DESUCLUB/Llama-3.1-8B-Instruct-bf16-quantized.w8a8

This is a custom W8A16 quantized version of meta-llama/Llama-3.1-8B-Instruct.

Quantization Details

• Method: Custom W8A16 (8-bit weights, 16-bit activations)
• Weight precision: INT8
• Scale precision: BF16
• Quantization: Symmetric per-channel
• Zero points: None (symmetric)

Model Structure

The quantized model contains:

• .weight: INT8 quantized weights
• .weight_scale: BF16 scale parameters (trainable)
• Standard embedding and normalization layers in original precision

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

# Note: This requires custom quantization code to load properly
model = AutoModelForCausalLM.from_pretrained("DESUCLUB/Llama-3.1-8B-Instruct-bf16-quantized.w8a8")
tokenizer = AutoTokenizer.from_pretrained("DESUCLUB/Llama-3.1-8B-Instruct-bf16-quantized.w8a8")