--- license: apache-2.0 library_name: transformers pipeline_tag: text-generation language: - en tags: - kozu - reasoning - trace-inversion - synthetic-reasoning - chain-of-thought datasets: - open-thoughts/OpenThoughts3-1.2M ---

Kuiper R1

Specialist Class · 9.41B · Apache 2.0
Trace inversion Synthetic reasoning Safetensors Public release
Release signal Expand the trace.
Preserve the answer. Weights available

Overview

01

Kuiper-R1 reconstructs full, readable reasoning from a problem, its final answer, and compressed reasoning bubbles.

Each result follows a three-tag contract: <synthetic_trace>, <verification_note>, and <final>. The final answer is preserved byte-for-byte, while the reconstructed trace is explicitly labeled synthetic.

Use Kuiper-R1 for reasoning-dataset augmentation, interpretability review, and research into reasoning compression and expansion. Do not present its synthetic traces as another model's hidden chain-of-thought.

Specifications

02
Architecture
TypeQwen3.5 conditional generation
Params9.41B
Context262,144 tokens
ArtifactBF16 safetensors
Training
MethodQLoRA SFT → ORPO → merge
SFT rows3,000
Preference87 scored pairs
DataOpenThoughts3-1.2M

Usage

03

Serve with vLLM. Disable the default thinking pass so the requested three-tag contract can use the full output budget.

Recommended Samplers
Temperature0.4
Repetition1.1
ThinkingDisabled
Model len8,192 served

Start the private checkpoint:

vllm serve Michael-Kozu/Kuiper-R1 \
  --served-model-name kuiper-r1 \
  --max-model-len 8192 \
  --gpu-memory-utilization 0.85 \
  --skip-mm-profiling \
  --trust-remote-code

Send an OpenAI-compatible request:

{
  "model": "kuiper-r1",
  "messages": [
    {
      "role": "system",
      "content": "You are Kuiper, a reasoning-trace inversion model."
    },
    {
      "role": "user",
      "content": "Problem:\n[problem text]\n\nAnswer:\n[exact final answer]\n\nReasoning Bubbles:\n- [bubble one]\n- [bubble two]\n\nReconstruct the full reasoning trace."
    }
  ],
  "temperature": 0.4,
  "repetition_penalty": 1.1,
  "chat_template_kwargs": {"enable_thinking": false}
}

Training Details

04

Training uses domain-balanced science, math, and code rows from OpenThoughts3-1.2M. Reasoning traces are compressed into extractive bubbles deterministically, without an additional model call.

3,000SFT examples
1SFT epoch
87ORPO pairs
  • Training split: 1,531 science, 1,001 math, and 468 code examples.
  • Compression is extractive at a target ratio of 0.28 with 3–24 bubbles.
  • The held-out DeepSeek-V4 benchmark lineage is disjoint from the QwQ-32B training lineage.

Benchmarks

05
Kuiper-R1 release · held-out n=200
Answer exact
0.820
Format valid
0.880
Expansion
0.769
Overall
0.504
MetricBaseKuiper-R1
Final answer preserved · exact0.1550.820
Format valid · 3-tag contract0.9800.880
Expansion quality0.7910.769
Semantic consistency0.7080.535
No invented tool outputs1.0001.000
Overall0.3290.504

Limitations & License

06
  • Bubble informativeness bounds trace quality; terse inputs provide little evidence to expand.
  • The verification note is a construction-time consistency statement, not an external verifier.
  • The release improves exact answer preservation but trails the untouched base on raw format validity and semantic consistency.

The checkpoint is released under the Apache License 2.0.

Kozu AI Turning the laws of reality into unparalleled creation.