Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT/README.md

---
library_name: transformers
pipeline_tag: text-generation
license: apache-2.0
language:
  - en
base_model: Qwen/Qwen3-0.6B
datasets:
  - 0xZee/dataset-CoT-Advanced-Calculus-268
  - 0xZee/dataset-CoT-Modern-Physics-177
  - 0xZee/dataset-CoT-Theoretical-Mechanics-307
  - 0xZee/dataset-CoT-Linear-Algebra-667
  - 0xZee/dataset-CoT-Electromagnetism-580
  - 0xZee/dataset-CoT-Molecular-Biology-71
  - 0xZee/dataset-CoT-Physiology-114
  - 0xZee/dataset-CoT-Classical-Mechanics-343
  - 0xZee/dataset-CoT-Differential-Equations-636
  - 0xZee/dataset-CoT-Physics-2254
  - 0xZee/dataset-CoT-Engineering-574
  - 0xZee/dataset-CoT-mathematics
  - Alignment-Lab-AI/Lawyer-Instruct
tags:
  - causal-lm
  - text-generation
  - distillation
  - knowledge-distillation
  - sft
  - reasoning
  - chain-of-thought
  - mathematics
  - physics
  - engineering
  - legal
  - stem
  - convergentintel
  - edge
---

# Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT

A 0.6B parameter model built in two stages: knowledge distillation from a 30B Thinking teacher to establish a structured reasoning backbone, then supervised fine-tuning on legal instruction data. 50x compression. Under 500MB quantized. Runs on a phone.

The training order is the thesis: teach the model *how to reason* first (distillation from Thinking teacher), then teach it *what to reason about* (legal SFT). The Thinking teacher's extended deliberation traces transfer deeper reasoning structure than an Instruct teacher — critical when the student has only 0.6B parameters to work with.

> *"Structure beats scale, collaboration beats hierarchy, observation beats theory."*
> — Convergent Intelligence LLC: Research Division

## Training Pipeline

### Stage 1: Knowledge Distillation (STEM Reasoning Backbone)

Qwen3-0.6B distilled from [Qwen3-30B-A3B-Thinking-2507](https://huggingface.co/Qwen/Qwen3-30B-A3B-Thinking-2507) — a Mixture-of-Experts model with 30B total parameters, ~3B active per token, using the Thinking variant that generates extended internal reasoning traces.

**Why the Thinking teacher matters at 0.6B:** The Thinking variant produces higher-entropy softmax distributions than the Instruct variant — it considers more reasoning paths before committing. At distillation temperature T=2.0, the 0.6B student sees a richer landscape of alternative derivation strategies. With only 0.6B parameters, every bit of transferred structure counts. The Thinking teacher gives more.

**Data:** 6,122 STEM chain-of-thought samples across 12 domains:

| Domain | Samples |
|---|---|
| Physics | 2,254 |
| Linear Algebra | 667 |
| Differential Equations | 636 |
| Electromagnetism | 580 |
| Mathematics | 576 |
| Engineering | 574 |
| Classical Mechanics | 343 |
| Theoretical Mechanics | 307 |
| Advanced Calculus | 268 |
| Modern Physics | 177 |
| Physiology | 114 |
| Molecular Biology | 71 |

All from [0xZee](https://huggingface.co/0xZee). Shuffled seed 42, split 95/5 train/eval.

**Loss function:**

1. **Proof-Weighted Cross-Entropy (55%)** — 2.5x weight on derivation tokens, decaying to 1.5x. Forces the student to allocate its limited capacity to reasoning steps, not answer formatting.
2. **Knowledge Distillation KL Divergence (45%)** — T=2.0, scaled by T². Transfers the Thinking teacher's full deliberation landscape.

**Training format:**

```
Solve the following problem carefully and show a rigorous derivation.

Problem:
{question}

Proof:
{CoT}

Final Answer:
{response}
```

**Stage 1 hyperparameters:**

| Parameter | Value |
|---|---|
| Epochs | 1 |
| Training samples | 5,815 |
| Effective batch size | 8 |
| Learning rate | 1.5e-5 → 1e-6 (cosine) |
| Temperature | 2.0 |
| Proof weight | 2.5 → 1.5 |
| Precision | bf16 |

---

### Stage 2: Supervised Fine-Tuning (Legal Domain)

The distilled model was fine-tuned on [Alignment-Lab-AI/Lawyer-Instruct](https://huggingface.co/datasets/Alignment-Lab-AI/Lawyer-Instruct) using TRL's SFTTrainer.

**Why legal on top of STEM:** Legal reasoning is structurally isomorphic to mathematical reasoning — premise identification, logical chaining, exception handling, structured argumentation toward a conclusion. A model that learned rigorous derivation transfers that structure to legal analysis rather than learning legal templates from scratch.

**Training format:**

```
### Instruction:
{instruction}

### Response:
{output}
```

**Stage 2 hyperparameters:**

| Parameter | Value |
|---|---|
| Epochs | 1 |
| Effective batch size | 8 |
| Learning rate | 5e-6 (lower than Stage 1 to preserve backbone) |
| Gradient checkpointing | Enabled |
| Precision | bf16 |

## Model Details

| Attribute | Value |
|---|---|
| **Architecture** | Qwen3 (causal LM, RoPE, GQA) |
| **Parameters** | 0.6B |
| **Base model** | [Qwen/Qwen3-0.6B](https://huggingface.co/Qwen/Qwen3-0.6B) |
| **Teacher model** | [Qwen/Qwen3-30B-A3B-Thinking-2507](https://huggingface.co/Qwen/Qwen3-30B-A3B-Thinking-2507) |
| **Compression ratio** | 50x |
| **Stage 1 data** | 6,122 STEM CoT samples (12 datasets) |
| **Stage 2 data** | [Alignment-Lab-AI/Lawyer-Instruct](https://huggingface.co/datasets/Alignment-Lab-AI/Lawyer-Instruct) |
| **Context length** | 1024 tokens (training) |
| **License** | Apache 2.0 |
| **Developer** | Reaperdoesntrun / [Convergent Intelligence LLC](https://convergentintel.com): Research Division |

## Usage

```python
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch

model_id = "reaperdoesntknow/Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16 if torch.cuda.is_available() else torch.float32,
    device_map="auto",
)

# Legal instruction-following
prompt = """### Instruction:
What is the difference between a felony and a misdemeanor?

### Response:
"""

# STEM derivation (Stage 1 format still works)
prompt_stem = """Solve the following problem carefully and show a rigorous derivation.

Problem:
Compute the determinant of the matrix [[1, 2], [3, 4]].

Proof:
"""

inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
with torch.no_grad():
    outputs = model.generate(**inputs, max_new_tokens=512, do_sample=False)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
```

### GGUF

Quantized versions at [reaperdoesntknow/Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT-GGUF](https://huggingface.co/reaperdoesntknow/Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT-GGUF).

## Prompt Formats

**STEM derivation (Stage 1):**

```
Solve the following problem carefully and show a rigorous derivation.

Problem:
[Your problem]

Proof:
```

**Instruction-following (Stage 2):**

```
### Instruction:
[Your question]

### Response:
```

## Intended Uses

**Good for:** Ultra-lightweight reasoning on mobile/edge/IoT, legal and STEM instruction-following, educational tutoring, embedded inference, component in multi-model pipelines, anywhere you need reasoning in under 500MB.

**Not for:** Formal proof verification, actual legal counsel, safety-critical analysis, complex multi-step proofs (>8 steps), or long-context tasks beyond 1024 tokens.

## Limitations

0.6B is a hard capacity constraint. The model trades depth for deployability. It will make reasoning errors that a larger model would not. Multi-step derivations beyond ~8 steps degrade. Legal reasoning covers general concepts but lacks the nuance of larger models. Performance is weakest on underrepresented domains (molecular biology, physiology). Always verify outputs.


## Mathematical Foundations: Discrepancy Calculus (DISC)

This model is part of a distillation chain built on Discrepancy Calculus — a measure-theoretic framework where the teacher's output distribution is decomposed via the Mesh Fundamental Identity into smooth (AC), jump, and Cantor components. The discrepancy operator $Df(x) = \lim_{\varepsilon \downarrow 0} \frac{1}{\varepsilon} \int_x^{x+\varepsilon} \frac{|f(t) - f(x)|}{|t - x|} dt$ quantifies local structural mismatch that standard KL divergence averages away.

Full theory: *"On the Formal Analysis of Discrepancy Calculus"* (Colca, 2026; Convergent Intelligence LLC: Research Division). Full methodology: [Structure Over Scale (DOI: 10.57967/hf/8165)](https://doi.org/10.57967/hf/8165).

## Related Models

| Model | Description |
|---|---|
| [Qwen3-0.6B-STEM-Proof-Distilled-Thinking](https://huggingface.co/reaperdoesntknow/Qwen3-0.6B-STEM-Proof-Distilled-Thinking) | Stage 1 only — pure STEM backbone |
| [Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT-GGUF](https://huggingface.co/reaperdoesntknow/Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT-GGUF) | This model quantized for edge deployment |
| [Qwen3-1.7B-STEM-Proof-Distilled](https://huggingface.co/reaperdoesntknow/Qwen3-1.7B-STEM-Proof-Distilled) | Larger 1.7B variant (Instruct teacher) |
| [Qwen3-1.7B-Distilled-30B-A3B-SFT](https://huggingface.co/reaperdoesntknow/Qwen3-1.7B-Distilled-30B-A3B-SFT) | Larger 1.7B variant + legal SFT |

## Citation

```bibtex
@misc{colca2026thinking06bsft,
  title={Two-Stage Reasoning Transfer at 0.6B: Thinking Teacher Distillation + Legal SFT},
  year={2026},
  publisher={HuggingFace},
  url={https://huggingface.co/reaperdoesntknow/Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT},
  note={Convergent Intelligence LLC: Research Division}
}
```

---

*Convergent Intelligence LLC: Research Division*
*"Where classical analysis fails to see, we begin."*

---

## Convergent Intelligence Portfolio

*Part of the [Qwen3 0.6B Distillation Series](https://huggingface.co/reaperdoesntknow) by [Convergent Intelligence LLC: Research Division](https://huggingface.co/reaperdoesntknow)*


#
## Mathematical Foundations: Discrepancy Calculus (DISC)

This model is part of a distillation chain built on Discrepancy Calculus — a measure-theoretic framework where the teacher's output distribution is decomposed via the Mesh Fundamental Identity into smooth (AC), jump, and Cantor components. The discrepancy operator $Df(x) = \lim_{\varepsilon \downarrow 0} \frac{1}{\varepsilon} \int_x^{x+\varepsilon} \frac{|f(t) - f(x)|}{|t - x|} dt$ quantifies local structural mismatch that standard KL divergence averages away.

Full theory: *"On the Formal Analysis of Discrepancy Calculus"* (Colca, 2026; Convergent Intelligence LLC: Research Division). Full methodology: [Structure Over Scale (DOI: 10.57967/hf/8165)](https://doi.org/10.57967/hf/8165).

## Related Models

| Model | Downloads | Format |
|-------|-----------|--------|
| [Qwen3-0.6B-Distilled-30B-A3B](https://huggingface.co/reaperdoesntknow/Qwen3-0.6B-Distilled-30B-A3B) | 36 | HF |
| [Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT-GGUF](https://huggingface.co/reaperdoesntknow/Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT-GGUF) | 203 | GGUF |

### Top Models from Our Lab

| Model | Downloads |
|-------|-----------|
| [Qwen3-1.7B-Thinking-Distil](https://huggingface.co/reaperdoesntknow/Qwen3-1.7B-Thinking-Distil) | 501 |
| [LFM2.5-1.2B-Distilled-SFT](https://huggingface.co/reaperdoesntknow/LFM2.5-1.2B-Distilled-SFT) | 342 |
| [Qwen3-1.7B-Coder-Distilled-SFT](https://huggingface.co/reaperdoesntknow/Qwen3-1.7B-Coder-Distilled-SFT) | 302 |
| [Qwen3-1.7B-Coder-Distilled-SFT-GGUF](https://huggingface.co/reaperdoesntknow/Qwen3-1.7B-Coder-Distilled-SFT-GGUF) | 194 |
| [Qwen3-1.7B-Distilled-30B-A3B-SFT-GGUF](https://huggingface.co/reaperdoesntknow/Qwen3-1.7B-Distilled-30B-A3B-SFT-GGUF) | 175 |

**Total Portfolio: 41 models | 2,781 total downloads**


*Last updated: 2026-03-28 12:56 UTC*

<!-- DISTILQWEN-SPOTLIGHT-START -->

## DistilQwen Collection

This model is part of the **[DistilQwen](https://huggingface.co/collections/reaperdoesntknow/distilqwen-69bf40ec669117e3f069ef1c)** proof-weighted distillation series.
Collection: **9 models** | **2,788 downloads**

### Teacher Variant Comparison

| Teacher | Student Size | Strength | Models |
|---------|-------------|----------|--------|
| Qwen3-30B-A3B (Instruct) | 1.7B | Instruction following, structured output, legal reasoning | 3 (833 DL) |
| Qwen3-30B-A3B (Thinking) | 0.6B | Extended deliberation, higher-entropy distributions, proof derivation | 3 (779 DL) **← this model** |
| Qwen3-30B-A3B (Coder) | 1.7B | Structured decomposition, STEM derivation, logical inference | 2 (825 DL) |

### Methodology

**The only BF16 collection in the portfolio.** While the broader Convergent Intelligence catalog (43 models, 12,000+ downloads) was trained on CPU at FP32 for $24 total compute, the DistilQwen series was trained on H100 at BF16 with a 30B-parameter teacher. Same methodology, premium hardware. This is what happens when you give the pipeline real compute.

All models use proof-weighted knowledge distillation: 55% cross-entropy with decaying proof weights (2.5× → 1.5×), 45% KL divergence at T=2.0. The proof weight amplifies loss on reasoning-critical tokens, forcing the student to allocate capacity to structural understanding rather than surface-level pattern matching.

Full methodology: [Structure Over Scale (DOI: 10.57967/hf/8165)](https://doi.org/10.57967/hf/8165)

### Related in this series

- [Qwen3-0.6B-Distilled-30B-A3B](https://huggingface.co/reaperdoesntknow/Qwen3-0.6B-Distilled-30B-A3B) (236 downloads)
- [Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT-GGUF](https://huggingface.co/reaperdoesntknow/Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT-GGUF) (316 downloads)

<!-- DISTILQWEN-SPOTLIGHT-END -->
<!-- cix-keeper-ts:2026-06-12T13:16:20Z -->
<!-- card-refresh: 2026-03-30 -->