This model is a 4-bit quantized version of the original Nanbeige/Nanbeige4.1-3B 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: 7867.27 MB (BFLOAT16)
Quantized Footprint: 3243.05 MB (UINT8)
Memory Reduction: 58.8%

Usage

python

from transformers import AutoModel, AutoTokenizer

model_name = "Nanbeige4.1-3B-bnb-4bit-nf4"
model = AutoModel.from_pretrained(
    "manu02/Nanbeige4.1-3B-bnb-4bit-nf4",
)
tokenizer = AutoTokenizer.from_pretrained("manu02/Nanbeige4.1-3B-bnb-4bit-nf4", use_fast=True)

manu02/Nanbeige4.1-3B-bnb-4bit-nf4-dq

Nanbeige4.1-3B (Quantized)

Description

Quantization Details

Usage

manu02/
Nanbeige4.1-3B-bnb-4bit-nf4-dq