Lion Optimized

AXL-Comment-5M

Code commenting. 7.2M params. PPL 1.16. Context 2048 bytes.

7M
Parameters
1.16
Perplexity
10 min
Training
14 MB
GGUF
PropertyValue
ArchitectureMulti-Scale Transformer
d_model?
Attention Heads?
Layers per Scale?
Context Window2048 bytes
Downsample Factors[1, 2, 4]
Vocab Size258 (byte-level)
OptimizerLion
Retrained with Lion on 20MB commenting pairs. 263 steps in 10 min.
MetricValue
Final Loss0.1476
Perplexity1.16
Training Steps263
Training Time10 min

Usage

ollama create axl-comment-5m -f Modelfile
ollama run axl-comment-5m "def fibonacci():"
Adds inline comments to explain code logic.
FileSizeFormat
F16 GGUF14 MBFull precision
Q4_K_M GGUF14 MB4-bit quantized
GGUF files work with Ollama and llama.cpp. Q4_K_M is about 3x smaller than F16.
← All AXL Models