CALLAII Edge Small Business v1 technical sheet
Language model · A compact CALLAII language model prepared to run on small businesses' own hardware.
CALLAII Edge Small Business v1
Released September 2026
CALLAII Edge Small Business is designed for businesses that want a compact CALLAII running on their own server. It automatically checks procedural requirements in legal questions and reads document images.
What improved in this version
- First release
- Procedural check in legal questions: performed in 67 out of 70 exam questions, none before training
- Procedural check points improved from 9 to 57 across 70 questions
- Text extraction and image description from document images were fully preserved after training
What it does
- Legal: petition and contract drafting, procedural check on legal questions
- Procedural check: competent and authorized court, conditions of action, and deadlines
- Call center: call summary, customer response, and quality evaluation
- Visual reading: understanding text and content in documents and screenshots
Limitations and responsible use
- Generated text is a draft; petitions, contracts, and official documents must be approved by an authorized person
- Must be used together with a legislation library for current laws, regulations, and rulings; the model's memory is not up to date
- Audio recordings are transcribed with CALLAII ASR; this model is not used for speech recognition
- Quality in languages other than Turkish has not been measured in as much detail as Turkish
Training and measurements
1Epoch
0,810 → 0,605Validation loss
In 67 out of 70 questionsProcedural check
| Training data | 45,847 samples |
|---|---|
| Validation set | 200 samples never seen during training |
| Steps | 2,861 · 1 epoch · 16 samples per step |
| Training context | 4,096 tokens |
| Validation loss | 0.810 → 0.605 |
| Exam | 70 procedural questions, same questions before and after training |
| Training date | September 2026 |
Training loss
Validation loss
Each point is the average of 5 consecutive records.
Show values as a table
| Training loss | |
|---|---|
| 0.01 | 2.0882 |
| 0.05 | 0.8227 |
| 0.10 | 0.7656 |
| 0.14 | 0.7346 |
| 0.18 | 0.6991 |
| 0.23 | 0.6836 |
| 0.27 | 0.6616 |
| 0.31 | 0.6667 |
| 0.36 | 0.6545 |
| 0.40 | 0.6291 |
| 0.45 | 0.6276 |
| 0.49 | 0.6401 |
| 0.53 | 0.6182 |
| 0.58 | 0.6186 |
| 0.62 | 0.6388 |
| 0.66 | 0.5980 |
| 0.71 | 0.6144 |
| 0.75 | 0.6025 |
| 0.80 | 0.6273 |
| 0.84 | 0.6025 |
| 0.88 | 0.6050 |
| 0.93 | 0.5961 |
| 0.97 | 0.6281 |
| 1.00 | 0.5878 |
| Validation loss | |
| 0.14 | 0.7797 |
| 0.21 | 0.7025 |
| 0.28 | 0.6809 |
| 0.35 | 0.6608 |
| 0.42 | 0.6528 |
| 0.49 | 0.6549 |
| 0.56 | 0.6427 |
| 0.63 | 0.6439 |
| 0.70 | 0.6169 |
| 0.77 | 0.6117 |
| 0.84 | 0.6079 |
| 0.91 | 0.6055 |
| 0.98 | 0.6049 |
| 1 | 0.6050 |
Before training
After training
70 procedural questions, same questions before and after training.
Show values as a table
| Before training | After training | |
|---|---|---|
| Court and jurisdiction | 0.00 | 95.70 |
| Procedural points | 12.90 | 81.40 |
Technical sheet
| Version | v1 |
|---|---|
| Release | September 2026 |
| Input | Text and image |
| Output | Text |
| Language | Turkish-focused; responds in the same language for other languages |
| Runs on | the business's own server |
Customer data is never used to train the models. This sheet was generated on 27.09.2026. See callmenta.com for the current version. Versions and improvements