dolev31

ProactiveInquirer-Qwen3-8B-Merged

README

License: apache-2.0

Results

The results are the trained questioner's, as the paper reports them: see the adapter card's Results. This merged model reproduces the adapter's output on that card's example, as the paragraph above says.

How to use it

The questioner reads the prompt template it was trained on, in prompts/, and replies with one JSON action per step: {"action": "ask", "question": ...} or {"action": "stop", ...}. Keep Qwen3's thinking off, as in training.

python
import re

import torch
from huggingface_hub import hf_hub_download
from transformers import AutoModelForCausalLM, AutoTokenizer

REPO = "dolev31/ProactiveInquirer-Qwen3-8B-Merged"
tok = AutoTokenizer.from_pretrained(REPO)
model = AutoModelForCausalLM.from_pretrained(REPO, dtype=torch.bfloat16, device_map="auto")
template = open(hf_hub_download(REPO, "prompts/inquirer_prompted.txt"), encoding="utf-8").read()
placebo = open(hf_hub_download(REPO, "prompts/fragment_user_channel_placebo.txt"), encoding="utf-8").read()


def next_action(**state):
    fields = dict(state, user_channel=placebo.strip())
    prompt = re.sub(r"\{\{(\w+)\}\}", lambda m: str(fields[m.group(1)]), template)
    ids = tok.apply_chat_template(
        [{"role": "user", "content": prompt}],
        add_generation_prompt=True,
        enable_thinking=False,
        return_tensors="pt",
        return_dict=True,
    ).to(model.device)
    out = model.generate(**ids, max_new_tokens=200, do_sample=False)
    return tok.decode(out[0, ids["input_ids"].shape[1] :], skip_special_tokens=True)


print(next_action(
    question="Who was the spouse of the director of the film The Great Flamarion?",
    instructions="Answer the question using a closed pool of 20 paragraphs. You may issue retrieval "
    "queries against that pool before answering; several paragraphs are distractors, and the answer "
    "usually requires composing facts from more than one of them.",
    evidence="(nothing retrieved yet)", draft="(no draft yet)", history="(nothing asked yet)",
))
# {"action": "ASK", "question": "Who directed the film The Great Flamarion?", "rationale": "Identify the director to later find their spouse"}

With vLLM, serve it and send the filled template as the user message, with thinking off:

bash
vllm serve dolev31/ProactiveInquirer-Qwen3-8B-Merged
# request body: {"messages": [{"role": "user", "content": "<the filled template>"}],
#                "chat_template_kwargs": {"enable_thinking": false}, "temperature": 0}

Limitations

The questioner's own limitations, from the paper: it has learned what to ask more readily than when to stop, the extra evidence it finds does not yet translate into better final answers, and its user-facing results come from a simulated customer, not from real people.
It is a component inside an agent, meant to be called with its prompt template. It is not a chat assistant, and it was trained and evaluated in English.
This is one training seed (seed 1), merged in float32 and stored in bfloat16. Its equality with the adapter was checked on the card's example, not on a benchmark.

Citation

bibtex
@article{levy2026asking,
  title   = {Asking for What Was Never Requested: Horizontal and Vertical Proactivity in Agents},
  author  = {Levy, Ido and Yehudai, Asaf and Shlomov, Segev and Adi, Asaf and Choshen, Leshem},
  journal = {arXiv preprint arXiv:2609.37236},
  url     = {https://arxiv.org/abs/2609.37236},
  year    = {2026}
}

License

Apache-2.0, like the base model Qwen3-8B.

Available on FriendliAI

Dedicated Endpoints

Run this model inference on single tenant GPU with unmatched speed and reliability at scale.

Learn more

Container

Run this model inference with full control and performance in your environment.

Learn more

Model Details

Model Provider

dolev31

Model Tree

Base

Qwen/Qwen3-8B

Fine-tuned

this model

Input Modalities

Text

Output Modalities