Models
Two tracks. No trained model yet.
Each model gets a card here once it exists. Until then this page shows the plan and what has been measured on the candidates.
Track B · fine-tune
A reviewer for apps built with AI tools
A LoRA adapter on a small open-weight model. The task is reviewing and repairing apps built on Supabase: missing or wrong row-level security policies, missing ownership checks, exposed keys. It has to work in English and in Arabic.
not trained yet
Table 2 · parent candidates
| Qwen3.5-4B | Gemma 4 E4B | |
|---|---|---|
| Licence | Apache-2.0 | Apache-2.0 |
| How it was loaded | 16-bit | 4-bit, with the embedding table and some layers in system RAM |
| Peak VRAM, first load | 8.72 GiB | 7.09 GiB |
| Peak VRAM, LoRA at sequence 512 | 8.81 GiB | 7.09 GiB |
| Training time, 100 steps | 159 s | 231 s |
| Levantine Arabic, one prompt | answered in Egyptian | Levantine, judged natural |
| Vulnerable trials repaired, untuned | 18 of 42 (43%) | not measured yet |
| Clean apps left working, untuned | 8 of 9 trials | not measured yet |
The first load used four prompts per model, which is a sanity check and does not decide the parent. The repair eval is a pilot: 17 tasks, 3 trials each. A third candidate, Jais 3, has not been tried.
What counts as a result
- An adapter has to beat the untuned parent’s 43% by about 30 points at this task count. Anything smaller is reported as no measurable difference.
- It also has to beat a stronger model with lint tools on the same tasks. That comparison has not been run.
- It must not break the clean apps.
Track A · from scratch
A 17M-parameter model, built whole
A decoder-only Transformer trained from random weights with our own tokenizer. The goal is to learn the full stack: tokenizer, pretraining, checkpointing, sampling. It will not be a usable product model.
not trained yet
| Layers | 8 blocks, width 384, 6 heads |
|---|---|
| Parameters | 17,301,504, counted by hand |
| Context | 512 tokens |
| Tokenizer | Byte-level BPE. Vocabulary size is still open. |
| Tokens per second on this card | not measured yet |