Downloads · 30 days
68
71% of all-time downloads
qwen2.5-0.5b-ok-ppo-kineteq
Ehraim/qwen2.5-0.5b-ok-ppo-kineteq
qwen2.5-0.5b-ok-ppo-kineteq is a reinforcement learning model from Ehraim. 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. The card lists the license as apache-2.0.
Kineteq file-manager policy trained with ok-ppo on Qwen/Qwen2.5-0.5B-Instruct.
All-time downloads
96
Public
Repo size
15.8 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.json11.4 MB · 72%
At a glance
- Task
- Reinforcement Learning
- Library
- peft
- License
- apache-2.0
- Access
- Public
- Created
- Aug 28, 2026
- Updated
- Aug 28, 2026
- SHA
- b5785b7a
Base models
Model card
From the Hugging Face model README
Ehraim__qwen2.5-0.5b-ok-ppo-kineteq
Kineteq file-manager policy trained with ok-ppo on Qwen/Qwen2.5-0.5B-Instruct.
The Hub folder includes the adapter (or stub weights) and the algorithm primitives used to train it (primitives/*.py): PPO critic + clip, GRPO group z-score, OK-PPO (PPO + spherical kernel-bank OOD reward), OK-GRPO.
Load
from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel
base = "Qwen/Qwen2.5-0.5B-Instruct"
tok = AutoTokenizer.from_pretrained("Ehraim__qwen2.5-0.5b-ok-ppo-kineteq")
model = AutoModelForCausalLM.from_pretrained(base)
model = PeftModel.from_pretrained(model, "Ehraim__qwen2.5-0.5b-ok-ppo-kineteq")
Held-out file-manager benchmark
See bench.json in this repo.
Source: ok-grpo. Stub=False.
- Task
- Reinforcement Learning
- Library
- peft
- License
- apache-2.0
- Languages
- en
- Created
- Aug 28, 2026
- Updated
- Aug 28, 2026