mlx-community/NeoHorse-1-4B-4bit - Каталог нейросетей
Генерация текста

mlx-community/NeoHorse-1-4B-4bit

Добавлено:
mlx-community/NeoHorse-1-4B-4bit

A 4-bit MLX port of NeoHorse-1-4B by TokenRhythm, quantized for Apple Silicon. This is the same fine-tuned model — just repacked as a uniform 4-bit safetensor so it runs on MLX. Vision weights are not included; this is text-only, matching the upstream release. The tensor values are unchanged. The only difference is the quantization and the format swap. NeoHorse-1-4B is a hybrid-attention model post-trained from Qwen3.5-4B: — 32 layers, mostly linear attention with full-attention layers every 4th position (4 total: layers 3, 7, 11, 15, 19, 23, 27, 31) — GQA: 16 query heads, 4 KV heads — Partial RoPE (0.25 rotary factor) — only a quarter of dimensions get positional encoding — 262K context natively The hybrid attention means the KV cache grows slowly — only the full-attention layers add per-token state, so long contexts stay cheap. The output was coherent chain-of-thought about special and general relativity. Quantization held up. You’ll also need to patch mlx-lm’s model remapping if it doesn’t already include qwen35text. In mlxlm/utils.py`, add: (If you’re on mlx-lm >= 0.31.3 with the patch already applied, this works out of the box.) The model fits comfortably in 8 GB RAM, but 16…

Модальности:
Генерация текста

Области применения:
Генерация кода Логика и рассуждение Диалог / чат Вызов функций (Tool use)


Задача: Генерация текста
Автор: mlx-community
Теги: mlx_lm, qwen3_5, mlx, 4-bit, apple-silicon, agentic, tool-use, coding
Лайков: 3  |  Загрузок: 160

Открыть на HuggingFace →

Описание основано на материалах HuggingFace. Перевод выполнен автоматически.