manu02/HY-1.8B-2Bit-bnb-4bit-nf4-dq

HY-1.8B-2Bit (Quantized)

Description

This model is a 4-bit quantized version of the original AngelSlim/HY-1.8B-2Bit model, optimized for reduced memory usage while maintaining performance.

Quantization Details

• Quantization Type: 4-bit
• bnb_4bit_quant_type: nf4
• bnb_4bit_use_double_quant: True
• bnb_4bit_compute_dtype: bfloat16
• bnb_4bit_quant_storage: uint8
• Original Footprint: 3582.16 MB (BFLOAT16)
• Quantized Footprint: 1761.78 MB (UINT8)
• Memory Reduction: 50.8%

Usage

from transformers import AutoModel, AutoTokenizer

model_name = "HY-1.8B-2Bit-bnb-4bit-nf4"
model = AutoModel.from_pretrained(
    "manu02/HY-1.8B-2Bit-bnb-4bit-nf4",
)
tokenizer = AutoTokenizer.from_pretrained("manu02/HY-1.8B-2Bit-bnb-4bit-nf4", use_fast=True)