该模型为使用ModelScope Trainer 微调的模型

• 基础模型：chatglm2-6b
• 任务类型：generation
• 任务名称：chatglm2-6b-11

评估结果

示例代码

此模型需要用Lora方式进行推理

from modelscope import Model, pipeline, read_config
from modelscope.metainfo import Models
from modelscope.swift import Swift
import torch
from modelscope.hub.snapshot_download import snapshot_download
import os.path as osp
from modelscope.swift.lora import LoRAConfig
from modelscope.utils.config import ConfigDict
from modelscope.hub.api import HubApi

YOUR_ACCESS_TOKEN = '请从ModelScope个人中心->访问令牌获取'
api = HubApi()
api.login(YOUR_ACCESS_TOKEN)

lora_config = LoRAConfig(
    replace_modules=['attention.query_key_value'],
    rank=32,
    lora_alpha=32,
    lora_dropout=0.05
    )

model_dir = 'ZhipuAI/chatglm2-6b'
model_config = read_config(model_dir)
model_config['model'] = ConfigDict({'type': Models.chatglm2_6b})
model = Model.from_pretrained(model_dir, cfg_dict=model_config)
model = model.bfloat16()
Swift.prepare_model(model, lora_config)
# flex train 训练得到的模型'xxxxxxxxxx'，snapshot_download不支持model_revision,所以用revision
work_dir = snapshot_download('xxxxxxxxxx',revision='xxx')
state_dict = torch.load(osp.join(work_dir, 'pytorch_model.bin'))
model.load_state_dict(state_dict)
pipe = pipeline('chat', model, pipeline_name='chatglm2_6b-text-generation')
result_zh = pipe({
        'text':
        '简述曼德拉效应',
        'history': []
    })
print(result_zh['response'])