Files
layerskip-llama3.2-1B/original/consolidate_params.json

7 lines
544 B
JSON
Raw Permalink Normal View History

{
"dtype": "bf16",
"model_parallel_size": 1,
"on_gpu": true,
"src": "/fsx-atom/melhoushi/xldumps/continual_llama3.2_1B_gpus_64_shared_emb_higher_lr_ee_0.1_ld_0.1_128_gpus/continual_llama3.2_1B_gpus_64_shared_emb_higher_lr_ee_0.1_ld_0.1_128_gpus_run000/checkpoints/checkpoint_0100000/",
"tgt": "/fsx-atom/melhoushi/xldumps/continual_llama3.2_1B_gpus_64_shared_emb_higher_lr_ee_0.1_ld_0.1_128_gpus/continual_llama3.2_1B_gpus_64_shared_emb_higher_lr_ee_0.1_ld_0.1_128_gpus_run000/checkpoints/checkpoint_0100000_consolidated/"
}