Highlights

• Faithful by design — answers only from the supplied context; achieves the best faithfulness (lowest memorization ratio) across all evaluated scales, including 32B models.
• Calibrated abstention — outputs Not enough information when the context does not support an answer.
• Structured, citable reasoning — every answer comes with a transparent trace (query analysis → source analysis → reasoning → status → answer) that cites sources by id.
• Compact — a small model that delivers chain-of-thought-level transparency at a fraction of full thinking-mode inference cost.

Model overview

OCC-RAG-1.7B is mid-trained from Qwen/Qwen3-1.7B-Base via supervised fine-tuning on a synthetic corpus of ~3.25M QA pairs (~2.78M single-hop, ~262k multi-hop single-context, ~165k multi-hop multi-context, and ~43k abstain examples), distilled from a larger teacher with citation-anchored reasoning traces. Multi-hop and multi-context subsets are oversampled to emphasize compositional reasoning. The prompt/response format is identical at training and inference time, so no train–test mismatch is introduced.

Evaluation

Evaluated across multi-hop reasoning (HotpotQA, MuSiQue, TAT-QA), faithfulness (ConFiQA), and refusal (MuSiQue-Un). In-Acc = the gold answer appears as a substring of the prediction; F1 = token-level overlap between prediction and gold answer; M_R = memorization ratio (lower = more faithful); R-Acc = refusal accuracy.

OCC-RAG-1.7B closes the gap with Qwen3-4B (thinking) on multi-hop reasoning while attaining the best faithfulness (highest ConFiQA In-Acc, lowest M_R) across all evaluated scales, and refusal accuracy on par with 8B+ models. Mid-training reduces the memorization ratio from 12.7 (8.3 in thinking mode) for Qwen3-1.7B down to 5.0.

Input / output format

OCC-RAG uses a structured prompt format with special tokens. The question is wrapped in <|query_start|> … <|query_end|> and each source in <|source_start|><|source_id|>N … <|source_end|>.

The response is split into five sections, each delimited by special tokens:

Quickstart (Transformers)

The chat template accepts a documents= kwarg and emits the structural tokens for the query and sources automatically — pass the user message as plain text and the sources as a list of dicts.

python
Copy code
import re
from transformers import AutoModelForCausalLM, AutoTokenizer

MODEL = "occ-ai/OCC-RAG-1.7B"

tokenizer = AutoTokenizer.from_pretrained(MODEL)
model = AutoModelForCausalLM.from_pretrained(MODEL, torch_dtype="auto", device_map="auto")

question = "Which country is the inventor of the telephone, Alexander Graham Bell, buried in?"
documents = [
    {"text": "Alexander Graham Bell was a Scottish-born inventor best known for patenting the first practical telephone."},
    {"text": "Bell died on August 2, 1922, at his estate Beinn Bhreagh, near Baddeck, Nova Scotia, and was buried there."},
    {"text": "Nova Scotia is a province on the east coast of Canada."},
]

text = tokenizer.apply_chat_template(
    [{"role": "user", "content": question}],
    documents=documents,
    tokenize=False,
    add_generation_prompt=True,
    enable_thinking=False,
)

# Alternative: assemble the structural tokens yourself.
#
# query_start, query_end = "<|query_start|>", "<|query_end|>"
# source_start, source_end, source_id = "<|source_start|>", "<|source_end|>", "<|source_id|>"
#
# def build_user_content(question, sources):
#     content = f"{query_start}{question}{query_end}\n"
#     for i, s in enumerate(sources, start=1):
#         content += f"{source_start}{source_id}{i} {s}{source_end}\n"
#     return content
#
# messages = [{"role": "user", "content": build_user_content(question, [d["text"] for d in documents])}]
# text = tokenizer.apply_chat_template(
#     messages, tokenize=False, add_generation_prompt=True, enable_thinking=False
# )

inputs = tokenizer([text], return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=2048)
response = tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=False)
print(response)

m = re.findall(r"<\|answer_start\|>(.*?)(?:<\|answer_end\|>|\Z)", response, re.DOTALL)
print("Answer:", m[-1].strip() if m else "")   # -> Canada

[!NOTE] We recommend greedy decoding (do_sample=False), which is the training/evaluation default and is baked into generation_config.json. Qwen3's default sampling parameters (best practices) also work fine.

Deployment

OCC-RAG-1.7B is a standard Qwen3 causal LM and is compatible with vLLM, SGLang, and other Transformers-based serving stacks. With only 1.7B parameters, it can be readily deployed in constrained infrastructure, including desktop systems running on CPU RAM. When serving, keep skip_special_tokens=False if you need to parse the structural tokens out of the raw output.

When using an OpenAI-compatible server (vLLM ≥0.6, SGLang ≥0.4.7), the documents= kwarg is reachable from the client via chat_template_kwargs:

python
Copy code
client.chat.completions.create(
    model="occ-ai/OCC-RAG-1.7B",
    messages=[{"role": "user", "content": question}],
    extra_body={"chat_template_kwargs": {"documents": documents}},
)

Limitations

• Context-grounded only. The model is trained to answer from the supplied sources and to ignore parametric knowledge. It is not a general-purpose chat or knowledge model.
• Reasoning depth. Training and evaluation are capped at three-hop reasoning; longer chains are out of distribution.

Citation

If you find our work helpful, feel free to give us a cite.

bibtex
Copy code
@misc{savkin2026occragoptimalcognitivecore,
  title         = {OCC-RAG: Optimal Cognitive Core for Faithful Question Answering},
  author        = {Maksim Savkin and Mikhail Goncharov and Alexander Gambashidze and Alla Chepurova and Dmitrii Tarasov and Nikita Andriianov and Daria Pugacheva and Vasily Konovalov and Andrey Galichin and Ivan Oseledets},
  year          = {2026},
  eprint        = {2606.00683},
  archivePrefix = {arXiv},
  primaryClass  = {cs.CL},
  url           = {https://arxiv.org/abs/2606.00683}
}