EVOKE-ALFWorld-3B/README.md

base_model, base_model_relation, library_name, pipeline_tag, license, license_name, license_link, tags

base_model	base_model_relation	library_name	pipeline_tag	license	license_name	license_link	tags
Qwen/Qwen2.5-3B-Instruct	finetune	transformers	text-generation	other	qwen-research	https://huggingface.co/Qwen/Qwen2.5-3B-Instruct/blob/main/LICENSE	evoke alfworld agent arxiv:2609.38334

EVOKE-ALFWorld-3B

EVOKE is a post-training method that elicits world knowledge for goal-directed decisions. This is the full model for ALFWorld, based on Qwen2.5-3B-Instruct.

Paper · Code and evaluation · Project page

Load

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "Gnonymous/EVOKE-ALFWorld-3B"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id, torch_dtype="auto", device_map="auto"
)

ALFWorld evaluation

Follow the installation and data instructions in the GitHub repository, then run:

.venv/inference/bin/python evaluate_alfworld.py \
  --model Gnonymous/EVOKE-ALFWorld-3B \
  --data-root /path/to/alfworld/json_2.1.1 \
  --output /path/to/results/evoke-3b \
  --gpus 0

Built with Qwen. See the included LICENSE and NOTICE files.