标签:

• 合并
• mergekit
• lazymergekit
• localfultonextractor/Erosumika-7B-v3-0.2
• KoboldAI/Mistral-7B-Holodeck-1 基础模型:
• localfultonextractor/Erosumika-7B-v3-0.2
• KoboldAI/Mistral-7B-Holodeck-1

HoloSumika-7B测试版

初步测试印象：这个模型似乎（不）稳定

HoloSumika-7B测试版是通过LazyMergekit合并以下模型而成：

• localfultonextractor/Erosumika-7B-v3-0.2
• KoboldAI/Mistral-7B-Holodeck-1

🧩 配置

切片:
  - 源模型:
      - 模型: localfultonextractor/Erosumika-7B-v3-0.2
        层范围: [0, 32]
      - 模型: KoboldAI/Mistral-7B-Holodeck-1
        层范围: [0, 32]
合并方法: slerp
基础模型: localfultonextractor/Erosumika-7B-v3-0.2
参数:
  t:
    - 值: 0.32
数据类型: bfloat16

💻 使用方式

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "son-of-man/HoloSumika-7B-test"
messages = [{"role": "user", "content": "什么是大语言模型？"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "文本生成",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])