Nexa_0.2_Enhanced / README.md
ItsMatti4's picture
Update README.md
d6f8cf5 verified
metadata
license: apache-2.0
base_model:
  - WeAreNexa/Nexa_0.1_First

Nexa 0.2 Enhanced πŸš€

Nexa 0.2 Enhanced is a high-performance, lightweight language model. This version is a direct evolution of the Nexa 0.1, optimized for better instruction following and faster inference speeds.


πŸ“Š Model Details

  • Model Name: Nexa 0.2 Enhanced
  • Base Model: Nexa 0.1
  • Library: Unsloth
  • License: Apache 2.0
  • Pipeline Tag: text-generation
  • Tags: unsloth, lora, peft, text-generation

πŸ›  Technical Specifications

This model was fine-tuned using LoRA (Low-Rank Adaptation) via the Unsloth library. It is specifically designed to run efficiently on consumer hardware.

  • Framework Version: PEFT 0.18.1
  • Quantization: 4-bit (bitsandbytes)
  • Architecture: Optimized for low VRAM usage

πŸš€ Quick Start

To use Nexa 0.2 Enhanced in your Python environment, you can use the following code:

from unsloth import FastLanguageModel
import torch

model, tokenizer = FastLanguageModel.from_pretrained(
    model_name = "your-username/nexa-0.2-enhanced",
    max_seq_length = 2048,
    load_in_4bit = True,
)
FastLanguageModel.for_inference(model)

# Example usage
inputs = tokenizer(["How does Nexa 0.2 perform?"], return_tensors = "pt").to("cuda")
outputs = model.generate(**inputs, max_new_tokens = 64)
print(tokenizer.batch_decode(outputs))