Language-model training and adaptation · MLD-0064
Assistant-only loss
Masking training loss to assistant-role spans in chat-formatted examples.
Original reference page · detailed expansion pending. Original definitions, examples, notes and citations are preserved below. New examples are illustrative, not benchmark results.
Original example
Train on multi-turn conversations while excluding user targets.
Original caveat
Correct role-boundary detection is essential.
Context and aliases
Level: Intermediate. Stages: Training · Adaptation · Evaluation.
Tags: DL · SLM · LLM · NLP.
Aliases: None recorded.
Related concepts
Original references
- Transformers glossary ↗ · Consulted
- PEFT: LoRA conceptual guide ↗ · Consulted
- TRL: supervised fine-tuning ↗ · Reference link
- TRL documentation ↗ · Consulted
- Original reference URL ↗