---
license: apache-2.0
language:
- en
base_model: tjarvis91/qovaryx-50m-scratch-base
base_model_relation: finetune
library_name: pytorch
pipeline_tag: text-generation
tags:
- text-generation
- qovaryx
- compact-cognition
- local-ai
- code
- python
- code-generation
- sovereign-base
---

# Q-Coder-50M-Sovereign — Python code one-liners + small function skeletons

> **Python task in. Smallest correct expression out. No fences. No prose.**

## What this model does, in one sentence

Given a short natural-language Python task, returns the smallest correct Python expression or function that solves it. Trained on aggregate ops (sum/min/max/len/avg over named lists), string ops (reverse/upper/lower/title/palindrome), list comprehensions (even/odd/positive/squares/doubles), dict .get(default), small function definitions, try/except wrappers, class skeletons, and basic file I/O. Designed for fast structured code emission, not free-form programming.

## Honest performance

- **Task:** compact Python code generation
- **Metric:** `exact_match` (string-equal after strip + lowercase)
- **Holdout:** n=53 rows, never seen in training, scored row-by-row
- **Score:** **100.0%** mean
- **Bootstrap CI 95% lower bound:** 1.000
- **Gate threshold:** 0.90
- **Verdict:** PASS at point estimate AND at bootstrap CI lower bound

## What it's used for — real workflows

- **Inline boilerplate emitter for IDEs** — Wire Q-Coder into an editor extension; ask 'reverse a string', get back 's[::-1]'. The point isn't autocomplete — it's deterministic emission of the small idioms you'd otherwise type.
- **Notebook one-liner generator** — Quick prompts in a notebook: 'sum of values', 'filter positives from nums'. Q-Coder returns the one-liner ready to paste.
- **Snippet expansion in chat ops** — Slack/Discord bot: 'qcoder def add' returns the canonical add(a,b). Cheap, deterministic, on-prem.
- **Test-skeleton generation** — Ask for a class skeleton with __init__ + a method; get a clean Python class body to fill in.

## What problem this actually solves

Coding LMs are usually trained to generate sprawling, fence-wrapped, explanation-heavy code. Q-Coder is the opposite: tight expressions for tight tasks, no markdown, no chat, no fences. Use it as the focused emission step inside a bigger workflow.

## Integration paths

- **Editor snippet engine** — Wire to a VS Code / JetBrains extension as a code-snippet generator.
- **Q-Office-Suite runtime** — POST /run/q-coder with the natural-language task.
- **Pair with Q-SheetExtract** — Q-SheetExtract gives you fields; Q-Coder gives you the expression that computes whatever aggregate you actually wanted.

## Example

Input:
```
Define a function `square` that returns x squared.
```

Output:
```
def square(x):
    return x * x
```

## What this is NOT

- **Not a general-purpose chatbot.** This head does one job and does it consistently. Free-text generation outside the trained task surface will degrade.
- **Not a replacement for a verifier.** This is one component in the Qovaryx cluster-shell architecture. The decision-acceptance discipline lives in the wrapper, not in the head.
- **Not reproducible from this card.** Weights and audit are public; the crystal corpus, eval gate constants, and training hyperparameters are not.

## Proprietary Qovaryx technology — built on our own scratch base

This is a **53.5M-parameter sovereign specialist** in the Qovaryx Compact Specialist Suite. It is full-fine-tuned from [`tjarvis91/qovaryx-50m-scratch-base`](https://huggingface.co/tjarvis91/qovaryx-50m-scratch-base) — **our own scratch-trained base, not a borrowed foundation model**.

- **Base:** Qovaryx 50M scratch base. Pretrained from random initialization on 491.5M tokens. **Not SmolLM2. Not Qwen. Not Llama. Not Mistral. Not Phi.** No HuggingFace foundation. No closed-source weights. Every parameter traces back to a Qovaryx training run on Qovaryx hardware.
- **Tokenizer:** Qovaryx `english_v1` BPE (vocab 32000), built in-house against our own pretraining corpus.
- **Architecture:** Qovaryx FinanceDecoder — 12 decoder blocks, GQA, RoPE, SwiGLU FFN, RMSNorm, MTP heads, decision head.
- **Recipe:** Qovaryx crystallization discipline — train the law before replaying the noise.
- **Runs on CPU.** No GPU required at inference.

## Architecture (Qovaryx proprietary)

- 53.5M parameters
- 12 decoder blocks, d_model=512, n_head=8, GQA n_kv_head=2
- SwiGLU FFN, RoPE positional, RMSNorm
- Multi-token prediction (MTP) auxiliary heads
- Decision head for routed-decision tasks
- Tokenizer: Qovaryx `english_v1` BPE, vocab 32000 (in-house build)
- Pretrained from `qovaryx-50m-scratch-base` step 60000 — 491.5M tokens
- Full fine-tune (no LoRA, no QLoRA, no adapter): every parameter was updated on the Qovaryx crystal corpus for this specialist

## How to load it (Python)

```python
import torch
from tokenizers import Tokenizer
from bleeding_edge.model.decoder import FinanceDecoder, DecoderConfig

tok = Tokenizer.from_file("tokenizer.json")
ckpt = torch.load("pytorch_model.pt", map_location="cpu", weights_only=False)
cfg = DecoderConfig(**{k: v for k, v in ckpt["model_cfg"].items() if k in DecoderConfig.__dataclass_fields__})
cfg.vocab_size = tok.get_vocab_size()
model = FinanceDecoder(cfg).eval()
state = {k.removeprefix("_orig_mod."): v for k, v in ckpt["model_state"].items()}
model.load_state_dict(state, strict=False)

prompt = "Define a function `square` that returns x squared."
ids = tok.encode(prompt).ids
cur = torch.tensor([ids], dtype=torch.long)
with torch.no_grad():
    for _ in range(120):
        nxt = int(torch.argmax(model(cur, return_decision=False).logits[:, -1, :], dim=-1))
        if nxt == 0: break
        cur = torch.cat([cur, torch.tensor([[nxt]])], dim=1)
print(tok.decode(cur[0].tolist()[len(ids):]))
```

## License & posture

Apache 2.0 for the published weights, model card, and example code.

The Qovaryx scratch base build pipeline, the crystallization corpus, the eval gate constants, the cluster routing policy, and the protected runtime entrypoint are **Qovaryx proprietary technology** and are not included in this release. Same posture as every previous Qovaryx public release: ship the weights and the audit, not the recipe.

## Sibling specialists in the Qovaryx Q-Office-Suite

All nine specialists share the `qovaryx-50m-scratch-base` and the same audit discipline. Use one directly; use all nine through the cluster shell.

- [Q-Triage](https://huggingface.co/tjarvis91/Q-Triage-50M-Sovereign) — ticket routing
- [Q-DocCite](https://huggingface.co/tjarvis91/Q-DocCite-50M-Sovereign) — document citation
- [Q-Invoice](https://huggingface.co/tjarvis91/Q-Invoice-50M-Sovereign) — invoice extraction
- [Q-ToolCall](https://huggingface.co/tjarvis91/Q-ToolCall-50M-Sovereign) — agent tool-calls
- [Q-Meeting](https://huggingface.co/tjarvis91/Q-Meeting-50M-Sovereign) — meeting structuring
- [Q-FinCite](https://huggingface.co/tjarvis91/Q-FinCite-50M-Sovereign) — 10-K/10-Q citation
- [Q-CmdSafe](https://huggingface.co/tjarvis91/Q-CmdSafe-50M-Sovereign) — command safety triage
- [Q-SheetExtract](https://huggingface.co/tjarvis91/Q-SheetExtract-50M-Sovereign) — spreadsheet extraction
- [Q-Coder](https://huggingface.co/tjarvis91/Q-Coder-50M-Sovereign) — Python code skeletons

## Watermark

This release carries a SHA256 issue fingerprint inside `release.json` for tamper-detection and attribution.

## Community & support

- **Research devlog:** https://github.com/thron-j/qovaryx-ai-research
- **Discord (Qovaryx community):** https://discord.gg/PtuHZDv5ju
- **Ko-fi (we cover GPU bills):** https://ko-fi.com/tjarvis91
- **Qovaryx options decoder runtime:** https://huggingface.co/Qovaryx/qovaryx-options-decoder-full-community

If you find a failure mode this card doesn't cover, open a discussion on this repo or come to the Discord — that's how the next crystal corpus gets written.