File size: 447 Bytes
d8c9a83 fa3ace4 d8c9a83 fa3ace4 d8c9a83 fa3ace4 d8c9a83 fa3ace4 d8c9a83 fa3ace4 d8c9a83 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 | # payelb/HHRLHF_TinyLlama-1.1B_aligned_with_baseline_deberta_RM
Base model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
Alignment dataset: Anthropic/hh-rlhf
Reward model: payelb/HHRLHF_reward-model-deberta-v3-base_1k_fixed_baseline
Method: PPO alignment with LoRA adapters.
Notes:
- Reward normalization and clipping enabled
- KL control enabled
- pad_token_id/eos_token_id explicitly set
- DeBERTa RM loaded on a single device (no device_map='auto')
|