ScriptSwiftAI v12 (Dikshan)
QLoRA fine-tuned full-stack code generation model by Dikshan (India). Covers Python, Django, Flask, FastAPI, React, Next.js, TypeScript, SQL, Docker.
License
CC BY-NC 4.0 — Non-commercial use only. Must credit Dikshan / ScriptSwiftAI. https://creativecommons.org/licenses/by-nc/4.0/
Copyright © 2026 Dikshan. All rights reserved.
Model Details
| Property | Value |
|---|---|
| Base Model | Qwen2.5-Coder-7B-Instruct |
| Method | QLoRA (4-bit, r=64) |
| Hardware | NVIDIA RTX 5070 Ti 16GB |
| Target Steps | 40,000 |
| Dataset | ~450-500k samples (17 datasets) |
| Max Length | 1536 tokens |
| Developer | Dikshan (India) |
| Training Started | June 2026 |
Changelog / Training Timeline
Auto-updated by train_v12.py on every checkpoint save. Append-only — entries never overwritten. HuggingFace commit history provides timestamps.
🔵 Checkpoint — Step 13
Date: June 22, 2026 at 18:44 IST Checkpoint: checkpoint-13 Step: 13 / 40,000 (0.0% complete) Loss: 0.6164 Epoch: 0.000 Zone: 📚 Learning Status: Training ongoing — auto-uploaded via train_v12.py
🔵 Checkpoint — Step 25
Date: June 22, 2026 at 18:50 IST Checkpoint: checkpoint-25 Step: 25 / 40,000 (0.1% complete) Loss: 0.6057 Epoch: 0.001 Zone: 📚 Learning Status: Training ongoing — auto-uploaded via train_v12.py
🔵 Checkpoint — Step 36
Date: June 22, 2026 at 18:58 IST Checkpoint: checkpoint-36 Step: 36 / 40,000 (0.1% complete) Loss: 0.5508 Epoch: 0.001 Zone: 📚 Learning Status: Training ongoing — auto-uploaded via train_v12.py
🔵 Checkpoint — Step 42
Date: June 22, 2026 at 19:08 IST Checkpoint: checkpoint-42 Step: 42 / 40,000 (0.1% complete) Loss: 0.5525 Epoch: 0.001 Zone: 📚 Learning Status: Training ongoing — auto-uploaded via train_v12.py
🔵 Checkpoint — Step 538
Date: June 22, 2026 at 23:38 IST Checkpoint: checkpoint-538 Step: 538 / 40,000 (1.3% complete) Loss: 0.4087 Epoch: 0.015 Zone: 📈 Good Status: Training ongoing — auto-uploaded via train_v12.py
🔵 Checkpoint — Step 1,000
Date: June 23, 2026 at 03:05 IST Checkpoint: checkpoint-1000 Step: 1,000 / 40,000 (2.5% complete) Loss: 0.4174 Epoch: 0.029 Zone: 📈 Good Status: Training ongoing — auto-uploaded via train_v12.py