QLoRA Fine-tuning & RAG-based Chat Demo using Custom R1 Model
Finetune 4-bit (QLoRA) on ServiceNow-AI/R1-Distill-SFT subset (up to 5 min)
Finetune Status
Direct Generation (No Retrieval) using Custom R1
Prompt
Temperature
↺
0
1.5
Top-p
↺
0
1
Min New Tokens
↺
1
2500
Max New Tokens
↺
1
2500
Custom R1 Output
Generate with Custom R1
Compare Custom R1 vs Official R1
Compare
Custom R1 Output
Official R1 Output
Chat with Retrieval-Augmented Memory
RAG Chat
Textbox
Send