Downloads · 30 days
0
farffadet/syllogym-judge-qwen3-4b-grpo
syllogym-judge-qwen3-4b-grpo is a machine learning model from farffadet. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
Fine-tuned from Qwen/Qwen3-4B on the SylloGym legal reasoning environment using GRPO (Group Relative Policy Optimization).
Downloads · 30 days
0
Access
Public
Updated Apr 12, 2026
Repo size
411 MB
Likes
0
Public
Click a slice to open those files.
.safetensors264 MB · 64%
From the Hugging Face model README
Fine-tuned from Qwen/Qwen3-4B on the SylloGym legal reasoning environment using GRPO (Group Relative Policy Optimization).
Training: 180 steps on an A100, multi-turn rollout via TRL + Unsloth
Environment: 12 legal domains, 45 tasks, deterministic Python verifiers
Result: +6.1 pp overall accuracy (61.7% → 67.8%), +8.3 pp on 5-turn episodes
This is a LoRA adapter — load it on top of the base Qwen3-4B model.
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer
base = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-4B")
model = PeftModel.from_pretrained(base, "farffadet/syllogym-judge-qwen3-4b-grpo")
tokenizer = AutoTokenizer.from_pretrained("farffadet/syllogym-judge-qwen3-4b-grpo")
| Parameter | Value |
|---|---|
| Base model | Qwen/Qwen3-4B (4-bit, LoRA r=32) |
| Steps | 180 (target: 300) |
| Batch size | 8 |
| Temperature | 0.9 |
| KL penalty (β) | 0.02 |
| Loss type | DR-GRPO |
| max_completion_length | 3072 tokens |