metadata

language:
  - en
license: apache-2.0
tags:
  - mlx
datasets:
  - Locutusque/hercules-v5.0
inference:
  parameters:
    do_sample: true
    temperature: 0.8
    top_p: 0.95
    top_k: 40
    min_p: 0.1
    max_new_tokens: 250
    repetition_penalty: 1.1

mlx-community/Hercules-5.0-Qwen2-1.5B-4bits

The Model mlx-community/Hercules-5.0-Qwen2-1.5B-4bits was converted to MLX format from M4-ai/Hercules-5.0-Qwen2-1.5B using mlx-lm version 0.14.0.

Use with mlx

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("mlx-community/Hercules-5.0-Qwen2-1.5B-4bits")
response = generate(model, tokenizer, prompt="hello", verbose=True)