Website  |  GitHub  |  Twitter  |  Dataset  |  YOLO-Coder-8B

License: MIT  |  Author: @erdemwrites

YOLO-Coder-1.5B

Fix broken CLI commands. One command output. Runs on any machine. Fine-tuned Qwen2.5-Coder-1.5B · MLX LoRA on Apple Silicon · No API key needed

🎯 Task	CLI error → single bare bash fix command
🏆 Accuracy	71.1% pipeline×3 · 42.2% raw LLM
💾 Size	~941MB Q4_K_M GGUF · ~2GB RAM
⚡ Speed	<1s on Apple Silicon
🔒 Privacy	100% local · no API key · no telemetry

Quickstart

ollama run hf.co/erdemozkan/YOLO-Coder-1.5B "ModuleNotFoundError: No module named 'flask'"
# → pip install flask

That's it. No account. No cloud. No cost per call.

Benchmark — YOLO-Bench

218 verified CLI errors · structural match scoring (flag-order-independent)

YOLO-Coder-8B  pipeline×3  ████████████████████  77.1%
YOLO-Coder-1.5B pipeline×3 ██████████████████    71.1%  ★ this model
Claude Sonnet  raw         ████████████████       60.1%
YOLO-Coder-8B  raw         ███████████████        59.2%
GPT-4o         raw         ████████████           48.6%
YOLO-Coder-1.5B raw        ██████████             42.2%

Mode	Structural Match
Raw LLM (no pipeline)	42.2%
Pipeline × 1 (interceptors + LLM)	66.5%
Pipeline × 3 (interceptors + memory + 3 LLM attempts)	71.1%

At ~941MB, YOLO-Coder-1.5B reaches 71.1% with the full pipeline — running entirely offline.

Scoring code and dataset: github.com/erdemozkan/YOLO-CODER/tree/main/benchmark

How the pipeline works

Your error → [91 interceptors <1ms] → [fix memory <5ms] → [LLM <1s] → Fix
                ↑ ~50% of fixes stop here

Half of all fixes never reach the LLM. The model is the safety net, not the first guess.

Usage with YOLO-CODER

pip install yolo-coder

yoco --model hf.co/erdemozkan/YOLO-Coder-1.5B python3 myapp.py
yoco --model hf.co/erdemozkan/YOLO-Coder-1.5B npm run dev

Prompt format (ChatML)

<|im_start|>system
You are a CLI repair tool. Output ONLY a single bare bash command to fix the error. No explanation. No markdown. No backticks.<|im_end|>
<|im_start|>user
[Linux] $ python3 myapp.py
Error:
ModuleNotFoundError: No module named 'requests'
FIX:<|im_end|>
<|im_start|>assistant
pip install requests<|im_end|>

Training

"Trained on a MacBook Air. No rented A100s."

Property	Value
Base model	Qwen/Qwen2.5-Coder-1.5B-Instruct
Fine-tune method	LoRA via MLX on Apple Silicon
LoRA rank / scale	8 / 20.0
Layers trained	16
Training iterations	500
Learning rate	1e-5
Training examples	6,719 error/fix pairs across 15 categories
Export	Merged weights → Q4_K_M GGUF for Ollama

Files

File	Description
YOLO-Coder-1.5B-Q4_K_M.gguf	Q4_K_M quantized GGUF (~941MB) — use this with Ollama
safetensors/	fp16 safetensors — for further fine-tuning

1.5B vs 8B

	YOLO-Coder-1.5B	YOLO-Coder-8B
Size	~941MB	~4.4GB
RAM needed	~2GB	~6GB
Speed	<1s on Apple Silicon	1–3s on Apple Silicon
Raw accuracy	42.2%	59.2%
Pipeline×3 accuracy	71.1%	77.1%
Best for	Speed, low-RAM machines	Hard errors, best accuracy

Limitations

• Single-command output only — not designed for multi-step fixes without a wrapper
• Complex or highly novel errors may produce suboptimal output
• Not a general-purpose coding assistant

License

MIT