Property	Value
Developed by	CubicLabs
Architecture	Multi-Scale Transformer
Parameters	26M
Optimizer	SGD
Attention	SDPA
Vocab Size	258 (byte-level)
Context Window	512 bytes
d_model	320
Attention Heads	5
Layers per Scale	4
Downsample Factors	[1, 2, 4]
License	Apache 2.0

Sources

Repository: GitHub
Organization: CubicLabs

Uses

Direct Use

Agentic coding with tool routing.

python

import torch
from multiscale_transformer.model.model import MultiScaleTransformer
from multiscale_transformer.training.tokenizer import ByteTokenizer
ckpt = torch.load("axl_coder_15m.pt", map_location="cpu")
model = MultiScaleTransformer(config)
model.load_state_dict(ckpt["model_state_dict"])
model.eval()
tokenizer = ByteTokenizer()
ids = torch.tensor([tokenizer.encode("def hello():")], dtype=torch.long)
with torch.no_grad():
    out = model.generate(ids, max_new_tokens=50, temperature=0.8)
print(tokenizer.decode(out[0].tolist()))

Out-of-Scope Use

Not for production code generation. Not for non-code NLP tasks. For integration with tools like Continue.dev, LlamaIndex, or LangChain, use the Python API server which provides OpenAI-compatible endpoints.

Bias, Risks, and Limitations

Byte-level perplexity is not comparable to BPE-level perplexity. Agentic model with tool routing. Max context 512 bytes. Note: GGUF files for Ollama use a simplified single-stack encoder. For full AXL quality, use the Python API server.

Recommendations

Use for prototyping and experimentation, not production code generation.
Byte-level perplexity (258 vocab) is not comparable to BPE-level perplexity (32K vocab).
For better results, use the Lion-optimized version if available.

Training Details

Training Data

Trained with SGD for 10 min. Tool router predicts 8 actions: generate, read, write, edit, run, search, think, done.

Preprocessing

Byte-level tokenization with vocabulary size 258 (256 bytes + BOS + EOS). No vocabulary training required.

Speeds, Sizes, Times

Metric	Value
Training Steps	?
Training Time	---
Final Loss	0.4331

Evaluation

Metrics

Perplexity on held-out Python code using byte-level tokenization.

Results

Metric	Value
Perplexity (byte-level)	1.54
Final Loss	0.4331
Training Steps	?
Training Time	---

Summary: Agentic model. Decides what action to take per coding task.

Environmental Impact

Property	Value
Hardware	AMD Ryzen 5 5600G
Hours Used	0.001
Carbon Emitted	0.0001 kg CO2
Cloud Provider	None (local CPU)

Technical Specifications

Model Architecture

Multi-Scale Transformer with three parallel encoder stacks at resolution scales 1x, 2x, and 4x. Cross-scale attention connects all scale pairs. Adaptive gating fusion. SwiGLU feed-forward. RoPE positional encoding.

Compute Infrastructure

Property	Value
Hardware	AMD Ryzen 5 5600G (6 cores, 12 threads)
RAM	16 GB
GPU	None (CPU-only)

Citation

bibtex

@misc{axl_2026,
  title={AXL: AXL-Coder-15M - Multi-Scale Transformer for CPU Code Generation},
  author={Cubic},
  year={2026},
  url={[https://huggingface.co/CubicLabs](https://huggingface.co/CubicLabs)}
}

How to Get Started

With Ollama

bash

ollama create axl-coder-15m -f Modelfile
ollama run axl-coder-15m "def fibonacci():"

With Python