license: apple-amlr
base_model:
- Qwen/Qwen3-4B-Instruct-2507
tags:
- self-distillation
- code-generation
- ssd
library_name: transformers
SSD-Qwen3-4B-Instruct
This model was produced using Simple Self-Distillation (SSD), a method that improves code generation by fine-tuning a language model on its own sampled outputs—without rewards, verifiers, teacher models, or reinforcement learning.
- Base model: Qwen/Qwen3-4B-Instruct-2507
- Variant: instruct
- Self-distillation sampling: temperature=1.6, top_p=0.8, top_k=20
- Evaluation sampling: temperature=1.1, top_p=0.8, top_k=20
Method
SSD samples solutions from the base model using non-unit temperature and top-k/top-p truncation, then fine-tunes on those samples via standard supervised learning. Despite its simplicity, SSD yields large gains on competitive programming benchmarks, with improvements concentrating on harder problems. The mechanism traces to resolving a precision–exploration conflict: SSD reshapes token distributions in a context-dependent way so that a single global decoding configuration becomes far more effective at evaluation time.
Results
LiveCodeBench (%)
| Model | LCBv6 pass@1 | LCBv6 pass@5 | LCBv5 pass@1 | LCBv5 pass@5 |
|---|---|---|---|---|
| Qwen3-4B-Instruct-2507 (base) | 34.0 | 41.0 | 34.3 | 45.4 |
| + SSD (this model) | 41.5 (+7.5) | 56.8 (+15.8) | 45.7 (+11.4) | 61.9 (+16.5) |
Paper
Embarrassingly Simple Self-Distillation Improves Code Generation
Ruixiang Zhang, Richard He Bai, Huangjie Zheng, Navdeep Jaitly, Ronan Collobert, Yizhe Zhang
Usage
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("apple/SSD-Qwen3-4B-Instruct")
tokenizer = AutoTokenizer.from_pretrained("apple/SSD-Qwen3-4B-Instruct")
License
This model is released under the Apple Machine Learning Research Model License.