline_tag: text-generation
tags:
- gptq
- int4
- yi1.5-6B-Chat
- pytorch
license: apache-2.0
license_name: apache-2.0
license_link: LICENSE
About Quantization
我们使用modelscope swift仓库进行GPTQ 4bit量化. 量化文档可以查看这里. 量化命令如下:
We use the modelscope swift repository to perform GPTQ 4bit quantization. Quantization documentation can be found here. The quantization command is as follows:
# Experimental Environment: A100
swift export \
--quant_bits 4 \
--model_type yi-1_5-6b-chat \
--quant_method gptq \
--dataset alpaca-zh alpaca-en sharegpt-gpt4-mini \
--quant_seqlen 4096
Inference:
CUDA_VISIBLE_DEVICES=0 swift infer --model_type yi-1_5-6b-chat-gptq-int4
SFT:
CUDA_VISIBLE_DEVICES=0 swift sft --model_type yi-1_5-6b-chat-gptq-int4 --dataset leetcode-python-en
Original Model: