Sakalti/SJT-1.5B-Alpha
Full fine-tune vs Qwen/Qwen2.5-1.5B. Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Embeddings 1/1, Norms 57/57, MLP 84/84, Attention 196/196
7 derivatives checked
- Sakalti/SJT-2.4B: Partial fine-tune (fewer layers)
- mradermacher/SJT-1.5B-Alpha-GGUF: Same weights, converted
- Sakalti/Saka-1.5B: Full fine-tune
- Sakalti/mine-1.5B: Same weights, repacked
- Sakalti/SJT-RP-Lora-1.5B: LoRA-style
- Sakalti/SakaRP-1-1.5B: Same weights, repacked
- Sakalti/SJT-2.1B: Same weights, repacked