Downloads · 30 days
12
15% of all-time downloads
dh_ministal_gpro
dmayboroda/dh_ministal_gpro
dh_ministal_gpro is a reinforcement learning model from dmayboroda. Use it for the reinforcement learning task on the model card, and read the license before you ship it in a product. It is set up for peft.
LoRA adapter for mistralai/Ministral-3-8B-Instruct-2512-BF16, fine-tuned with GRPO to play Duck Hunt.
All-time downloads
78
Public
Repo size
158 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors70.6 MB · 81%
At a glance
- Task
- Reinforcement Learning
- Library
- peft
- Access
- Public
- Created
- Mar 1, 2026
- Updated
- Mar 1, 2026
- SHA
- e9c3dc83
Base models
Model card
From the Hugging Face model README
dh_ministal_gpro
LoRA adapter for mistralai/Ministral-3-8B-Instruct-2512-BF16, fine-tuned with GRPO to play Duck Hunt.
Training
- Method: Group Relative Policy Optimization (GRPO)
- LoRA rank: 16, alpha: 32
- Target modules: q_proj, k_proj, v_proj, o_proj
- Learning rate: 5e-06
- Generations per prompt: 4
- Best eval hit rate: -100.0%
Usage
The adapter produces Mistral native tool calls ([TOOL_CALLS]), compatible with
OpenAI SDK via vLLM or TGI.
from peft import AutoPeftModelForCausalLM
from transformers import AutoProcessor
model = AutoPeftModelForCausalLM.from_pretrained("dmayboroda/dh_ministal_gpro")
processor = AutoProcessor.from_pretrained("dmayboroda/dh_ministal_gpro")
- Task
- Reinforcement Learning
- Library
- peft
- Created
- Mar 1, 2026
- Updated
- Mar 1, 2026