The Model mlx-community/SorcererLM-8x22b-4bit was converted to MLX format from rAIfle/SorcererLM-8x22b-bf16 using mlx-lm version 0.19.0.
pip install mlx-lm
from mlx_lm import load, generate
model, tokenizer = load("mlx-community/SorcererLM-8x22b-4bit")
prompt="hello"
if hasattr(tokenizer, "apply_chat_template") and tokenizer.chat_template is not None:
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
messages, tokenize=False, add_generation_prompt=True
)
response = generate(model, tokenizer, prompt=prompt, verbose=True)
model-00001-of-00016.safetensors 5.3 GBmodel-00014-of-00016.safetensors 5.2 GBmodel-00011-of-00016.safetensors 5.2 GBmodel-00008-of-00016.safetensors 5.2 GBmodel-00005-of-00016.safetensors 5.2 GBmodel-00015-of-00016.safetensors 5.2 GBmodel-00006-of-00016.safetensors 5.2 GBmodel-00009-of-00016.safetensors 5.2 GBmodel-00012-of-00016.safetensors 5.2 GBmodel-00002-of-00016.safetensors 5.2 GBmodel-00003-of-00016.safetensors 5.2 GBmodel-00007-of-00016.safetensors 5.1 GBmodel-00013-of-00016.safetensors 5.1 GBmodel-00004-of-00016.safetensors 5.1 GBmodel-00010-of-00016.safetensors 5.1 GBmodel-00016-of-00016.safetensors 1.5 GB有文件?提交种子。
提交种子