You can deploy this model and start earning money today!

jeongseokoh/
llama3.1_8b_sft_SPEED-16-BoS_HotpotQA_lower_freeze

Name: llama3.1_8b_sft_SPEED-16-BoS_HotpotQA_lower_freeze
Availability: InStock

Views

No views yet

Recovered LoRA Adapter

Base model: jeongseokoh/llama3.1_8b_sft-llopa-k16-no_system
Source checkpoint: /data3/LLoPA/Capsule/model_from_downstream/llama3.1_8b_sft-llopa-k16-no_system-teacher_cache_overall.correct_only.all.q60000.r1-llopa-freeze-lower-k16-no_system/epoch_0
Recovery note: LoRA was trained only on upper layers 16-31 (train_upper_only=16) and resized tokenizer embeddings are included.