---
license: apache-2.0
base_model: AnkitAI/Parable-Qwen3-4B-Claude-Fable-5
tags:
- mlx
- apple-silicon
- qwen3
- agentic
- coding
- lora
library_name: mlx
pipeline_tag: text-generation
---

# Parable-Qwen3-4B-Claude-Fable-5 — MLX 4-bit

Apple Silicon build of [Parable-Qwen3-4B-Claude-Fable-5](https://huggingface.co/AnkitAI/Parable-Qwen3-4B-Claude-Fable-5),
a Qwen3-4B fine-tuned on execution-verified agent traces.

**2.1 GB, 4.501 bits per weight.** Runs on any M-series Mac with room to spare.

## Use it

```bash
pip install mlx-lm
mlx_lm.generate --model AnkitAI/Parable-Qwen3-4B-Claude-Fable-5-MLX-4bit \
  --prompt "Write a Python function that retries an HTTP call with backoff."
```

Or in Python:

```python
from mlx_lm import load, generate
model, tokenizer = load("AnkitAI/Parable-Qwen3-4B-Claude-Fable-5-MLX-4bit")
print(generate(model, tokenizer, prompt="...", max_tokens=512))
```

## What it is

Same weights as the source model, quantised to 4-bit for MLX. The recipe behind
it is v3.1: LoRA on agent traces plus a replay mix, completion-only loss, two
seeds souped, then merged into the base at scale 0.6 to limit drift.

Measured on the 4B, base against tuned in one session on one harness:

| | base | v3.1 |
|---|---|---|
| HumanEval+ | 0.616 | **0.683** |
| MBPP+ | 0.603 | **0.638** |

Those numbers are from the full-precision model. Quantisation to 4 bits costs
some accuracy; they are the ceiling, not a promise for this build.

## Other formats

- [GGUF](https://huggingface.co/AnkitAI/Parable-Qwen3-4B-Claude-Fable-5-GGUF) — llama.cpp, LM Studio, Ollama
- [safetensors](https://huggingface.co/AnkitAI/Parable-Qwen3-4B-Claude-Fable-5) — transformers

Apache-2.0, same as the base.