--- base_model: Qwen/Qwen2.5-3B-Instruct library_name: transformers pipeline_tag: text-generation tags: - uopd - alfworld - on-policy-distillation - llm-agent --- # UOPD-Qwen2.5-3B-Instruct-ALFWorld Final UOPD student checkpoint trained on ALFWorld with the frozen teacher [`langfeng01/GiGPO-Qwen2.5-7B-Instruct-ALFWorld`](https://huggingface.co/langfeng01/GiGPO-Qwen2.5-7B-Instruct-ALFWorld). ## Usage ```python import torch from transformers import AutoModelForCausalLM, AutoTokenizer model_id = "Wenboz/UOPD-Qwen2.5-3B-Instruct-ALFWorld" model = AutoModelForCausalLM.from_pretrained( model_id, torch_dtype=torch.bfloat16, device_map="auto", ) tokenizer = AutoTokenizer.from_pretrained(model_id) ``` ## Training configuration | Setting | Value | |---|---| | Student initialization | `Qwen/Qwen2.5-3B-Instruct` | | Teacher | `langfeng01/GiGPO-Qwen2.5-7B-Instruct-ALFWorld` | | Training steps | 250 | | Rollout / training batch | 16 / 64 | | Environment step limit | 50 | | Max prompt / response tokens | 2,048 / 512 | | Optimizer | AdamW, learning rate 1e-6 | | Target intervention rate | linear 0.30 → 0.05 over the first 120 training steps, then 0.05 | | Triggered-turn loss | Plain SFT, weight 1.0 | | Teacher takeover | Enabled |