how to get the 473M model from Qwen/Qwen1.5-0.5B-Chat

#1
by chengfeng17 - opened

I quantize Qwen/Qwen1.5-0.5B-Chat to int4
But my quantized model is 747M
my quantize_config.json
{
"bits": 4,
"group_size": 128,
"damp_percent": 0.01,
"desc_act": false,
"static_groups": false,
"sym": true,
"true_sequential": true,
"model_name_or_path": null,
"model_file_base_name": "model",
"is_marlin_format": false,
"quant_method": "gptq"
}
Thanks for advice

0.5B models use tied word embeddings; that's about 155M parameters and 310MB storage.

chengfeng17 changed discussion status to closed

Sign up or log in to comment