README.md

---
license: apache-2.0
base_model: Qwen/Qwen3.5-122B-A10B
library_name: peft
pipeline_tag: text-generation
tags: [lora, peft, grpo, swe-bench, code, newtype, rlvr, dense-reward, ablation]
---

# Qwen3.5-122B-A10B-Newtype-Tuned-v20-grpo-ckpt80

Checkpoint at step 80 of v20 GRPO training (intermediate snapshot).

- **init**: v5 LoRA (mix corpus, 30% Pass@1 baseline)
- **trainer**: TRL GRPOTrainer
- **rollout**: HF model.generate (k=8, T=1.0)
- **reward**: dense [0, 1.0] = parse 0.05 + grep 0.05 + file 0.10 + func 0.10 + harness 0.30/0.70
- **MoE safeguards**: output_router_logits + aux loss + explicit router freeze
- **corpus**: SWE-bench-Lite 50-task train pool (subset of 270 non-eval)
- **hyperparams**: β=0.1, ε=0.2, lr=1e-6, 80 steps (intermediate, full = 100)

**30-task smoke result**: 7/30 = 23.3% Pass@1 (same as final/step 100).

**Specialization finding**: this checkpoint and the step-100 final share
only 3/7 PASS tasks. Together they cover 11/30 = 36.7% (oracle ensemble).
See companion repo
[`Qwen3.5-122B-A10B-Newtype-Tuned-v20-grpo`](https://huggingface.co/newtype-2038/Qwen3.5-122B-A10B-Newtype-Tuned-v20-grpo)
for step-100 final.

## v20 training journey

See [Newtype SFT method doc](https://github.com/kr-ai-dev-association/newtype-framework/blob/main/agent-evaluation/docs/sft-dense-grpo.md)
for full v5 → v20 → v21 pipeline + ablation context.