Specialized Optimized

AXL-Vision-v2

UI vision. 4.1M params. Screen-to-code prototype. Trained on synthetic UI screenshots.

4M
Parameters
---
Perplexity
10 min
Training
---
GGUF
PropertyValue
ArchitectureMulti-Scale Transformer
d_model?
Attention Heads?
Layers per Scale?
Context Window256 bytes
Downsample Factors[1, 2, 4]
Vocab Size258 (byte-level)
OptimizerSGD
Trained for 20 min on 2000 synthetic UI screenshots. Element classification (buttons, textboxes, navbars).
MetricValue
Final Loss2.7868
Perplexity---
Training Steps1418
Training Time10 min

Usage

ollama create axl-vision-v2 -f Modelfile
ollama run axl-vision-v2 "def fibonacci():"
Screen-to-code. Trained on synthetic UI elements.
FileSizeFormat
F16 GGUF---Full precision
Q4_K_M GGUF---4-bit quantized
GGUF files work with Ollama and llama.cpp. Q4_K_M is about 3x smaller than F16.
← All AXL Models