| --- |
| base_model: Qwen/Qwen3-4B-Instruct-2507 |
| library_name: peft |
| tags: |
| - lora |
| - qwen3 |
| - grpo |
| - anonymous-submission |
| --- |
| |
| # Anonymous Chainer GRPO Checkpoint 2 |
|
|
| This repository contains a LoRA adapter checkpoint for the Chainer policy after the later GRPO stage checkpoint used in the paper workflow. |
|
|
| ## Base model |
|
|
| - `Qwen/Qwen3-4B-Instruct-2507` |
|
|
| ## Artifact type |
|
|
| - PEFT LoRA adapter |
| - Inference-only checkpoint export |
|
|
| ## Notes |
|
|
| - This repository is provided as anonymous supplementary material for peer review. |
| - The checkpoint is released for verification and follow-up research use. |
| - The repository intentionally omits author-identifying metadata. |
|
|
| ## Loading example |
|
|
| ```python |
| from transformers import AutoModelForCausalLM, AutoTokenizer |
| from peft import PeftModel |
| |
| base_model_id = "Qwen/Qwen3-4B-Instruct-2507" |
| adapter_repo_id = "Anonymous5345345/anonymous-chainer-grpo-epoch2" |
| |
| tokenizer = AutoTokenizer.from_pretrained(base_model_id) |
| base_model = AutoModelForCausalLM.from_pretrained(base_model_id) |
| model = PeftModel.from_pretrained(base_model, adapter_repo_id) |
| ``` |
|
|
