opus-4b-py-mixed-step150-2026-04-29

LoRA adapter (rank 32) trained with RL on a custom Opus-Magnum-style motion-planning task using the py answer representation. Snapshot at training step 150 / 300.

Source training run

• wandb: opus-4b-py-mixed-2026-04-29 (0hlmymwt)
• tinker checkpoint: tinker://55a5ea09-5855-5635-99bc-5aa39f35d08a:train:0/sampler_weights/000150
• distances: 1, 2, 3, 4
• task types: move, transmute, bond
• hard_task_fraction: 0.15 (cap on bond + d=4 share of train pool)
• learning rate: 1e-5
• group size: 8, groups per batch: 16
• renderer: qwen3_5_disable_thinking

Usage

from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

base = "Qwen/Qwen3.5-4B"
adapter = "maxbittker/opus-4b-py-mixed-step150-2026-04-29"
tok = AutoTokenizer.from_pretrained(base)
model = AutoModelForCausalLM.from_pretrained(base, device_map="auto")
model = PeftModel.from_pretrained(model, adapter)