Text Generation
MLX
Safetensors
qwen3_5_mtp
qwen3.8
multi-token-prediction
speculative-decoding
qwen3_5
apple-silicon
4-bit precision
Instructions to use junafinity/qwen38-mtp-head-fc-bf16-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use junafinity/qwen38-mtp-head-fc-bf16-4bit with MLX:
# Make sure mlx-lm is installed # pip install --upgrade mlx-lm # if on a CUDA device, also pip install mlx[cuda] # Generate text with mlx-lm from mlx_lm import load, generate model, tokenizer = load("junafinity/qwen38-mtp-head-fc-bf16-4bit") prompt = "Once upon a time in" text = generate(model, tokenizer, prompt=prompt, verbose=True) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- MLX LM
How to use junafinity/qwen38-mtp-head-fc-bf16-4bit with MLX LM:
Generate or start a chat session
# Install MLX LM uv tool install mlx-lm # Generate some text mlx_lm.generate --model "junafinity/qwen38-mtp-head-fc-bf16-4bit" --prompt "Once upon a time"
- Atomic Chat
docs: link Hub collection on the model card
Browse files
README.md
CHANGED
|
@@ -21,6 +21,8 @@ Apple Silicon. Derived from the organizer-pinned
|
|
| 21 |
|
| 22 |
|
| 23 |
## Pairing
|
|
|
|
|
|
|
| 24 |
|
| 25 |
This is a **drafting head only**, not a standalone model. Pair it with a Qwen3.8-27B MLX backbone (for example [Qwen-3.8-27B-Uncensored-8-Bit-MLX](https://huggingface.co/junafinity/Qwen-3.8-27B-Uncensored-8-Bit-MLX)) in a loader that accepts a scales-keyed mixed-precision head.
|
| 26 |
|
|
|
|
| 21 |
|
| 22 |
|
| 23 |
## Pairing
|
| 24 |
+
Hub collection: [https://huggingface.co/collections/junafinity/qwen-38-27b-uncensored-apple-silicon-6a896c726b52be3a0b63400e](https://huggingface.co/collections/junafinity/qwen-38-27b-uncensored-apple-silicon-6a896c726b52be3a0b63400e)
|
| 25 |
+
|
| 26 |
|
| 27 |
This is a **drafting head only**, not a standalone model. Pair it with a Qwen3.8-27B MLX backbone (for example [Qwen-3.8-27B-Uncensored-8-Bit-MLX](https://huggingface.co/junafinity/Qwen-3.8-27B-Uncensored-8-Bit-MLX)) in a loader that accepts a scales-keyed mixed-precision head.
|
| 28 |
|