Upload folder using huggingface_hub
Browse files- client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_0625_seed1/0_trainer_state.json +3776 -0
- client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_0625_seed1/server_model_round0.pth +3 -0
- client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_0625_seed1/server_model_round1.pth +3 -0
- client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_0625_seed1/server_model_round2.pth +3 -0
- client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_0625_seed1/server_model_round3.pth +3 -0
client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_0625_seed1/0_trainer_state.json
ADDED
@@ -0,0 +1,3776 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_global_step": null,
|
3 |
+
"best_metric": null,
|
4 |
+
"best_model_checkpoint": null,
|
5 |
+
"epoch": 1.0,
|
6 |
+
"eval_steps": 500,
|
7 |
+
"global_step": 1249,
|
8 |
+
"is_hyper_param_search": false,
|
9 |
+
"is_local_process_zero": true,
|
10 |
+
"is_world_process_zero": true,
|
11 |
+
"log_history": [
|
12 |
+
{
|
13 |
+
"epoch": 0.0016012810248198558,
|
14 |
+
"learning_rate": 2.4341906163790364e-06,
|
15 |
+
"loss": 0.1514,
|
16 |
+
"step": 2
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.0032025620496397116,
|
20 |
+
"learning_rate": 2.4708617956148052e-06,
|
21 |
+
"loss": 0.1172,
|
22 |
+
"step": 4
|
23 |
+
},
|
24 |
+
{
|
25 |
+
"epoch": 0.004803843074459567,
|
26 |
+
"learning_rate": 2.507768247396697e-06,
|
27 |
+
"loss": 0.139,
|
28 |
+
"step": 6
|
29 |
+
},
|
30 |
+
{
|
31 |
+
"epoch": 0.006405124099279423,
|
32 |
+
"learning_rate": 2.5449088184619065e-06,
|
33 |
+
"loss": 0.0545,
|
34 |
+
"step": 8
|
35 |
+
},
|
36 |
+
{
|
37 |
+
"epoch": 0.008006405124099279,
|
38 |
+
"learning_rate": 2.5822823482318517e-06,
|
39 |
+
"loss": 0.6831,
|
40 |
+
"step": 10
|
41 |
+
},
|
42 |
+
{
|
43 |
+
"epoch": 0.009607686148919135,
|
44 |
+
"learning_rate": 2.6198876688483453e-06,
|
45 |
+
"loss": 0.1763,
|
46 |
+
"step": 12
|
47 |
+
},
|
48 |
+
{
|
49 |
+
"epoch": 0.01120896717373899,
|
50 |
+
"learning_rate": 2.6577236052101764e-06,
|
51 |
+
"loss": 0.1585,
|
52 |
+
"step": 14
|
53 |
+
},
|
54 |
+
{
|
55 |
+
"epoch": 0.012810248198558846,
|
56 |
+
"learning_rate": 2.6957889750097866e-06,
|
57 |
+
"loss": 0.0462,
|
58 |
+
"step": 16
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.014411529223378704,
|
62 |
+
"learning_rate": 2.7340825887701848e-06,
|
63 |
+
"loss": 0.2059,
|
64 |
+
"step": 18
|
65 |
+
},
|
66 |
+
{
|
67 |
+
"epoch": 0.016012810248198558,
|
68 |
+
"learning_rate": 2.772603249882202e-06,
|
69 |
+
"loss": 0.0941,
|
70 |
+
"step": 20
|
71 |
+
},
|
72 |
+
{
|
73 |
+
"epoch": 0.017614091273018415,
|
74 |
+
"learning_rate": 2.81134975464178e-06,
|
75 |
+
"loss": 0.1517,
|
76 |
+
"step": 22
|
77 |
+
},
|
78 |
+
{
|
79 |
+
"epoch": 0.01921537229783827,
|
80 |
+
"learning_rate": 2.850320892287688e-06,
|
81 |
+
"loss": 0.1519,
|
82 |
+
"step": 24
|
83 |
+
},
|
84 |
+
{
|
85 |
+
"epoch": 0.020816653322658127,
|
86 |
+
"learning_rate": 2.889515445039256e-06,
|
87 |
+
"loss": 0.1036,
|
88 |
+
"step": 26
|
89 |
+
},
|
90 |
+
{
|
91 |
+
"epoch": 0.02241793434747798,
|
92 |
+
"learning_rate": 2.928932188134529e-06,
|
93 |
+
"loss": 0.1438,
|
94 |
+
"step": 28
|
95 |
+
},
|
96 |
+
{
|
97 |
+
"epoch": 0.02401921537229784,
|
98 |
+
"learning_rate": 2.9685698898684355e-06,
|
99 |
+
"loss": 0.4228,
|
100 |
+
"step": 30
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.025620496397117692,
|
104 |
+
"learning_rate": 3.00842731163137e-06,
|
105 |
+
"loss": 0.1722,
|
106 |
+
"step": 32
|
107 |
+
},
|
108 |
+
{
|
109 |
+
"epoch": 0.02722177742193755,
|
110 |
+
"learning_rate": 3.048503207947854e-06,
|
111 |
+
"loss": 0.0769,
|
112 |
+
"step": 34
|
113 |
+
},
|
114 |
+
{
|
115 |
+
"epoch": 0.028823058446757407,
|
116 |
+
"learning_rate": 3.0887963265154187e-06,
|
117 |
+
"loss": 0.4029,
|
118 |
+
"step": 36
|
119 |
+
},
|
120 |
+
{
|
121 |
+
"epoch": 0.03042433947157726,
|
122 |
+
"learning_rate": 3.129305408243829e-06,
|
123 |
+
"loss": 0.1135,
|
124 |
+
"step": 38
|
125 |
+
},
|
126 |
+
{
|
127 |
+
"epoch": 0.032025620496397116,
|
128 |
+
"learning_rate": 3.17002918729432e-06,
|
129 |
+
"loss": 0.185,
|
130 |
+
"step": 40
|
131 |
+
},
|
132 |
+
{
|
133 |
+
"epoch": 0.03362690152121697,
|
134 |
+
"learning_rate": 3.2109663911192622e-06,
|
135 |
+
"loss": 0.153,
|
136 |
+
"step": 42
|
137 |
+
},
|
138 |
+
{
|
139 |
+
"epoch": 0.03522818254603683,
|
140 |
+
"learning_rate": 3.2521157405018146e-06,
|
141 |
+
"loss": 0.1267,
|
142 |
+
"step": 44
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.03682946357085669,
|
146 |
+
"learning_rate": 3.293475949595998e-06,
|
147 |
+
"loss": 0.1423,
|
148 |
+
"step": 46
|
149 |
+
},
|
150 |
+
{
|
151 |
+
"epoch": 0.03843074459567654,
|
152 |
+
"learning_rate": 3.335045725966829e-06,
|
153 |
+
"loss": 0.2389,
|
154 |
+
"step": 48
|
155 |
+
},
|
156 |
+
{
|
157 |
+
"epoch": 0.040032025620496396,
|
158 |
+
"learning_rate": 3.3768237706306716e-06,
|
159 |
+
"loss": 0.2257,
|
160 |
+
"step": 50
|
161 |
+
},
|
162 |
+
{
|
163 |
+
"epoch": 0.041633306645316254,
|
164 |
+
"learning_rate": 3.418808778095917e-06,
|
165 |
+
"loss": 0.1678,
|
166 |
+
"step": 52
|
167 |
+
},
|
168 |
+
{
|
169 |
+
"epoch": 0.04323458767013611,
|
170 |
+
"learning_rate": 3.460999436403676e-06,
|
171 |
+
"loss": 0.1747,
|
172 |
+
"step": 54
|
173 |
+
},
|
174 |
+
{
|
175 |
+
"epoch": 0.04483586869495596,
|
176 |
+
"learning_rate": 3.5033944271688624e-06,
|
177 |
+
"loss": 0.1452,
|
178 |
+
"step": 56
|
179 |
+
},
|
180 |
+
{
|
181 |
+
"epoch": 0.04643714971977582,
|
182 |
+
"learning_rate": 3.5459924256213596e-06,
|
183 |
+
"loss": 0.0932,
|
184 |
+
"step": 58
|
185 |
+
},
|
186 |
+
{
|
187 |
+
"epoch": 0.04803843074459568,
|
188 |
+
"learning_rate": 3.588792100647368e-06,
|
189 |
+
"loss": 0.1592,
|
190 |
+
"step": 60
|
191 |
+
},
|
192 |
+
{
|
193 |
+
"epoch": 0.049639711769415534,
|
194 |
+
"learning_rate": 3.6317921148310965e-06,
|
195 |
+
"loss": 0.1281,
|
196 |
+
"step": 62
|
197 |
+
},
|
198 |
+
{
|
199 |
+
"epoch": 0.051240992794235385,
|
200 |
+
"learning_rate": 3.674991124496452e-06,
|
201 |
+
"loss": 0.2977,
|
202 |
+
"step": 64
|
203 |
+
},
|
204 |
+
{
|
205 |
+
"epoch": 0.05284227381905524,
|
206 |
+
"learning_rate": 3.7183877797491143e-06,
|
207 |
+
"loss": 0.5645,
|
208 |
+
"step": 66
|
209 |
+
},
|
210 |
+
{
|
211 |
+
"epoch": 0.0544435548438751,
|
212 |
+
"learning_rate": 3.7619807245186824e-06,
|
213 |
+
"loss": 0.1164,
|
214 |
+
"step": 68
|
215 |
+
},
|
216 |
+
{
|
217 |
+
"epoch": 0.05604483586869496,
|
218 |
+
"learning_rate": 3.8057685966010025e-06,
|
219 |
+
"loss": 0.3493,
|
220 |
+
"step": 70
|
221 |
+
},
|
222 |
+
{
|
223 |
+
"epoch": 0.057646116893514815,
|
224 |
+
"learning_rate": 3.849750027700842e-06,
|
225 |
+
"loss": 0.1356,
|
226 |
+
"step": 72
|
227 |
+
},
|
228 |
+
{
|
229 |
+
"epoch": 0.059247397918334666,
|
230 |
+
"learning_rate": 3.8939236434745184e-06,
|
231 |
+
"loss": 0.0712,
|
232 |
+
"step": 74
|
233 |
+
},
|
234 |
+
{
|
235 |
+
"epoch": 0.06084867894315452,
|
236 |
+
"learning_rate": 3.938288063572962e-06,
|
237 |
+
"loss": 0.4313,
|
238 |
+
"step": 76
|
239 |
+
},
|
240 |
+
{
|
241 |
+
"epoch": 0.06244995996797438,
|
242 |
+
"learning_rate": 3.982841901684792e-06,
|
243 |
+
"loss": 0.2522,
|
244 |
+
"step": 78
|
245 |
+
},
|
246 |
+
{
|
247 |
+
"epoch": 0.06405124099279423,
|
248 |
+
"learning_rate": 4.027583765579601e-06,
|
249 |
+
"loss": 0.2175,
|
250 |
+
"step": 80
|
251 |
+
},
|
252 |
+
{
|
253 |
+
"epoch": 0.0656525220176141,
|
254 |
+
"learning_rate": 4.072512257151546e-06,
|
255 |
+
"loss": 0.0657,
|
256 |
+
"step": 82
|
257 |
+
},
|
258 |
+
{
|
259 |
+
"epoch": 0.06725380304243395,
|
260 |
+
"learning_rate": 4.117625972462988e-06,
|
261 |
+
"loss": 0.2479,
|
262 |
+
"step": 84
|
263 |
+
},
|
264 |
+
{
|
265 |
+
"epoch": 0.0688550840672538,
|
266 |
+
"learning_rate": 4.1629235017883285e-06,
|
267 |
+
"loss": 0.1743,
|
268 |
+
"step": 86
|
269 |
+
},
|
270 |
+
{
|
271 |
+
"epoch": 0.07045636509207366,
|
272 |
+
"learning_rate": 4.208403429658151e-06,
|
273 |
+
"loss": 0.6238,
|
274 |
+
"step": 88
|
275 |
+
},
|
276 |
+
{
|
277 |
+
"epoch": 0.07205764611689351,
|
278 |
+
"learning_rate": 4.254064334903347e-06,
|
279 |
+
"loss": 0.0513,
|
280 |
+
"step": 90
|
281 |
+
},
|
282 |
+
{
|
283 |
+
"epoch": 0.07365892714171338,
|
284 |
+
"learning_rate": 4.299904790699619e-06,
|
285 |
+
"loss": 0.0817,
|
286 |
+
"step": 92
|
287 |
+
},
|
288 |
+
{
|
289 |
+
"epoch": 0.07526020816653323,
|
290 |
+
"learning_rate": 4.345923364612024e-06,
|
291 |
+
"loss": 0.0745,
|
292 |
+
"step": 94
|
293 |
+
},
|
294 |
+
{
|
295 |
+
"epoch": 0.07686148919135308,
|
296 |
+
"learning_rate": 4.392118618639698e-06,
|
297 |
+
"loss": 0.2478,
|
298 |
+
"step": 96
|
299 |
+
},
|
300 |
+
{
|
301 |
+
"epoch": 0.07846277021617294,
|
302 |
+
"learning_rate": 4.4384891092608795e-06,
|
303 |
+
"loss": 0.3022,
|
304 |
+
"step": 98
|
305 |
+
},
|
306 |
+
{
|
307 |
+
"epoch": 0.08006405124099279,
|
308 |
+
"learning_rate": 4.485033387477915e-06,
|
309 |
+
"loss": 0.283,
|
310 |
+
"step": 100
|
311 |
+
},
|
312 |
+
{
|
313 |
+
"epoch": 0.08166533226581266,
|
314 |
+
"learning_rate": 4.531749998862628e-06,
|
315 |
+
"loss": 0.0965,
|
316 |
+
"step": 102
|
317 |
+
},
|
318 |
+
{
|
319 |
+
"epoch": 0.08326661329063251,
|
320 |
+
"learning_rate": 4.578637483601732e-06,
|
321 |
+
"loss": 0.2106,
|
322 |
+
"step": 104
|
323 |
+
},
|
324 |
+
{
|
325 |
+
"epoch": 0.08486789431545236,
|
326 |
+
"learning_rate": 4.625694376542399e-06,
|
327 |
+
"loss": 0.0995,
|
328 |
+
"step": 106
|
329 |
+
},
|
330 |
+
{
|
331 |
+
"epoch": 0.08646917534027222,
|
332 |
+
"learning_rate": 4.672919207238145e-06,
|
333 |
+
"loss": 0.1367,
|
334 |
+
"step": 108
|
335 |
+
},
|
336 |
+
{
|
337 |
+
"epoch": 0.08807045636509207,
|
338 |
+
"learning_rate": 4.720310499994664e-06,
|
339 |
+
"loss": 0.0185,
|
340 |
+
"step": 110
|
341 |
+
},
|
342 |
+
{
|
343 |
+
"epoch": 0.08967173738991192,
|
344 |
+
"learning_rate": 4.767866773916041e-06,
|
345 |
+
"loss": 0.0643,
|
346 |
+
"step": 112
|
347 |
+
},
|
348 |
+
{
|
349 |
+
"epoch": 0.09127301841473179,
|
350 |
+
"learning_rate": 4.81558654295099e-06,
|
351 |
+
"loss": 0.4434,
|
352 |
+
"step": 114
|
353 |
+
},
|
354 |
+
{
|
355 |
+
"epoch": 0.09287429943955164,
|
356 |
+
"learning_rate": 4.863468315939234e-06,
|
357 |
+
"loss": 0.19,
|
358 |
+
"step": 116
|
359 |
+
},
|
360 |
+
{
|
361 |
+
"epoch": 0.0944755804643715,
|
362 |
+
"learning_rate": 4.911510596658202e-06,
|
363 |
+
"loss": 0.668,
|
364 |
+
"step": 118
|
365 |
+
},
|
366 |
+
{
|
367 |
+
"epoch": 0.09607686148919135,
|
368 |
+
"learning_rate": 4.959711883869734e-06,
|
369 |
+
"loss": 0.111,
|
370 |
+
"step": 120
|
371 |
+
},
|
372 |
+
{
|
373 |
+
"epoch": 0.0976781425140112,
|
374 |
+
"learning_rate": 5.0080706713669435e-06,
|
375 |
+
"loss": 0.0766,
|
376 |
+
"step": 122
|
377 |
+
},
|
378 |
+
{
|
379 |
+
"epoch": 0.09927942353883107,
|
380 |
+
"learning_rate": 5.056585448021398e-06,
|
381 |
+
"loss": 0.0974,
|
382 |
+
"step": 124
|
383 |
+
},
|
384 |
+
{
|
385 |
+
"epoch": 0.10088070456365092,
|
386 |
+
"learning_rate": 5.105254697830208e-06,
|
387 |
+
"loss": 0.1732,
|
388 |
+
"step": 126
|
389 |
+
},
|
390 |
+
{
|
391 |
+
"epoch": 0.10248198558847077,
|
392 |
+
"learning_rate": 5.154076899963514e-06,
|
393 |
+
"loss": 0.1209,
|
394 |
+
"step": 128
|
395 |
+
},
|
396 |
+
{
|
397 |
+
"epoch": 0.10408326661329063,
|
398 |
+
"learning_rate": 5.203050528811959e-06,
|
399 |
+
"loss": 0.6952,
|
400 |
+
"step": 130
|
401 |
+
},
|
402 |
+
{
|
403 |
+
"epoch": 0.10568454763811048,
|
404 |
+
"learning_rate": 5.2521740540343205e-06,
|
405 |
+
"loss": 0.1867,
|
406 |
+
"step": 132
|
407 |
+
},
|
408 |
+
{
|
409 |
+
"epoch": 0.10728582866293035,
|
410 |
+
"learning_rate": 5.3014459406054295e-06,
|
411 |
+
"loss": 0.1691,
|
412 |
+
"step": 134
|
413 |
+
},
|
414 |
+
{
|
415 |
+
"epoch": 0.1088871096877502,
|
416 |
+
"learning_rate": 5.350864648864026e-06,
|
417 |
+
"loss": 0.1333,
|
418 |
+
"step": 136
|
419 |
+
},
|
420 |
+
{
|
421 |
+
"epoch": 0.11048839071257005,
|
422 |
+
"learning_rate": 5.4004286345609665e-06,
|
423 |
+
"loss": 0.1076,
|
424 |
+
"step": 138
|
425 |
+
},
|
426 |
+
{
|
427 |
+
"epoch": 0.11208967173738991,
|
428 |
+
"learning_rate": 5.450136348907444e-06,
|
429 |
+
"loss": 0.1348,
|
430 |
+
"step": 140
|
431 |
+
},
|
432 |
+
{
|
433 |
+
"epoch": 0.11369095276220977,
|
434 |
+
"learning_rate": 5.499986238623329e-06,
|
435 |
+
"loss": 0.1674,
|
436 |
+
"step": 142
|
437 |
+
},
|
438 |
+
{
|
439 |
+
"epoch": 0.11529223378702963,
|
440 |
+
"learning_rate": 5.549976745985809e-06,
|
441 |
+
"loss": 0.0837,
|
442 |
+
"step": 144
|
443 |
+
},
|
444 |
+
{
|
445 |
+
"epoch": 0.11689351481184948,
|
446 |
+
"learning_rate": 5.6001063088780085e-06,
|
447 |
+
"loss": 0.052,
|
448 |
+
"step": 146
|
449 |
+
},
|
450 |
+
{
|
451 |
+
"epoch": 0.11849479583666933,
|
452 |
+
"learning_rate": 5.650373360837763e-06,
|
453 |
+
"loss": 0.0444,
|
454 |
+
"step": 148
|
455 |
+
},
|
456 |
+
{
|
457 |
+
"epoch": 0.1200960768614892,
|
458 |
+
"learning_rate": 5.700776331106674e-06,
|
459 |
+
"loss": 0.015,
|
460 |
+
"step": 150
|
461 |
+
},
|
462 |
+
{
|
463 |
+
"epoch": 0.12169735788630905,
|
464 |
+
"learning_rate": 5.751313644679071e-06,
|
465 |
+
"loss": 0.0531,
|
466 |
+
"step": 152
|
467 |
+
},
|
468 |
+
{
|
469 |
+
"epoch": 0.1232986389111289,
|
470 |
+
"learning_rate": 5.8019837223513295e-06,
|
471 |
+
"loss": 0.1197,
|
472 |
+
"step": 154
|
473 |
+
},
|
474 |
+
{
|
475 |
+
"epoch": 0.12489991993594876,
|
476 |
+
"learning_rate": 5.852784980771182e-06,
|
477 |
+
"loss": 0.2765,
|
478 |
+
"step": 156
|
479 |
+
},
|
480 |
+
{
|
481 |
+
"epoch": 0.1265012009607686,
|
482 |
+
"learning_rate": 5.903715832487138e-06,
|
483 |
+
"loss": 0.2867,
|
484 |
+
"step": 158
|
485 |
+
},
|
486 |
+
{
|
487 |
+
"epoch": 0.12810248198558846,
|
488 |
+
"learning_rate": 5.954774685998206e-06,
|
489 |
+
"loss": 0.1788,
|
490 |
+
"step": 160
|
491 |
+
},
|
492 |
+
{
|
493 |
+
"epoch": 0.1297037630104083,
|
494 |
+
"learning_rate": 6.005959945803494e-06,
|
495 |
+
"loss": 0.2799,
|
496 |
+
"step": 162
|
497 |
+
},
|
498 |
+
{
|
499 |
+
"epoch": 0.1313050440352282,
|
500 |
+
"learning_rate": 6.057270012452186e-06,
|
501 |
+
"loss": 0.1319,
|
502 |
+
"step": 164
|
503 |
+
},
|
504 |
+
{
|
505 |
+
"epoch": 0.13290632506004804,
|
506 |
+
"learning_rate": 6.108703282593461e-06,
|
507 |
+
"loss": 0.3217,
|
508 |
+
"step": 166
|
509 |
+
},
|
510 |
+
{
|
511 |
+
"epoch": 0.1345076060848679,
|
512 |
+
"learning_rate": 6.160258149026557e-06,
|
513 |
+
"loss": 0.1849,
|
514 |
+
"step": 168
|
515 |
+
},
|
516 |
+
{
|
517 |
+
"epoch": 0.13610888710968774,
|
518 |
+
"learning_rate": 6.2119330007511014e-06,
|
519 |
+
"loss": 0.4212,
|
520 |
+
"step": 170
|
521 |
+
},
|
522 |
+
{
|
523 |
+
"epoch": 0.1377101681345076,
|
524 |
+
"learning_rate": 6.263726223017326e-06,
|
525 |
+
"loss": 0.1075,
|
526 |
+
"step": 172
|
527 |
+
},
|
528 |
+
{
|
529 |
+
"epoch": 0.13931144915932747,
|
530 |
+
"learning_rate": 6.315636197376634e-06,
|
531 |
+
"loss": 0.1672,
|
532 |
+
"step": 174
|
533 |
+
},
|
534 |
+
{
|
535 |
+
"epoch": 0.14091273018414732,
|
536 |
+
"learning_rate": 6.3676613017321305e-06,
|
537 |
+
"loss": 0.4658,
|
538 |
+
"step": 176
|
539 |
+
},
|
540 |
+
{
|
541 |
+
"epoch": 0.14251401120896717,
|
542 |
+
"learning_rate": 6.419799910389257e-06,
|
543 |
+
"loss": 0.1964,
|
544 |
+
"step": 178
|
545 |
+
},
|
546 |
+
{
|
547 |
+
"epoch": 0.14411529223378702,
|
548 |
+
"learning_rate": 6.472050394106689e-06,
|
549 |
+
"loss": 0.042,
|
550 |
+
"step": 180
|
551 |
+
},
|
552 |
+
{
|
553 |
+
"epoch": 0.14571657325860687,
|
554 |
+
"learning_rate": 6.524411120147204e-06,
|
555 |
+
"loss": 0.0147,
|
556 |
+
"step": 182
|
557 |
+
},
|
558 |
+
{
|
559 |
+
"epoch": 0.14731785428342675,
|
560 |
+
"learning_rate": 6.576880452328645e-06,
|
561 |
+
"loss": 0.1999,
|
562 |
+
"step": 184
|
563 |
+
},
|
564 |
+
{
|
565 |
+
"epoch": 0.1489191353082466,
|
566 |
+
"learning_rate": 6.6294567510751675e-06,
|
567 |
+
"loss": 0.8296,
|
568 |
+
"step": 186
|
569 |
+
},
|
570 |
+
{
|
571 |
+
"epoch": 0.15052041633306645,
|
572 |
+
"learning_rate": 6.682138373468341e-06,
|
573 |
+
"loss": 0.1899,
|
574 |
+
"step": 188
|
575 |
+
},
|
576 |
+
{
|
577 |
+
"epoch": 0.1521216973578863,
|
578 |
+
"learning_rate": 6.734923673298605e-06,
|
579 |
+
"loss": 0.27,
|
580 |
+
"step": 190
|
581 |
+
},
|
582 |
+
{
|
583 |
+
"epoch": 0.15372297838270615,
|
584 |
+
"learning_rate": 6.787811001116654e-06,
|
585 |
+
"loss": 0.0206,
|
586 |
+
"step": 192
|
587 |
+
},
|
588 |
+
{
|
589 |
+
"epoch": 0.15532425940752603,
|
590 |
+
"learning_rate": 6.840798704284939e-06,
|
591 |
+
"loss": 0.305,
|
592 |
+
"step": 194
|
593 |
+
},
|
594 |
+
{
|
595 |
+
"epoch": 0.15692554043234588,
|
596 |
+
"learning_rate": 6.893885127029419e-06,
|
597 |
+
"loss": 0.1658,
|
598 |
+
"step": 196
|
599 |
+
},
|
600 |
+
{
|
601 |
+
"epoch": 0.15852682145716573,
|
602 |
+
"learning_rate": 6.94706861049117e-06,
|
603 |
+
"loss": 0.1486,
|
604 |
+
"step": 198
|
605 |
+
},
|
606 |
+
{
|
607 |
+
"epoch": 0.16012810248198558,
|
608 |
+
"learning_rate": 7.000347492778341e-06,
|
609 |
+
"loss": 0.1772,
|
610 |
+
"step": 200
|
611 |
+
},
|
612 |
+
{
|
613 |
+
"epoch": 0.16172938350680544,
|
614 |
+
"learning_rate": 7.05372010901803e-06,
|
615 |
+
"loss": 0.4549,
|
616 |
+
"step": 202
|
617 |
+
},
|
618 |
+
{
|
619 |
+
"epoch": 0.1633306645316253,
|
620 |
+
"learning_rate": 7.1071847914082605e-06,
|
621 |
+
"loss": 0.2341,
|
622 |
+
"step": 204
|
623 |
+
},
|
624 |
+
{
|
625 |
+
"epoch": 0.16493194555644516,
|
626 |
+
"learning_rate": 7.160739869270219e-06,
|
627 |
+
"loss": 0.0896,
|
628 |
+
"step": 206
|
629 |
+
},
|
630 |
+
{
|
631 |
+
"epoch": 0.16653322658126501,
|
632 |
+
"learning_rate": 7.214383669100317e-06,
|
633 |
+
"loss": 0.1866,
|
634 |
+
"step": 208
|
635 |
+
},
|
636 |
+
{
|
637 |
+
"epoch": 0.16813450760608487,
|
638 |
+
"learning_rate": 7.268114514622635e-06,
|
639 |
+
"loss": 0.7071,
|
640 |
+
"step": 210
|
641 |
+
},
|
642 |
+
{
|
643 |
+
"epoch": 0.16973578863090472,
|
644 |
+
"learning_rate": 7.321930726841144e-06,
|
645 |
+
"loss": 0.1683,
|
646 |
+
"step": 212
|
647 |
+
},
|
648 |
+
{
|
649 |
+
"epoch": 0.17133706965572457,
|
650 |
+
"learning_rate": 7.375830624092336e-06,
|
651 |
+
"loss": 0.0287,
|
652 |
+
"step": 214
|
653 |
+
},
|
654 |
+
{
|
655 |
+
"epoch": 0.17293835068054444,
|
656 |
+
"learning_rate": 7.429812522097613e-06,
|
657 |
+
"loss": 0.1243,
|
658 |
+
"step": 216
|
659 |
+
},
|
660 |
+
{
|
661 |
+
"epoch": 0.1745396317053643,
|
662 |
+
"learning_rate": 7.4838747340160475e-06,
|
663 |
+
"loss": 0.204,
|
664 |
+
"step": 218
|
665 |
+
},
|
666 |
+
{
|
667 |
+
"epoch": 0.17614091273018415,
|
668 |
+
"learning_rate": 7.538015570497046e-06,
|
669 |
+
"loss": 0.2267,
|
670 |
+
"step": 220
|
671 |
+
},
|
672 |
+
{
|
673 |
+
"epoch": 0.177742193755004,
|
674 |
+
"learning_rate": 7.592233339733077e-06,
|
675 |
+
"loss": 0.3241,
|
676 |
+
"step": 222
|
677 |
+
},
|
678 |
+
{
|
679 |
+
"epoch": 0.17934347477982385,
|
680 |
+
"learning_rate": 7.646526347512665e-06,
|
681 |
+
"loss": 0.492,
|
682 |
+
"step": 224
|
683 |
+
},
|
684 |
+
{
|
685 |
+
"epoch": 0.18094475580464373,
|
686 |
+
"learning_rate": 7.70089289727319e-06,
|
687 |
+
"loss": 0.0819,
|
688 |
+
"step": 226
|
689 |
+
},
|
690 |
+
{
|
691 |
+
"epoch": 0.18254603682946358,
|
692 |
+
"learning_rate": 7.755331290154041e-06,
|
693 |
+
"loss": 0.1381,
|
694 |
+
"step": 228
|
695 |
+
},
|
696 |
+
{
|
697 |
+
"epoch": 0.18414731785428343,
|
698 |
+
"learning_rate": 7.809839825049565e-06,
|
699 |
+
"loss": 0.0647,
|
700 |
+
"step": 230
|
701 |
+
},
|
702 |
+
{
|
703 |
+
"epoch": 0.18574859887910328,
|
704 |
+
"learning_rate": 7.864416798662347e-06,
|
705 |
+
"loss": 0.1587,
|
706 |
+
"step": 232
|
707 |
+
},
|
708 |
+
{
|
709 |
+
"epoch": 0.18734987990392313,
|
710 |
+
"learning_rate": 7.919060505556376e-06,
|
711 |
+
"loss": 0.0265,
|
712 |
+
"step": 234
|
713 |
+
},
|
714 |
+
{
|
715 |
+
"epoch": 0.188951160928743,
|
716 |
+
"learning_rate": 7.973769238210291e-06,
|
717 |
+
"loss": 0.4234,
|
718 |
+
"step": 236
|
719 |
+
},
|
720 |
+
{
|
721 |
+
"epoch": 0.19055244195356286,
|
722 |
+
"learning_rate": 8.028541287070858e-06,
|
723 |
+
"loss": 0.2822,
|
724 |
+
"step": 238
|
725 |
+
},
|
726 |
+
{
|
727 |
+
"epoch": 0.1921537229783827,
|
728 |
+
"learning_rate": 8.083374940606256e-06,
|
729 |
+
"loss": 0.2843,
|
730 |
+
"step": 240
|
731 |
+
},
|
732 |
+
{
|
733 |
+
"epoch": 0.19375500400320256,
|
734 |
+
"learning_rate": 8.138268485359684e-06,
|
735 |
+
"loss": 0.5846,
|
736 |
+
"step": 242
|
737 |
+
},
|
738 |
+
{
|
739 |
+
"epoch": 0.1953562850280224,
|
740 |
+
"learning_rate": 8.193220206002785e-06,
|
741 |
+
"loss": 0.227,
|
742 |
+
"step": 244
|
743 |
+
},
|
744 |
+
{
|
745 |
+
"epoch": 0.1969575660528423,
|
746 |
+
"learning_rate": 8.248228385389349e-06,
|
747 |
+
"loss": 0.225,
|
748 |
+
"step": 246
|
749 |
+
},
|
750 |
+
{
|
751 |
+
"epoch": 0.19855884707766214,
|
752 |
+
"learning_rate": 8.303291304608936e-06,
|
753 |
+
"loss": 0.1251,
|
754 |
+
"step": 248
|
755 |
+
},
|
756 |
+
{
|
757 |
+
"epoch": 0.200160128102482,
|
758 |
+
"learning_rate": 8.358407243040524e-06,
|
759 |
+
"loss": 0.2108,
|
760 |
+
"step": 250
|
761 |
+
},
|
762 |
+
{
|
763 |
+
"epoch": 0.20176140912730184,
|
764 |
+
"learning_rate": 8.413574478406386e-06,
|
765 |
+
"loss": 0.0656,
|
766 |
+
"step": 252
|
767 |
+
},
|
768 |
+
{
|
769 |
+
"epoch": 0.2033626901521217,
|
770 |
+
"learning_rate": 8.468791286825856e-06,
|
771 |
+
"loss": 0.5503,
|
772 |
+
"step": 254
|
773 |
+
},
|
774 |
+
{
|
775 |
+
"epoch": 0.20496397117694154,
|
776 |
+
"learning_rate": 8.524055942869135e-06,
|
777 |
+
"loss": 0.1978,
|
778 |
+
"step": 256
|
779 |
+
},
|
780 |
+
{
|
781 |
+
"epoch": 0.20656525220176142,
|
782 |
+
"learning_rate": 8.579366719611353e-06,
|
783 |
+
"loss": 0.0869,
|
784 |
+
"step": 258
|
785 |
+
},
|
786 |
+
{
|
787 |
+
"epoch": 0.20816653322658127,
|
788 |
+
"learning_rate": 8.634721888686368e-06,
|
789 |
+
"loss": 0.1127,
|
790 |
+
"step": 260
|
791 |
+
},
|
792 |
+
{
|
793 |
+
"epoch": 0.20976781425140112,
|
794 |
+
"learning_rate": 8.690119720340907e-06,
|
795 |
+
"loss": 0.1418,
|
796 |
+
"step": 262
|
797 |
+
},
|
798 |
+
{
|
799 |
+
"epoch": 0.21136909527622097,
|
800 |
+
"learning_rate": 8.74555848348857e-06,
|
801 |
+
"loss": 0.2559,
|
802 |
+
"step": 264
|
803 |
+
},
|
804 |
+
{
|
805 |
+
"epoch": 0.21297037630104082,
|
806 |
+
"learning_rate": 8.801036445763858e-06,
|
807 |
+
"loss": 0.092,
|
808 |
+
"step": 266
|
809 |
+
},
|
810 |
+
{
|
811 |
+
"epoch": 0.2145716573258607,
|
812 |
+
"learning_rate": 8.856551873576448e-06,
|
813 |
+
"loss": 0.109,
|
814 |
+
"step": 268
|
815 |
+
},
|
816 |
+
{
|
817 |
+
"epoch": 0.21617293835068055,
|
818 |
+
"learning_rate": 8.912103032165206e-06,
|
819 |
+
"loss": 0.2023,
|
820 |
+
"step": 270
|
821 |
+
},
|
822 |
+
{
|
823 |
+
"epoch": 0.2177742193755004,
|
824 |
+
"learning_rate": 8.967688185652527e-06,
|
825 |
+
"loss": 0.1366,
|
826 |
+
"step": 272
|
827 |
+
},
|
828 |
+
{
|
829 |
+
"epoch": 0.21937550040032025,
|
830 |
+
"learning_rate": 9.023305597098526e-06,
|
831 |
+
"loss": 0.124,
|
832 |
+
"step": 274
|
833 |
+
},
|
834 |
+
{
|
835 |
+
"epoch": 0.2209767814251401,
|
836 |
+
"learning_rate": 9.078953528555258e-06,
|
837 |
+
"loss": 0.6121,
|
838 |
+
"step": 276
|
839 |
+
},
|
840 |
+
{
|
841 |
+
"epoch": 0.22257806244995998,
|
842 |
+
"learning_rate": 9.134630241121135e-06,
|
843 |
+
"loss": 0.1334,
|
844 |
+
"step": 278
|
845 |
+
},
|
846 |
+
{
|
847 |
+
"epoch": 0.22417934347477983,
|
848 |
+
"learning_rate": 9.190333994995208e-06,
|
849 |
+
"loss": 0.2663,
|
850 |
+
"step": 280
|
851 |
+
},
|
852 |
+
{
|
853 |
+
"epoch": 0.22578062449959968,
|
854 |
+
"learning_rate": 9.24606304953148e-06,
|
855 |
+
"loss": 0.0988,
|
856 |
+
"step": 282
|
857 |
+
},
|
858 |
+
{
|
859 |
+
"epoch": 0.22738190552441953,
|
860 |
+
"learning_rate": 9.301815663293426e-06,
|
861 |
+
"loss": 0.409,
|
862 |
+
"step": 284
|
863 |
+
},
|
864 |
+
{
|
865 |
+
"epoch": 0.22898318654923938,
|
866 |
+
"learning_rate": 9.35759009410826e-06,
|
867 |
+
"loss": 0.2903,
|
868 |
+
"step": 286
|
869 |
+
},
|
870 |
+
{
|
871 |
+
"epoch": 0.23058446757405926,
|
872 |
+
"learning_rate": 9.41338459912151e-06,
|
873 |
+
"loss": 1.129,
|
874 |
+
"step": 288
|
875 |
+
},
|
876 |
+
{
|
877 |
+
"epoch": 0.2321857485988791,
|
878 |
+
"learning_rate": 9.469197434851414e-06,
|
879 |
+
"loss": 0.0543,
|
880 |
+
"step": 290
|
881 |
+
},
|
882 |
+
{
|
883 |
+
"epoch": 0.23378702962369896,
|
884 |
+
"learning_rate": 9.52502685724336e-06,
|
885 |
+
"loss": 0.37,
|
886 |
+
"step": 292
|
887 |
+
},
|
888 |
+
{
|
889 |
+
"epoch": 0.2353883106485188,
|
890 |
+
"learning_rate": 9.580871121724498e-06,
|
891 |
+
"loss": 0.6854,
|
892 |
+
"step": 294
|
893 |
+
},
|
894 |
+
{
|
895 |
+
"epoch": 0.23698959167333866,
|
896 |
+
"learning_rate": 9.636728483258116e-06,
|
897 |
+
"loss": 0.5659,
|
898 |
+
"step": 296
|
899 |
+
},
|
900 |
+
{
|
901 |
+
"epoch": 0.23859087269815854,
|
902 |
+
"learning_rate": 9.692597196398302e-06,
|
903 |
+
"loss": 0.2975,
|
904 |
+
"step": 298
|
905 |
+
},
|
906 |
+
{
|
907 |
+
"epoch": 0.2401921537229784,
|
908 |
+
"learning_rate": 9.748475515344416e-06,
|
909 |
+
"loss": 0.2951,
|
910 |
+
"step": 300
|
911 |
+
},
|
912 |
+
{
|
913 |
+
"epoch": 0.24179343474779824,
|
914 |
+
"learning_rate": 9.80436169399561e-06,
|
915 |
+
"loss": 0.5071,
|
916 |
+
"step": 302
|
917 |
+
},
|
918 |
+
{
|
919 |
+
"epoch": 0.2433947157726181,
|
920 |
+
"learning_rate": 9.8602539860055e-06,
|
921 |
+
"loss": 0.0566,
|
922 |
+
"step": 304
|
923 |
+
},
|
924 |
+
{
|
925 |
+
"epoch": 0.24499599679743794,
|
926 |
+
"learning_rate": 9.916150644836596e-06,
|
927 |
+
"loss": 0.1737,
|
928 |
+
"step": 306
|
929 |
+
},
|
930 |
+
{
|
931 |
+
"epoch": 0.2465972778222578,
|
932 |
+
"learning_rate": 9.972049923815011e-06,
|
933 |
+
"loss": 0.2336,
|
934 |
+
"step": 308
|
935 |
+
},
|
936 |
+
{
|
937 |
+
"epoch": 0.24819855884707767,
|
938 |
+
"learning_rate": 1.0027950076184982e-05,
|
939 |
+
"loss": 0.3613,
|
940 |
+
"step": 310
|
941 |
+
},
|
942 |
+
{
|
943 |
+
"epoch": 0.24979983987189752,
|
944 |
+
"learning_rate": 1.0083849355163397e-05,
|
945 |
+
"loss": 0.521,
|
946 |
+
"step": 312
|
947 |
+
},
|
948 |
+
{
|
949 |
+
"epoch": 0.2514011208967174,
|
950 |
+
"learning_rate": 1.0139746013994493e-05,
|
951 |
+
"loss": 0.1591,
|
952 |
+
"step": 314
|
953 |
+
},
|
954 |
+
{
|
955 |
+
"epoch": 0.2530024019215372,
|
956 |
+
"learning_rate": 1.0195638306004383e-05,
|
957 |
+
"loss": 0.2951,
|
958 |
+
"step": 316
|
959 |
+
},
|
960 |
+
{
|
961 |
+
"epoch": 0.2546036829463571,
|
962 |
+
"learning_rate": 1.0251524484655577e-05,
|
963 |
+
"loss": 0.0635,
|
964 |
+
"step": 318
|
965 |
+
},
|
966 |
+
{
|
967 |
+
"epoch": 0.2562049639711769,
|
968 |
+
"learning_rate": 1.0307402803601691e-05,
|
969 |
+
"loss": 0.1459,
|
970 |
+
"step": 320
|
971 |
+
},
|
972 |
+
{
|
973 |
+
"epoch": 0.2578062449959968,
|
974 |
+
"learning_rate": 1.0363271516741877e-05,
|
975 |
+
"loss": 0.156,
|
976 |
+
"step": 322
|
977 |
+
},
|
978 |
+
{
|
979 |
+
"epoch": 0.2594075260208166,
|
980 |
+
"learning_rate": 1.0419128878275495e-05,
|
981 |
+
"loss": 0.3462,
|
982 |
+
"step": 324
|
983 |
+
},
|
984 |
+
{
|
985 |
+
"epoch": 0.2610088070456365,
|
986 |
+
"learning_rate": 1.0474973142756632e-05,
|
987 |
+
"loss": 0.1901,
|
988 |
+
"step": 326
|
989 |
+
},
|
990 |
+
{
|
991 |
+
"epoch": 0.2626100880704564,
|
992 |
+
"learning_rate": 1.053080256514858e-05,
|
993 |
+
"loss": 0.2888,
|
994 |
+
"step": 328
|
995 |
+
},
|
996 |
+
{
|
997 |
+
"epoch": 0.2642113690952762,
|
998 |
+
"learning_rate": 1.0586615400878484e-05,
|
999 |
+
"loss": 0.2525,
|
1000 |
+
"step": 330
|
1001 |
+
},
|
1002 |
+
{
|
1003 |
+
"epoch": 0.2658126501200961,
|
1004 |
+
"learning_rate": 1.0642409905891733e-05,
|
1005 |
+
"loss": 0.2064,
|
1006 |
+
"step": 332
|
1007 |
+
},
|
1008 |
+
{
|
1009 |
+
"epoch": 0.2674139311449159,
|
1010 |
+
"learning_rate": 1.0698184336706567e-05,
|
1011 |
+
"loss": 0.318,
|
1012 |
+
"step": 334
|
1013 |
+
},
|
1014 |
+
{
|
1015 |
+
"epoch": 0.2690152121697358,
|
1016 |
+
"learning_rate": 1.0753936950468513e-05,
|
1017 |
+
"loss": 0.3186,
|
1018 |
+
"step": 336
|
1019 |
+
},
|
1020 |
+
{
|
1021 |
+
"epoch": 0.27061649319455566,
|
1022 |
+
"learning_rate": 1.0809666005004787e-05,
|
1023 |
+
"loss": 0.3997,
|
1024 |
+
"step": 338
|
1025 |
+
},
|
1026 |
+
{
|
1027 |
+
"epoch": 0.2722177742193755,
|
1028 |
+
"learning_rate": 1.0865369758878858e-05,
|
1029 |
+
"loss": 0.1775,
|
1030 |
+
"step": 340
|
1031 |
+
},
|
1032 |
+
{
|
1033 |
+
"epoch": 0.27381905524419536,
|
1034 |
+
"learning_rate": 1.0921046471444737e-05,
|
1035 |
+
"loss": 0.2227,
|
1036 |
+
"step": 342
|
1037 |
+
},
|
1038 |
+
{
|
1039 |
+
"epoch": 0.2754203362690152,
|
1040 |
+
"learning_rate": 1.0976694402901467e-05,
|
1041 |
+
"loss": 0.1278,
|
1042 |
+
"step": 344
|
1043 |
+
},
|
1044 |
+
{
|
1045 |
+
"epoch": 0.27702161729383507,
|
1046 |
+
"learning_rate": 1.1032311814347467e-05,
|
1047 |
+
"loss": 0.2362,
|
1048 |
+
"step": 346
|
1049 |
+
},
|
1050 |
+
{
|
1051 |
+
"epoch": 0.27862289831865494,
|
1052 |
+
"learning_rate": 1.1087896967834787e-05,
|
1053 |
+
"loss": 0.2938,
|
1054 |
+
"step": 348
|
1055 |
+
},
|
1056 |
+
{
|
1057 |
+
"epoch": 0.28022417934347477,
|
1058 |
+
"learning_rate": 1.1143448126423545e-05,
|
1059 |
+
"loss": 0.3197,
|
1060 |
+
"step": 350
|
1061 |
+
},
|
1062 |
+
{
|
1063 |
+
"epoch": 0.28182546036829464,
|
1064 |
+
"learning_rate": 1.1198963554236135e-05,
|
1065 |
+
"loss": 0.111,
|
1066 |
+
"step": 352
|
1067 |
+
},
|
1068 |
+
{
|
1069 |
+
"epoch": 0.28342674139311447,
|
1070 |
+
"learning_rate": 1.1254441516511425e-05,
|
1071 |
+
"loss": 0.1264,
|
1072 |
+
"step": 354
|
1073 |
+
},
|
1074 |
+
{
|
1075 |
+
"epoch": 0.28502802241793435,
|
1076 |
+
"learning_rate": 1.1309880279659087e-05,
|
1077 |
+
"loss": 0.2787,
|
1078 |
+
"step": 356
|
1079 |
+
},
|
1080 |
+
{
|
1081 |
+
"epoch": 0.2866293034427542,
|
1082 |
+
"learning_rate": 1.1365278111313625e-05,
|
1083 |
+
"loss": 0.2473,
|
1084 |
+
"step": 358
|
1085 |
+
},
|
1086 |
+
{
|
1087 |
+
"epoch": 0.28823058446757405,
|
1088 |
+
"learning_rate": 1.142063328038864e-05,
|
1089 |
+
"loss": 0.3255,
|
1090 |
+
"step": 360
|
1091 |
+
},
|
1092 |
+
{
|
1093 |
+
"epoch": 0.2898318654923939,
|
1094 |
+
"learning_rate": 1.1475944057130856e-05,
|
1095 |
+
"loss": 0.239,
|
1096 |
+
"step": 362
|
1097 |
+
},
|
1098 |
+
{
|
1099 |
+
"epoch": 0.29143314651721375,
|
1100 |
+
"learning_rate": 1.1531208713174138e-05,
|
1101 |
+
"loss": 0.5546,
|
1102 |
+
"step": 364
|
1103 |
+
},
|
1104 |
+
{
|
1105 |
+
"epoch": 0.2930344275420336,
|
1106 |
+
"learning_rate": 1.1586425521593607e-05,
|
1107 |
+
"loss": 0.0541,
|
1108 |
+
"step": 366
|
1109 |
+
},
|
1110 |
+
{
|
1111 |
+
"epoch": 0.2946357085668535,
|
1112 |
+
"learning_rate": 1.1641592756959467e-05,
|
1113 |
+
"loss": 0.1592,
|
1114 |
+
"step": 368
|
1115 |
+
},
|
1116 |
+
{
|
1117 |
+
"epoch": 0.2962369895916733,
|
1118 |
+
"learning_rate": 1.1696708695391057e-05,
|
1119 |
+
"loss": 0.2305,
|
1120 |
+
"step": 370
|
1121 |
+
},
|
1122 |
+
{
|
1123 |
+
"epoch": 0.2978382706164932,
|
1124 |
+
"learning_rate": 1.1751771614610643e-05,
|
1125 |
+
"loss": 0.1104,
|
1126 |
+
"step": 372
|
1127 |
+
},
|
1128 |
+
{
|
1129 |
+
"epoch": 0.29943955164131303,
|
1130 |
+
"learning_rate": 1.180677979399721e-05,
|
1131 |
+
"loss": 0.1474,
|
1132 |
+
"step": 374
|
1133 |
+
},
|
1134 |
+
{
|
1135 |
+
"epoch": 0.3010408326661329,
|
1136 |
+
"learning_rate": 1.1861731514640309e-05,
|
1137 |
+
"loss": 0.1575,
|
1138 |
+
"step": 376
|
1139 |
+
},
|
1140 |
+
{
|
1141 |
+
"epoch": 0.3026421136909528,
|
1142 |
+
"learning_rate": 1.1916625059393739e-05,
|
1143 |
+
"loss": 0.21,
|
1144 |
+
"step": 378
|
1145 |
+
},
|
1146 |
+
{
|
1147 |
+
"epoch": 0.3042433947157726,
|
1148 |
+
"learning_rate": 1.1971458712929133e-05,
|
1149 |
+
"loss": 0.0702,
|
1150 |
+
"step": 380
|
1151 |
+
},
|
1152 |
+
{
|
1153 |
+
"epoch": 0.3058446757405925,
|
1154 |
+
"learning_rate": 1.2026230761789702e-05,
|
1155 |
+
"loss": 0.1996,
|
1156 |
+
"step": 382
|
1157 |
+
},
|
1158 |
+
{
|
1159 |
+
"epoch": 0.3074459567654123,
|
1160 |
+
"learning_rate": 1.2080939494443618e-05,
|
1161 |
+
"loss": 0.2965,
|
1162 |
+
"step": 384
|
1163 |
+
},
|
1164 |
+
{
|
1165 |
+
"epoch": 0.3090472377902322,
|
1166 |
+
"learning_rate": 1.2135583201337646e-05,
|
1167 |
+
"loss": 0.1831,
|
1168 |
+
"step": 386
|
1169 |
+
},
|
1170 |
+
{
|
1171 |
+
"epoch": 0.31064851881505207,
|
1172 |
+
"learning_rate": 1.2190160174950428e-05,
|
1173 |
+
"loss": 0.1602,
|
1174 |
+
"step": 388
|
1175 |
+
},
|
1176 |
+
{
|
1177 |
+
"epoch": 0.3122497998398719,
|
1178 |
+
"learning_rate": 1.2244668709845952e-05,
|
1179 |
+
"loss": 0.0055,
|
1180 |
+
"step": 390
|
1181 |
+
},
|
1182 |
+
{
|
1183 |
+
"epoch": 0.31385108086469177,
|
1184 |
+
"learning_rate": 1.2299107102726804e-05,
|
1185 |
+
"loss": 0.7346,
|
1186 |
+
"step": 392
|
1187 |
+
},
|
1188 |
+
{
|
1189 |
+
"epoch": 0.3154523618895116,
|
1190 |
+
"learning_rate": 1.2353473652487329e-05,
|
1191 |
+
"loss": 0.043,
|
1192 |
+
"step": 394
|
1193 |
+
},
|
1194 |
+
{
|
1195 |
+
"epoch": 0.31705364291433147,
|
1196 |
+
"learning_rate": 1.2407766660266916e-05,
|
1197 |
+
"loss": 0.0878,
|
1198 |
+
"step": 396
|
1199 |
+
},
|
1200 |
+
{
|
1201 |
+
"epoch": 0.31865492393915135,
|
1202 |
+
"learning_rate": 1.2461984429502947e-05,
|
1203 |
+
"loss": 0.769,
|
1204 |
+
"step": 398
|
1205 |
+
},
|
1206 |
+
{
|
1207 |
+
"epoch": 0.32025620496397117,
|
1208 |
+
"learning_rate": 1.2516125265983945e-05,
|
1209 |
+
"loss": 0.1737,
|
1210 |
+
"step": 400
|
1211 |
+
},
|
1212 |
+
{
|
1213 |
+
"epoch": 0.32185748598879105,
|
1214 |
+
"learning_rate": 1.257018747790238e-05,
|
1215 |
+
"loss": 0.2405,
|
1216 |
+
"step": 402
|
1217 |
+
},
|
1218 |
+
{
|
1219 |
+
"epoch": 0.32345876701361087,
|
1220 |
+
"learning_rate": 1.2624169375907657e-05,
|
1221 |
+
"loss": 0.4884,
|
1222 |
+
"step": 404
|
1223 |
+
},
|
1224 |
+
{
|
1225 |
+
"epoch": 0.32506004803843075,
|
1226 |
+
"learning_rate": 1.2678069273158849e-05,
|
1227 |
+
"loss": 0.194,
|
1228 |
+
"step": 406
|
1229 |
+
},
|
1230 |
+
{
|
1231 |
+
"epoch": 0.3266613290632506,
|
1232 |
+
"learning_rate": 1.273188548537736e-05,
|
1233 |
+
"loss": 0.3149,
|
1234 |
+
"step": 408
|
1235 |
+
},
|
1236 |
+
{
|
1237 |
+
"epoch": 0.32826261008807045,
|
1238 |
+
"learning_rate": 1.2785616330899676e-05,
|
1239 |
+
"loss": 1.8402,
|
1240 |
+
"step": 410
|
1241 |
+
},
|
1242 |
+
{
|
1243 |
+
"epoch": 0.32986389111289033,
|
1244 |
+
"learning_rate": 1.2839260130729776e-05,
|
1245 |
+
"loss": 0.1369,
|
1246 |
+
"step": 412
|
1247 |
+
},
|
1248 |
+
{
|
1249 |
+
"epoch": 0.33146517213771015,
|
1250 |
+
"learning_rate": 1.2892815208591734e-05,
|
1251 |
+
"loss": 0.5522,
|
1252 |
+
"step": 414
|
1253 |
+
},
|
1254 |
+
{
|
1255 |
+
"epoch": 0.33306645316253003,
|
1256 |
+
"learning_rate": 1.2946279890981966e-05,
|
1257 |
+
"loss": 0.2254,
|
1258 |
+
"step": 416
|
1259 |
+
},
|
1260 |
+
{
|
1261 |
+
"epoch": 0.33466773418734985,
|
1262 |
+
"learning_rate": 1.2999652507221652e-05,
|
1263 |
+
"loss": 0.243,
|
1264 |
+
"step": 418
|
1265 |
+
},
|
1266 |
+
{
|
1267 |
+
"epoch": 0.33626901521216973,
|
1268 |
+
"learning_rate": 1.3052931389508822e-05,
|
1269 |
+
"loss": 0.2589,
|
1270 |
+
"step": 420
|
1271 |
+
},
|
1272 |
+
{
|
1273 |
+
"epoch": 0.3378702962369896,
|
1274 |
+
"learning_rate": 1.3106114872970575e-05,
|
1275 |
+
"loss": 0.189,
|
1276 |
+
"step": 422
|
1277 |
+
},
|
1278 |
+
{
|
1279 |
+
"epoch": 0.33947157726180943,
|
1280 |
+
"learning_rate": 1.3159201295715054e-05,
|
1281 |
+
"loss": 0.5865,
|
1282 |
+
"step": 424
|
1283 |
+
},
|
1284 |
+
{
|
1285 |
+
"epoch": 0.3410728582866293,
|
1286 |
+
"learning_rate": 1.321218899888334e-05,
|
1287 |
+
"loss": 0.1693,
|
1288 |
+
"step": 426
|
1289 |
+
},
|
1290 |
+
{
|
1291 |
+
"epoch": 0.34267413931144913,
|
1292 |
+
"learning_rate": 1.326507632670139e-05,
|
1293 |
+
"loss": 0.4265,
|
1294 |
+
"step": 428
|
1295 |
+
},
|
1296 |
+
{
|
1297 |
+
"epoch": 0.344275420336269,
|
1298 |
+
"learning_rate": 1.3317861626531652e-05,
|
1299 |
+
"loss": 0.3939,
|
1300 |
+
"step": 430
|
1301 |
+
},
|
1302 |
+
{
|
1303 |
+
"epoch": 0.3458767013610889,
|
1304 |
+
"learning_rate": 1.3370543248924826e-05,
|
1305 |
+
"loss": 0.4309,
|
1306 |
+
"step": 432
|
1307 |
+
},
|
1308 |
+
{
|
1309 |
+
"epoch": 0.3474779823859087,
|
1310 |
+
"learning_rate": 1.3423119547671348e-05,
|
1311 |
+
"loss": 0.2518,
|
1312 |
+
"step": 434
|
1313 |
+
},
|
1314 |
+
{
|
1315 |
+
"epoch": 0.3490792634107286,
|
1316 |
+
"learning_rate": 1.347558887985279e-05,
|
1317 |
+
"loss": 0.3826,
|
1318 |
+
"step": 436
|
1319 |
+
},
|
1320 |
+
{
|
1321 |
+
"epoch": 0.3506805444355484,
|
1322 |
+
"learning_rate": 1.3527949605893305e-05,
|
1323 |
+
"loss": 0.1508,
|
1324 |
+
"step": 438
|
1325 |
+
},
|
1326 |
+
{
|
1327 |
+
"epoch": 0.3522818254603683,
|
1328 |
+
"learning_rate": 1.3580200089610739e-05,
|
1329 |
+
"loss": 0.4003,
|
1330 |
+
"step": 440
|
1331 |
+
},
|
1332 |
+
{
|
1333 |
+
"epoch": 0.35388310648518817,
|
1334 |
+
"learning_rate": 1.3632338698267863e-05,
|
1335 |
+
"loss": 0.2644,
|
1336 |
+
"step": 442
|
1337 |
+
},
|
1338 |
+
{
|
1339 |
+
"epoch": 0.355484387510008,
|
1340 |
+
"learning_rate": 1.368436380262336e-05,
|
1341 |
+
"loss": 0.3196,
|
1342 |
+
"step": 444
|
1343 |
+
},
|
1344 |
+
{
|
1345 |
+
"epoch": 0.35708566853482787,
|
1346 |
+
"learning_rate": 1.3736273776982667e-05,
|
1347 |
+
"loss": 0.1834,
|
1348 |
+
"step": 446
|
1349 |
+
},
|
1350 |
+
{
|
1351 |
+
"epoch": 0.3586869495596477,
|
1352 |
+
"learning_rate": 1.3788066999248893e-05,
|
1353 |
+
"loss": 0.2493,
|
1354 |
+
"step": 448
|
1355 |
+
},
|
1356 |
+
{
|
1357 |
+
"epoch": 0.3602882305844676,
|
1358 |
+
"learning_rate": 1.3839741850973435e-05,
|
1359 |
+
"loss": 0.1866,
|
1360 |
+
"step": 450
|
1361 |
+
},
|
1362 |
+
{
|
1363 |
+
"epoch": 0.36188951160928745,
|
1364 |
+
"learning_rate": 1.3891296717406533e-05,
|
1365 |
+
"loss": 0.1911,
|
1366 |
+
"step": 452
|
1367 |
+
},
|
1368 |
+
{
|
1369 |
+
"epoch": 0.3634907926341073,
|
1370 |
+
"learning_rate": 1.3942729987547808e-05,
|
1371 |
+
"loss": 0.3398,
|
1372 |
+
"step": 454
|
1373 |
+
},
|
1374 |
+
{
|
1375 |
+
"epoch": 0.36509207365892715,
|
1376 |
+
"learning_rate": 1.3994040054196498e-05,
|
1377 |
+
"loss": 0.4,
|
1378 |
+
"step": 456
|
1379 |
+
},
|
1380 |
+
{
|
1381 |
+
"epoch": 0.366693354683747,
|
1382 |
+
"learning_rate": 1.4045225314001789e-05,
|
1383 |
+
"loss": 0.1742,
|
1384 |
+
"step": 458
|
1385 |
+
},
|
1386 |
+
{
|
1387 |
+
"epoch": 0.36829463570856685,
|
1388 |
+
"learning_rate": 1.4096284167512856e-05,
|
1389 |
+
"loss": 0.161,
|
1390 |
+
"step": 460
|
1391 |
+
},
|
1392 |
+
{
|
1393 |
+
"epoch": 0.36989591673338673,
|
1394 |
+
"learning_rate": 1.4147215019228813e-05,
|
1395 |
+
"loss": 0.1911,
|
1396 |
+
"step": 462
|
1397 |
+
},
|
1398 |
+
{
|
1399 |
+
"epoch": 0.37149719775820655,
|
1400 |
+
"learning_rate": 1.4198016277648665e-05,
|
1401 |
+
"loss": 0.1084,
|
1402 |
+
"step": 464
|
1403 |
+
},
|
1404 |
+
{
|
1405 |
+
"epoch": 0.37309847878302643,
|
1406 |
+
"learning_rate": 1.4248686355320922e-05,
|
1407 |
+
"loss": 0.3494,
|
1408 |
+
"step": 466
|
1409 |
+
},
|
1410 |
+
{
|
1411 |
+
"epoch": 0.37469975980784626,
|
1412 |
+
"learning_rate": 1.429922366889332e-05,
|
1413 |
+
"loss": 0.4,
|
1414 |
+
"step": 468
|
1415 |
+
},
|
1416 |
+
{
|
1417 |
+
"epoch": 0.37630104083266613,
|
1418 |
+
"learning_rate": 1.4349626639162231e-05,
|
1419 |
+
"loss": 0.3706,
|
1420 |
+
"step": 470
|
1421 |
+
},
|
1422 |
+
{
|
1423 |
+
"epoch": 0.377902321857486,
|
1424 |
+
"learning_rate": 1.4399893691121985e-05,
|
1425 |
+
"loss": 0.4302,
|
1426 |
+
"step": 472
|
1427 |
+
},
|
1428 |
+
{
|
1429 |
+
"epoch": 0.37950360288230583,
|
1430 |
+
"learning_rate": 1.4450023254014185e-05,
|
1431 |
+
"loss": 0.2218,
|
1432 |
+
"step": 474
|
1433 |
+
},
|
1434 |
+
{
|
1435 |
+
"epoch": 0.3811048839071257,
|
1436 |
+
"learning_rate": 1.4500013761376663e-05,
|
1437 |
+
"loss": 0.2239,
|
1438 |
+
"step": 476
|
1439 |
+
},
|
1440 |
+
{
|
1441 |
+
"epoch": 0.38270616493194554,
|
1442 |
+
"learning_rate": 1.454986365109255e-05,
|
1443 |
+
"loss": 0.1917,
|
1444 |
+
"step": 478
|
1445 |
+
},
|
1446 |
+
{
|
1447 |
+
"epoch": 0.3843074459567654,
|
1448 |
+
"learning_rate": 1.4599571365439027e-05,
|
1449 |
+
"loss": 0.514,
|
1450 |
+
"step": 480
|
1451 |
+
},
|
1452 |
+
{
|
1453 |
+
"epoch": 0.3859087269815853,
|
1454 |
+
"learning_rate": 1.4649135351135968e-05,
|
1455 |
+
"loss": 0.0929,
|
1456 |
+
"step": 482
|
1457 |
+
},
|
1458 |
+
{
|
1459 |
+
"epoch": 0.3875100080064051,
|
1460 |
+
"learning_rate": 1.4698554059394563e-05,
|
1461 |
+
"loss": 0.2467,
|
1462 |
+
"step": 484
|
1463 |
+
},
|
1464 |
+
{
|
1465 |
+
"epoch": 0.389111289031225,
|
1466 |
+
"learning_rate": 1.4747825945965675e-05,
|
1467 |
+
"loss": 0.2526,
|
1468 |
+
"step": 486
|
1469 |
+
},
|
1470 |
+
{
|
1471 |
+
"epoch": 0.3907125700560448,
|
1472 |
+
"learning_rate": 1.4796949471188033e-05,
|
1473 |
+
"loss": 0.3333,
|
1474 |
+
"step": 488
|
1475 |
+
},
|
1476 |
+
{
|
1477 |
+
"epoch": 0.3923138510808647,
|
1478 |
+
"learning_rate": 1.4845923100036479e-05,
|
1479 |
+
"loss": 0.2936,
|
1480 |
+
"step": 490
|
1481 |
+
},
|
1482 |
+
{
|
1483 |
+
"epoch": 0.3939151321056846,
|
1484 |
+
"learning_rate": 1.4894745302169786e-05,
|
1485 |
+
"loss": 0.168,
|
1486 |
+
"step": 492
|
1487 |
+
},
|
1488 |
+
{
|
1489 |
+
"epoch": 0.3955164131305044,
|
1490 |
+
"learning_rate": 1.4943414551978597e-05,
|
1491 |
+
"loss": 0.2293,
|
1492 |
+
"step": 494
|
1493 |
+
},
|
1494 |
+
{
|
1495 |
+
"epoch": 0.3971176941553243,
|
1496 |
+
"learning_rate": 1.499192932863305e-05,
|
1497 |
+
"loss": 0.2809,
|
1498 |
+
"step": 496
|
1499 |
+
},
|
1500 |
+
{
|
1501 |
+
"epoch": 0.3987189751801441,
|
1502 |
+
"learning_rate": 1.5040288116130261e-05,
|
1503 |
+
"loss": 0.4036,
|
1504 |
+
"step": 498
|
1505 |
+
},
|
1506 |
+
{
|
1507 |
+
"epoch": 0.400320256204964,
|
1508 |
+
"learning_rate": 1.5088489403341793e-05,
|
1509 |
+
"loss": 0.3494,
|
1510 |
+
"step": 500
|
1511 |
+
},
|
1512 |
+
{
|
1513 |
+
"epoch": 0.40192153722978385,
|
1514 |
+
"learning_rate": 1.513653168406076e-05,
|
1515 |
+
"loss": 0.1981,
|
1516 |
+
"step": 502
|
1517 |
+
},
|
1518 |
+
{
|
1519 |
+
"epoch": 0.4035228182546037,
|
1520 |
+
"learning_rate": 1.5184413457049006e-05,
|
1521 |
+
"loss": 0.4441,
|
1522 |
+
"step": 504
|
1523 |
+
},
|
1524 |
+
{
|
1525 |
+
"epoch": 0.40512409927942356,
|
1526 |
+
"learning_rate": 1.5232133226083954e-05,
|
1527 |
+
"loss": 0.1469,
|
1528 |
+
"step": 506
|
1529 |
+
},
|
1530 |
+
{
|
1531 |
+
"epoch": 0.4067253803042434,
|
1532 |
+
"learning_rate": 1.527968950000533e-05,
|
1533 |
+
"loss": 0.3933,
|
1534 |
+
"step": 508
|
1535 |
+
},
|
1536 |
+
{
|
1537 |
+
"epoch": 0.40832666132906326,
|
1538 |
+
"learning_rate": 1.532708079276185e-05,
|
1539 |
+
"loss": 0.1119,
|
1540 |
+
"step": 510
|
1541 |
+
},
|
1542 |
+
{
|
1543 |
+
"epoch": 0.4099279423538831,
|
1544 |
+
"learning_rate": 1.5374305623457594e-05,
|
1545 |
+
"loss": 0.4375,
|
1546 |
+
"step": 512
|
1547 |
+
},
|
1548 |
+
{
|
1549 |
+
"epoch": 0.41152922337870296,
|
1550 |
+
"learning_rate": 1.542136251639826e-05,
|
1551 |
+
"loss": 0.1107,
|
1552 |
+
"step": 514
|
1553 |
+
},
|
1554 |
+
{
|
1555 |
+
"epoch": 0.41313050440352284,
|
1556 |
+
"learning_rate": 1.5468250001137368e-05,
|
1557 |
+
"loss": 0.1726,
|
1558 |
+
"step": 516
|
1559 |
+
},
|
1560 |
+
{
|
1561 |
+
"epoch": 0.41473178542834266,
|
1562 |
+
"learning_rate": 1.551496661252208e-05,
|
1563 |
+
"loss": 0.3434,
|
1564 |
+
"step": 518
|
1565 |
+
},
|
1566 |
+
{
|
1567 |
+
"epoch": 0.41633306645316254,
|
1568 |
+
"learning_rate": 1.5561510890739113e-05,
|
1569 |
+
"loss": 0.3449,
|
1570 |
+
"step": 520
|
1571 |
+
},
|
1572 |
+
{
|
1573 |
+
"epoch": 0.41793434747798236,
|
1574 |
+
"learning_rate": 1.5607881381360296e-05,
|
1575 |
+
"loss": 0.1956,
|
1576 |
+
"step": 522
|
1577 |
+
},
|
1578 |
+
{
|
1579 |
+
"epoch": 0.41953562850280224,
|
1580 |
+
"learning_rate": 1.565407663538797e-05,
|
1581 |
+
"loss": 0.2114,
|
1582 |
+
"step": 524
|
1583 |
+
},
|
1584 |
+
{
|
1585 |
+
"epoch": 0.4211369095276221,
|
1586 |
+
"learning_rate": 1.5700095209300376e-05,
|
1587 |
+
"loss": 0.2248,
|
1588 |
+
"step": 526
|
1589 |
+
},
|
1590 |
+
{
|
1591 |
+
"epoch": 0.42273819055244194,
|
1592 |
+
"learning_rate": 1.5745935665096647e-05,
|
1593 |
+
"loss": 0.1809,
|
1594 |
+
"step": 528
|
1595 |
+
},
|
1596 |
+
{
|
1597 |
+
"epoch": 0.4243394715772618,
|
1598 |
+
"learning_rate": 1.5791596570341844e-05,
|
1599 |
+
"loss": 0.7246,
|
1600 |
+
"step": 530
|
1601 |
+
},
|
1602 |
+
{
|
1603 |
+
"epoch": 0.42594075260208164,
|
1604 |
+
"learning_rate": 1.5837076498211666e-05,
|
1605 |
+
"loss": 0.4036,
|
1606 |
+
"step": 532
|
1607 |
+
},
|
1608 |
+
{
|
1609 |
+
"epoch": 0.4275420336269015,
|
1610 |
+
"learning_rate": 1.5882374027537005e-05,
|
1611 |
+
"loss": 0.1076,
|
1612 |
+
"step": 534
|
1613 |
+
},
|
1614 |
+
{
|
1615 |
+
"epoch": 0.4291433146517214,
|
1616 |
+
"learning_rate": 1.5927487742848448e-05,
|
1617 |
+
"loss": 0.676,
|
1618 |
+
"step": 536
|
1619 |
+
},
|
1620 |
+
{
|
1621 |
+
"epoch": 0.4307445956765412,
|
1622 |
+
"learning_rate": 1.5972416234420393e-05,
|
1623 |
+
"loss": 0.5394,
|
1624 |
+
"step": 538
|
1625 |
+
},
|
1626 |
+
{
|
1627 |
+
"epoch": 0.4323458767013611,
|
1628 |
+
"learning_rate": 1.60171580983152e-05,
|
1629 |
+
"loss": 0.2514,
|
1630 |
+
"step": 540
|
1631 |
+
},
|
1632 |
+
{
|
1633 |
+
"epoch": 0.4339471577261809,
|
1634 |
+
"learning_rate": 1.606171193642703e-05,
|
1635 |
+
"loss": 0.2542,
|
1636 |
+
"step": 542
|
1637 |
+
},
|
1638 |
+
{
|
1639 |
+
"epoch": 0.4355484387510008,
|
1640 |
+
"learning_rate": 1.6106076356525474e-05,
|
1641 |
+
"loss": 0.4332,
|
1642 |
+
"step": 544
|
1643 |
+
},
|
1644 |
+
{
|
1645 |
+
"epoch": 0.4371497197758207,
|
1646 |
+
"learning_rate": 1.6150249972299153e-05,
|
1647 |
+
"loss": 0.1449,
|
1648 |
+
"step": 546
|
1649 |
+
},
|
1650 |
+
{
|
1651 |
+
"epoch": 0.4387510008006405,
|
1652 |
+
"learning_rate": 1.6194231403398994e-05,
|
1653 |
+
"loss": 0.417,
|
1654 |
+
"step": 548
|
1655 |
+
},
|
1656 |
+
{
|
1657 |
+
"epoch": 0.4403522818254604,
|
1658 |
+
"learning_rate": 1.6238019275481313e-05,
|
1659 |
+
"loss": 0.2424,
|
1660 |
+
"step": 550
|
1661 |
+
},
|
1662 |
+
{
|
1663 |
+
"epoch": 0.4419535628502802,
|
1664 |
+
"learning_rate": 1.6281612220250883e-05,
|
1665 |
+
"loss": 0.242,
|
1666 |
+
"step": 552
|
1667 |
+
},
|
1668 |
+
{
|
1669 |
+
"epoch": 0.4435548438751001,
|
1670 |
+
"learning_rate": 1.6325008875503543e-05,
|
1671 |
+
"loss": 0.2277,
|
1672 |
+
"step": 554
|
1673 |
+
},
|
1674 |
+
{
|
1675 |
+
"epoch": 0.44515612489991996,
|
1676 |
+
"learning_rate": 1.6368207885168897e-05,
|
1677 |
+
"loss": 0.3083,
|
1678 |
+
"step": 556
|
1679 |
+
},
|
1680 |
+
{
|
1681 |
+
"epoch": 0.4467574059247398,
|
1682 |
+
"learning_rate": 1.641120789935263e-05,
|
1683 |
+
"loss": 0.3959,
|
1684 |
+
"step": 558
|
1685 |
+
},
|
1686 |
+
{
|
1687 |
+
"epoch": 0.44835868694955966,
|
1688 |
+
"learning_rate": 1.6454007574378637e-05,
|
1689 |
+
"loss": 0.3832,
|
1690 |
+
"step": 560
|
1691 |
+
},
|
1692 |
+
{
|
1693 |
+
"epoch": 0.4499599679743795,
|
1694 |
+
"learning_rate": 1.6496605572831134e-05,
|
1695 |
+
"loss": 0.365,
|
1696 |
+
"step": 562
|
1697 |
+
},
|
1698 |
+
{
|
1699 |
+
"epoch": 0.45156124899919936,
|
1700 |
+
"learning_rate": 1.6539000563596318e-05,
|
1701 |
+
"loss": 0.2909,
|
1702 |
+
"step": 564
|
1703 |
+
},
|
1704 |
+
{
|
1705 |
+
"epoch": 0.45316253002401924,
|
1706 |
+
"learning_rate": 1.6581191221904077e-05,
|
1707 |
+
"loss": 0.1024,
|
1708 |
+
"step": 566
|
1709 |
+
},
|
1710 |
+
{
|
1711 |
+
"epoch": 0.45476381104883906,
|
1712 |
+
"learning_rate": 1.6623176229369324e-05,
|
1713 |
+
"loss": 0.2723,
|
1714 |
+
"step": 568
|
1715 |
+
},
|
1716 |
+
{
|
1717 |
+
"epoch": 0.45636509207365894,
|
1718 |
+
"learning_rate": 1.6664954274033168e-05,
|
1719 |
+
"loss": 0.3812,
|
1720 |
+
"step": 570
|
1721 |
+
},
|
1722 |
+
{
|
1723 |
+
"epoch": 0.45796637309847876,
|
1724 |
+
"learning_rate": 1.6706524050403996e-05,
|
1725 |
+
"loss": 0.0368,
|
1726 |
+
"step": 572
|
1727 |
+
},
|
1728 |
+
{
|
1729 |
+
"epoch": 0.45956765412329864,
|
1730 |
+
"learning_rate": 1.674788425949818e-05,
|
1731 |
+
"loss": 0.3394,
|
1732 |
+
"step": 574
|
1733 |
+
},
|
1734 |
+
{
|
1735 |
+
"epoch": 0.4611689351481185,
|
1736 |
+
"learning_rate": 1.6789033608880735e-05,
|
1737 |
+
"loss": 0.0958,
|
1738 |
+
"step": 576
|
1739 |
+
},
|
1740 |
+
{
|
1741 |
+
"epoch": 0.46277021617293834,
|
1742 |
+
"learning_rate": 1.6829970812705674e-05,
|
1743 |
+
"loss": 0.1656,
|
1744 |
+
"step": 578
|
1745 |
+
},
|
1746 |
+
{
|
1747 |
+
"epoch": 0.4643714971977582,
|
1748 |
+
"learning_rate": 1.6870694591756165e-05,
|
1749 |
+
"loss": 0.4163,
|
1750 |
+
"step": 580
|
1751 |
+
},
|
1752 |
+
{
|
1753 |
+
"epoch": 0.46597277822257804,
|
1754 |
+
"learning_rate": 1.6911203673484577e-05,
|
1755 |
+
"loss": 0.0417,
|
1756 |
+
"step": 582
|
1757 |
+
},
|
1758 |
+
{
|
1759 |
+
"epoch": 0.4675740592473979,
|
1760 |
+
"learning_rate": 1.695149679205214e-05,
|
1761 |
+
"loss": 0.7376,
|
1762 |
+
"step": 584
|
1763 |
+
},
|
1764 |
+
{
|
1765 |
+
"epoch": 0.4691753402722178,
|
1766 |
+
"learning_rate": 1.6991572688368628e-05,
|
1767 |
+
"loss": 0.3798,
|
1768 |
+
"step": 586
|
1769 |
+
},
|
1770 |
+
{
|
1771 |
+
"epoch": 0.4707766212970376,
|
1772 |
+
"learning_rate": 1.7031430110131562e-05,
|
1773 |
+
"loss": 0.4544,
|
1774 |
+
"step": 588
|
1775 |
+
},
|
1776 |
+
{
|
1777 |
+
"epoch": 0.4723779023218575,
|
1778 |
+
"learning_rate": 1.7071067811865467e-05,
|
1779 |
+
"loss": 0.3427,
|
1780 |
+
"step": 590
|
1781 |
+
},
|
1782 |
+
{
|
1783 |
+
"epoch": 0.4739791833466773,
|
1784 |
+
"learning_rate": 1.7110484554960738e-05,
|
1785 |
+
"loss": 0.0882,
|
1786 |
+
"step": 592
|
1787 |
+
},
|
1788 |
+
{
|
1789 |
+
"epoch": 0.4755804643714972,
|
1790 |
+
"learning_rate": 1.7149679107712306e-05,
|
1791 |
+
"loss": 0.5429,
|
1792 |
+
"step": 594
|
1793 |
+
},
|
1794 |
+
{
|
1795 |
+
"epoch": 0.4771817453963171,
|
1796 |
+
"learning_rate": 1.7188650245358215e-05,
|
1797 |
+
"loss": 0.0116,
|
1798 |
+
"step": 596
|
1799 |
+
},
|
1800 |
+
{
|
1801 |
+
"epoch": 0.4787830264211369,
|
1802 |
+
"learning_rate": 1.722739675011779e-05,
|
1803 |
+
"loss": 0.35,
|
1804 |
+
"step": 598
|
1805 |
+
},
|
1806 |
+
{
|
1807 |
+
"epoch": 0.4803843074459568,
|
1808 |
+
"learning_rate": 1.726591741122981e-05,
|
1809 |
+
"loss": 0.2514,
|
1810 |
+
"step": 600
|
1811 |
+
},
|
1812 |
+
{
|
1813 |
+
"epoch": 0.4819855884707766,
|
1814 |
+
"learning_rate": 1.730421102499021e-05,
|
1815 |
+
"loss": 0.4055,
|
1816 |
+
"step": 602
|
1817 |
+
},
|
1818 |
+
{
|
1819 |
+
"epoch": 0.4835868694955965,
|
1820 |
+
"learning_rate": 1.734227639478982e-05,
|
1821 |
+
"loss": 0.4334,
|
1822 |
+
"step": 604
|
1823 |
+
},
|
1824 |
+
{
|
1825 |
+
"epoch": 0.4851881505204163,
|
1826 |
+
"learning_rate": 1.738011233115165e-05,
|
1827 |
+
"loss": 0.2647,
|
1828 |
+
"step": 606
|
1829 |
+
},
|
1830 |
+
{
|
1831 |
+
"epoch": 0.4867894315452362,
|
1832 |
+
"learning_rate": 1.7417717651768144e-05,
|
1833 |
+
"loss": 0.2909,
|
1834 |
+
"step": 608
|
1835 |
+
},
|
1836 |
+
{
|
1837 |
+
"epoch": 0.48839071257005606,
|
1838 |
+
"learning_rate": 1.7455091181538087e-05,
|
1839 |
+
"loss": 0.0694,
|
1840 |
+
"step": 610
|
1841 |
+
},
|
1842 |
+
{
|
1843 |
+
"epoch": 0.4899919935948759,
|
1844 |
+
"learning_rate": 1.74922317526033e-05,
|
1845 |
+
"loss": 0.0212,
|
1846 |
+
"step": 612
|
1847 |
+
},
|
1848 |
+
{
|
1849 |
+
"epoch": 0.49159327461969576,
|
1850 |
+
"learning_rate": 1.752913820438519e-05,
|
1851 |
+
"loss": 0.1282,
|
1852 |
+
"step": 614
|
1853 |
+
},
|
1854 |
+
{
|
1855 |
+
"epoch": 0.4931945556445156,
|
1856 |
+
"learning_rate": 1.756580938362096e-05,
|
1857 |
+
"loss": 0.2675,
|
1858 |
+
"step": 616
|
1859 |
+
},
|
1860 |
+
{
|
1861 |
+
"epoch": 0.49479583666933546,
|
1862 |
+
"learning_rate": 1.7602244144399693e-05,
|
1863 |
+
"loss": 0.5806,
|
1864 |
+
"step": 618
|
1865 |
+
},
|
1866 |
+
{
|
1867 |
+
"epoch": 0.49639711769415534,
|
1868 |
+
"learning_rate": 1.7638441348198147e-05,
|
1869 |
+
"loss": 0.0669,
|
1870 |
+
"step": 620
|
1871 |
+
},
|
1872 |
+
{
|
1873 |
+
"epoch": 0.49799839871897517,
|
1874 |
+
"learning_rate": 1.7674399863916295e-05,
|
1875 |
+
"loss": 0.5326,
|
1876 |
+
"step": 622
|
1877 |
+
},
|
1878 |
+
{
|
1879 |
+
"epoch": 0.49959967974379504,
|
1880 |
+
"learning_rate": 1.771011856791273e-05,
|
1881 |
+
"loss": 0.058,
|
1882 |
+
"step": 624
|
1883 |
+
},
|
1884 |
+
{
|
1885 |
+
"epoch": 0.5012009607686149,
|
1886 |
+
"learning_rate": 1.7745596344039712e-05,
|
1887 |
+
"loss": 0.1398,
|
1888 |
+
"step": 626
|
1889 |
+
},
|
1890 |
+
{
|
1891 |
+
"epoch": 0.5028022417934348,
|
1892 |
+
"learning_rate": 1.7780832083678116e-05,
|
1893 |
+
"loss": 0.1384,
|
1894 |
+
"step": 628
|
1895 |
+
},
|
1896 |
+
{
|
1897 |
+
"epoch": 0.5044035228182546,
|
1898 |
+
"learning_rate": 1.7815824685772035e-05,
|
1899 |
+
"loss": 0.3157,
|
1900 |
+
"step": 630
|
1901 |
+
},
|
1902 |
+
{
|
1903 |
+
"epoch": 0.5060048038430744,
|
1904 |
+
"learning_rate": 1.7850573056863156e-05,
|
1905 |
+
"loss": 0.1699,
|
1906 |
+
"step": 632
|
1907 |
+
},
|
1908 |
+
{
|
1909 |
+
"epoch": 0.5076060848678943,
|
1910 |
+
"learning_rate": 1.7885076111125004e-05,
|
1911 |
+
"loss": 0.5987,
|
1912 |
+
"step": 634
|
1913 |
+
},
|
1914 |
+
{
|
1915 |
+
"epoch": 0.5092073658927142,
|
1916 |
+
"learning_rate": 1.791933277039679e-05,
|
1917 |
+
"loss": 0.3425,
|
1918 |
+
"step": 636
|
1919 |
+
},
|
1920 |
+
{
|
1921 |
+
"epoch": 0.510808646917534,
|
1922 |
+
"learning_rate": 1.7953341964217183e-05,
|
1923 |
+
"loss": 0.0768,
|
1924 |
+
"step": 638
|
1925 |
+
},
|
1926 |
+
{
|
1927 |
+
"epoch": 0.5124099279423538,
|
1928 |
+
"learning_rate": 1.7987102629857696e-05,
|
1929 |
+
"loss": 0.0989,
|
1930 |
+
"step": 640
|
1931 |
+
},
|
1932 |
+
{
|
1933 |
+
"epoch": 0.5140112089671738,
|
1934 |
+
"learning_rate": 1.8020613712355912e-05,
|
1935 |
+
"loss": 0.129,
|
1936 |
+
"step": 642
|
1937 |
+
},
|
1938 |
+
{
|
1939 |
+
"epoch": 0.5156124899919936,
|
1940 |
+
"learning_rate": 1.805387416454847e-05,
|
1941 |
+
"loss": 0.252,
|
1942 |
+
"step": 644
|
1943 |
+
},
|
1944 |
+
{
|
1945 |
+
"epoch": 0.5172137710168134,
|
1946 |
+
"learning_rate": 1.8086882947103787e-05,
|
1947 |
+
"loss": 0.2271,
|
1948 |
+
"step": 646
|
1949 |
+
},
|
1950 |
+
{
|
1951 |
+
"epoch": 0.5188150520416333,
|
1952 |
+
"learning_rate": 1.811963902855447e-05,
|
1953 |
+
"loss": 0.253,
|
1954 |
+
"step": 648
|
1955 |
+
},
|
1956 |
+
{
|
1957 |
+
"epoch": 0.5204163330664532,
|
1958 |
+
"learning_rate": 1.8152141385329658e-05,
|
1959 |
+
"loss": 0.3181,
|
1960 |
+
"step": 650
|
1961 |
+
},
|
1962 |
+
{
|
1963 |
+
"epoch": 0.522017614091273,
|
1964 |
+
"learning_rate": 1.8184389001786895e-05,
|
1965 |
+
"loss": 0.1894,
|
1966 |
+
"step": 652
|
1967 |
+
},
|
1968 |
+
{
|
1969 |
+
"epoch": 0.5236188951160928,
|
1970 |
+
"learning_rate": 1.821638087024396e-05,
|
1971 |
+
"loss": 0.1544,
|
1972 |
+
"step": 654
|
1973 |
+
},
|
1974 |
+
{
|
1975 |
+
"epoch": 0.5252201761409128,
|
1976 |
+
"learning_rate": 1.8248115991010296e-05,
|
1977 |
+
"loss": 0.2476,
|
1978 |
+
"step": 656
|
1979 |
+
},
|
1980 |
+
{
|
1981 |
+
"epoch": 0.5268214571657326,
|
1982 |
+
"learning_rate": 1.8279593372418264e-05,
|
1983 |
+
"loss": 0.4412,
|
1984 |
+
"step": 658
|
1985 |
+
},
|
1986 |
+
{
|
1987 |
+
"epoch": 0.5284227381905524,
|
1988 |
+
"learning_rate": 1.8310812030854155e-05,
|
1989 |
+
"loss": 0.0098,
|
1990 |
+
"step": 660
|
1991 |
+
},
|
1992 |
+
{
|
1993 |
+
"epoch": 0.5300240192153723,
|
1994 |
+
"learning_rate": 1.834177099078887e-05,
|
1995 |
+
"loss": 0.1315,
|
1996 |
+
"step": 662
|
1997 |
+
},
|
1998 |
+
{
|
1999 |
+
"epoch": 0.5316253002401922,
|
2000 |
+
"learning_rate": 1.8372469284808465e-05,
|
2001 |
+
"loss": 0.1228,
|
2002 |
+
"step": 664
|
2003 |
+
},
|
2004 |
+
{
|
2005 |
+
"epoch": 0.533226581265012,
|
2006 |
+
"learning_rate": 1.840290595364436e-05,
|
2007 |
+
"loss": 0.9322,
|
2008 |
+
"step": 666
|
2009 |
+
},
|
2010 |
+
{
|
2011 |
+
"epoch": 0.5348278622898318,
|
2012 |
+
"learning_rate": 1.8433080046203286e-05,
|
2013 |
+
"loss": 0.3805,
|
2014 |
+
"step": 668
|
2015 |
+
},
|
2016 |
+
{
|
2017 |
+
"epoch": 0.5364291433146517,
|
2018 |
+
"learning_rate": 1.8462990619597054e-05,
|
2019 |
+
"loss": 0.1881,
|
2020 |
+
"step": 670
|
2021 |
+
},
|
2022 |
+
{
|
2023 |
+
"epoch": 0.5380304243394716,
|
2024 |
+
"learning_rate": 1.8492636739171966e-05,
|
2025 |
+
"loss": 0.3647,
|
2026 |
+
"step": 672
|
2027 |
+
},
|
2028 |
+
{
|
2029 |
+
"epoch": 0.5396317053642914,
|
2030 |
+
"learning_rate": 1.8522017478538067e-05,
|
2031 |
+
"loss": 0.3042,
|
2032 |
+
"step": 674
|
2033 |
+
},
|
2034 |
+
{
|
2035 |
+
"epoch": 0.5412329863891113,
|
2036 |
+
"learning_rate": 1.855113191959808e-05,
|
2037 |
+
"loss": 0.4389,
|
2038 |
+
"step": 676
|
2039 |
+
},
|
2040 |
+
{
|
2041 |
+
"epoch": 0.5428342674139311,
|
2042 |
+
"learning_rate": 1.8579979152576063e-05,
|
2043 |
+
"loss": 0.1327,
|
2044 |
+
"step": 678
|
2045 |
+
},
|
2046 |
+
{
|
2047 |
+
"epoch": 0.544435548438751,
|
2048 |
+
"learning_rate": 1.8608558276045895e-05,
|
2049 |
+
"loss": 0.2536,
|
2050 |
+
"step": 680
|
2051 |
+
},
|
2052 |
+
{
|
2053 |
+
"epoch": 0.5460368294635709,
|
2054 |
+
"learning_rate": 1.86368683969594e-05,
|
2055 |
+
"loss": 0.1594,
|
2056 |
+
"step": 682
|
2057 |
+
},
|
2058 |
+
{
|
2059 |
+
"epoch": 0.5476381104883907,
|
2060 |
+
"learning_rate": 1.866490863067425e-05,
|
2061 |
+
"loss": 0.1688,
|
2062 |
+
"step": 684
|
2063 |
+
},
|
2064 |
+
{
|
2065 |
+
"epoch": 0.5492393915132106,
|
2066 |
+
"learning_rate": 1.8692678100981663e-05,
|
2067 |
+
"loss": 0.0977,
|
2068 |
+
"step": 686
|
2069 |
+
},
|
2070 |
+
{
|
2071 |
+
"epoch": 0.5508406725380304,
|
2072 |
+
"learning_rate": 1.8720175940133705e-05,
|
2073 |
+
"loss": 0.4264,
|
2074 |
+
"step": 688
|
2075 |
+
},
|
2076 |
+
{
|
2077 |
+
"epoch": 0.5524419535628503,
|
2078 |
+
"learning_rate": 1.8747401288870472e-05,
|
2079 |
+
"loss": 0.0024,
|
2080 |
+
"step": 690
|
2081 |
+
},
|
2082 |
+
{
|
2083 |
+
"epoch": 0.5540432345876701,
|
2084 |
+
"learning_rate": 1.877435329644691e-05,
|
2085 |
+
"loss": 0.2463,
|
2086 |
+
"step": 692
|
2087 |
+
},
|
2088 |
+
{
|
2089 |
+
"epoch": 0.55564451561249,
|
2090 |
+
"learning_rate": 1.8801031120659393e-05,
|
2091 |
+
"loss": 0.1728,
|
2092 |
+
"step": 694
|
2093 |
+
},
|
2094 |
+
{
|
2095 |
+
"epoch": 0.5572457966373099,
|
2096 |
+
"learning_rate": 1.8827433927872066e-05,
|
2097 |
+
"loss": 0.3265,
|
2098 |
+
"step": 696
|
2099 |
+
},
|
2100 |
+
{
|
2101 |
+
"epoch": 0.5588470776621297,
|
2102 |
+
"learning_rate": 1.8853560893042854e-05,
|
2103 |
+
"loss": 0.3823,
|
2104 |
+
"step": 698
|
2105 |
+
},
|
2106 |
+
{
|
2107 |
+
"epoch": 0.5604483586869495,
|
2108 |
+
"learning_rate": 1.8879411199749303e-05,
|
2109 |
+
"loss": 0.1635,
|
2110 |
+
"step": 700
|
2111 |
+
},
|
2112 |
+
{
|
2113 |
+
"epoch": 0.5620496397117695,
|
2114 |
+
"learning_rate": 1.8904984040214037e-05,
|
2115 |
+
"loss": 0.2391,
|
2116 |
+
"step": 702
|
2117 |
+
},
|
2118 |
+
{
|
2119 |
+
"epoch": 0.5636509207365893,
|
2120 |
+
"learning_rate": 1.893027861533002e-05,
|
2121 |
+
"loss": 0.0641,
|
2122 |
+
"step": 704
|
2123 |
+
},
|
2124 |
+
{
|
2125 |
+
"epoch": 0.5652522017614091,
|
2126 |
+
"learning_rate": 1.8955294134685528e-05,
|
2127 |
+
"loss": 0.0634,
|
2128 |
+
"step": 706
|
2129 |
+
},
|
2130 |
+
{
|
2131 |
+
"epoch": 0.5668534827862289,
|
2132 |
+
"learning_rate": 1.898002981658886e-05,
|
2133 |
+
"loss": 0.3219,
|
2134 |
+
"step": 708
|
2135 |
+
},
|
2136 |
+
{
|
2137 |
+
"epoch": 0.5684547638110489,
|
2138 |
+
"learning_rate": 1.9004484888092724e-05,
|
2139 |
+
"loss": 0.2285,
|
2140 |
+
"step": 710
|
2141 |
+
},
|
2142 |
+
{
|
2143 |
+
"epoch": 0.5700560448358687,
|
2144 |
+
"learning_rate": 1.9028658585018455e-05,
|
2145 |
+
"loss": 0.2359,
|
2146 |
+
"step": 712
|
2147 |
+
},
|
2148 |
+
{
|
2149 |
+
"epoch": 0.5716573258606885,
|
2150 |
+
"learning_rate": 1.9052550151979816e-05,
|
2151 |
+
"loss": 0.0369,
|
2152 |
+
"step": 714
|
2153 |
+
},
|
2154 |
+
{
|
2155 |
+
"epoch": 0.5732586068855084,
|
2156 |
+
"learning_rate": 1.9076158842406674e-05,
|
2157 |
+
"loss": 0.024,
|
2158 |
+
"step": 716
|
2159 |
+
},
|
2160 |
+
{
|
2161 |
+
"epoch": 0.5748598879103283,
|
2162 |
+
"learning_rate": 1.909948391856829e-05,
|
2163 |
+
"loss": 0.0789,
|
2164 |
+
"step": 718
|
2165 |
+
},
|
2166 |
+
{
|
2167 |
+
"epoch": 0.5764611689351481,
|
2168 |
+
"learning_rate": 1.912252465159637e-05,
|
2169 |
+
"loss": 0.4336,
|
2170 |
+
"step": 720
|
2171 |
+
},
|
2172 |
+
{
|
2173 |
+
"epoch": 0.578062449959968,
|
2174 |
+
"learning_rate": 1.9145280321507872e-05,
|
2175 |
+
"loss": 0.3042,
|
2176 |
+
"step": 722
|
2177 |
+
},
|
2178 |
+
{
|
2179 |
+
"epoch": 0.5796637309847879,
|
2180 |
+
"learning_rate": 1.9167750217227454e-05,
|
2181 |
+
"loss": 0.245,
|
2182 |
+
"step": 724
|
2183 |
+
},
|
2184 |
+
{
|
2185 |
+
"epoch": 0.5812650120096077,
|
2186 |
+
"learning_rate": 1.9189933636609747e-05,
|
2187 |
+
"loss": 0.0515,
|
2188 |
+
"step": 726
|
2189 |
+
},
|
2190 |
+
{
|
2191 |
+
"epoch": 0.5828662930344275,
|
2192 |
+
"learning_rate": 1.9211829886461274e-05,
|
2193 |
+
"loss": 0.2889,
|
2194 |
+
"step": 728
|
2195 |
+
},
|
2196 |
+
{
|
2197 |
+
"epoch": 0.5844675740592474,
|
2198 |
+
"learning_rate": 1.9233438282562085e-05,
|
2199 |
+
"loss": 0.0882,
|
2200 |
+
"step": 730
|
2201 |
+
},
|
2202 |
+
{
|
2203 |
+
"epoch": 0.5860688550840673,
|
2204 |
+
"learning_rate": 1.925475814968719e-05,
|
2205 |
+
"loss": 0.0002,
|
2206 |
+
"step": 732
|
2207 |
+
},
|
2208 |
+
{
|
2209 |
+
"epoch": 0.5876701361088871,
|
2210 |
+
"learning_rate": 1.9275788821627607e-05,
|
2211 |
+
"loss": 0.0721,
|
2212 |
+
"step": 734
|
2213 |
+
},
|
2214 |
+
{
|
2215 |
+
"epoch": 0.589271417133707,
|
2216 |
+
"learning_rate": 1.9296529641211215e-05,
|
2217 |
+
"loss": 0.01,
|
2218 |
+
"step": 736
|
2219 |
+
},
|
2220 |
+
{
|
2221 |
+
"epoch": 0.5908726981585268,
|
2222 |
+
"learning_rate": 1.9316979960323286e-05,
|
2223 |
+
"loss": 0.021,
|
2224 |
+
"step": 738
|
2225 |
+
},
|
2226 |
+
{
|
2227 |
+
"epoch": 0.5924739791833467,
|
2228 |
+
"learning_rate": 1.9337139139926707e-05,
|
2229 |
+
"loss": 0.5328,
|
2230 |
+
"step": 740
|
2231 |
+
},
|
2232 |
+
{
|
2233 |
+
"epoch": 0.5940752602081665,
|
2234 |
+
"learning_rate": 1.935700655008199e-05,
|
2235 |
+
"loss": 0.4535,
|
2236 |
+
"step": 742
|
2237 |
+
},
|
2238 |
+
{
|
2239 |
+
"epoch": 0.5956765412329864,
|
2240 |
+
"learning_rate": 1.9376581569966933e-05,
|
2241 |
+
"loss": 0.2055,
|
2242 |
+
"step": 744
|
2243 |
+
},
|
2244 |
+
{
|
2245 |
+
"epoch": 0.5972778222578062,
|
2246 |
+
"learning_rate": 1.939586358789602e-05,
|
2247 |
+
"loss": 0.0065,
|
2248 |
+
"step": 746
|
2249 |
+
},
|
2250 |
+
{
|
2251 |
+
"epoch": 0.5988791032826261,
|
2252 |
+
"learning_rate": 1.9414852001339547e-05,
|
2253 |
+
"loss": 0.7148,
|
2254 |
+
"step": 748
|
2255 |
+
},
|
2256 |
+
{
|
2257 |
+
"epoch": 0.600480384307446,
|
2258 |
+
"learning_rate": 1.9433546216942423e-05,
|
2259 |
+
"loss": 0.0204,
|
2260 |
+
"step": 750
|
2261 |
+
},
|
2262 |
+
{
|
2263 |
+
"epoch": 0.6020816653322658,
|
2264 |
+
"learning_rate": 1.945194565054276e-05,
|
2265 |
+
"loss": 0.8355,
|
2266 |
+
"step": 752
|
2267 |
+
},
|
2268 |
+
{
|
2269 |
+
"epoch": 0.6036829463570856,
|
2270 |
+
"learning_rate": 1.9470049727190073e-05,
|
2271 |
+
"loss": 0.5139,
|
2272 |
+
"step": 754
|
2273 |
+
},
|
2274 |
+
{
|
2275 |
+
"epoch": 0.6052842273819056,
|
2276 |
+
"learning_rate": 1.948785788116329e-05,
|
2277 |
+
"loss": 0.0002,
|
2278 |
+
"step": 756
|
2279 |
+
},
|
2280 |
+
{
|
2281 |
+
"epoch": 0.6068855084067254,
|
2282 |
+
"learning_rate": 1.9505369555988395e-05,
|
2283 |
+
"loss": 0.2042,
|
2284 |
+
"step": 758
|
2285 |
+
},
|
2286 |
+
{
|
2287 |
+
"epoch": 0.6084867894315452,
|
2288 |
+
"learning_rate": 1.952258420445583e-05,
|
2289 |
+
"loss": 0.0885,
|
2290 |
+
"step": 760
|
2291 |
+
},
|
2292 |
+
{
|
2293 |
+
"epoch": 0.610088070456365,
|
2294 |
+
"learning_rate": 1.953950128863762e-05,
|
2295 |
+
"loss": 0.0436,
|
2296 |
+
"step": 762
|
2297 |
+
},
|
2298 |
+
{
|
2299 |
+
"epoch": 0.611689351481185,
|
2300 |
+
"learning_rate": 1.9556120279904144e-05,
|
2301 |
+
"loss": 0.5328,
|
2302 |
+
"step": 764
|
2303 |
+
},
|
2304 |
+
{
|
2305 |
+
"epoch": 0.6132906325060048,
|
2306 |
+
"learning_rate": 1.957244065894066e-05,
|
2307 |
+
"loss": 0.2262,
|
2308 |
+
"step": 766
|
2309 |
+
},
|
2310 |
+
{
|
2311 |
+
"epoch": 0.6148919135308246,
|
2312 |
+
"learning_rate": 1.9588461915763566e-05,
|
2313 |
+
"loss": 0.3054,
|
2314 |
+
"step": 768
|
2315 |
+
},
|
2316 |
+
{
|
2317 |
+
"epoch": 0.6164931945556446,
|
2318 |
+
"learning_rate": 1.9604183549736283e-05,
|
2319 |
+
"loss": 0.3508,
|
2320 |
+
"step": 770
|
2321 |
+
},
|
2322 |
+
{
|
2323 |
+
"epoch": 0.6180944755804644,
|
2324 |
+
"learning_rate": 1.9619605069584954e-05,
|
2325 |
+
"loss": 0.5364,
|
2326 |
+
"step": 772
|
2327 |
+
},
|
2328 |
+
{
|
2329 |
+
"epoch": 0.6196957566052842,
|
2330 |
+
"learning_rate": 1.9634725993413744e-05,
|
2331 |
+
"loss": 0.1168,
|
2332 |
+
"step": 774
|
2333 |
+
},
|
2334 |
+
{
|
2335 |
+
"epoch": 0.6212970376301041,
|
2336 |
+
"learning_rate": 1.964954584871995e-05,
|
2337 |
+
"loss": 0.4526,
|
2338 |
+
"step": 776
|
2339 |
+
},
|
2340 |
+
{
|
2341 |
+
"epoch": 0.622898318654924,
|
2342 |
+
"learning_rate": 1.966406417240872e-05,
|
2343 |
+
"loss": 0.2059,
|
2344 |
+
"step": 778
|
2345 |
+
},
|
2346 |
+
{
|
2347 |
+
"epoch": 0.6244995996797438,
|
2348 |
+
"learning_rate": 1.967828051080755e-05,
|
2349 |
+
"loss": 0.2176,
|
2350 |
+
"step": 780
|
2351 |
+
},
|
2352 |
+
{
|
2353 |
+
"epoch": 0.6261008807045636,
|
2354 |
+
"learning_rate": 1.969219441968046e-05,
|
2355 |
+
"loss": 0.0018,
|
2356 |
+
"step": 782
|
2357 |
+
},
|
2358 |
+
{
|
2359 |
+
"epoch": 0.6277021617293835,
|
2360 |
+
"learning_rate": 1.9705805464241856e-05,
|
2361 |
+
"loss": 0.1679,
|
2362 |
+
"step": 784
|
2363 |
+
},
|
2364 |
+
{
|
2365 |
+
"epoch": 0.6293034427542034,
|
2366 |
+
"learning_rate": 1.971911321917015e-05,
|
2367 |
+
"loss": 0.1927,
|
2368 |
+
"step": 786
|
2369 |
+
},
|
2370 |
+
{
|
2371 |
+
"epoch": 0.6309047237790232,
|
2372 |
+
"learning_rate": 1.9732117268621005e-05,
|
2373 |
+
"loss": 0.4173,
|
2374 |
+
"step": 788
|
2375 |
+
},
|
2376 |
+
{
|
2377 |
+
"epoch": 0.6325060048038431,
|
2378 |
+
"learning_rate": 1.9744817206240377e-05,
|
2379 |
+
"loss": 0.3615,
|
2380 |
+
"step": 790
|
2381 |
+
},
|
2382 |
+
{
|
2383 |
+
"epoch": 0.6341072858286629,
|
2384 |
+
"learning_rate": 1.9757212635177177e-05,
|
2385 |
+
"loss": 0.1433,
|
2386 |
+
"step": 792
|
2387 |
+
},
|
2388 |
+
{
|
2389 |
+
"epoch": 0.6357085668534828,
|
2390 |
+
"learning_rate": 1.976930316809569e-05,
|
2391 |
+
"loss": 0.2274,
|
2392 |
+
"step": 794
|
2393 |
+
},
|
2394 |
+
{
|
2395 |
+
"epoch": 0.6373098478783027,
|
2396 |
+
"learning_rate": 1.978108842718768e-05,
|
2397 |
+
"loss": 0.4554,
|
2398 |
+
"step": 796
|
2399 |
+
},
|
2400 |
+
{
|
2401 |
+
"epoch": 0.6389111289031225,
|
2402 |
+
"learning_rate": 1.9792568044184176e-05,
|
2403 |
+
"loss": 0.309,
|
2404 |
+
"step": 798
|
2405 |
+
},
|
2406 |
+
{
|
2407 |
+
"epoch": 0.6405124099279423,
|
2408 |
+
"learning_rate": 1.9803741660367015e-05,
|
2409 |
+
"loss": 0.125,
|
2410 |
+
"step": 800
|
2411 |
+
},
|
2412 |
+
{
|
2413 |
+
"epoch": 0.6421136909527622,
|
2414 |
+
"learning_rate": 1.9814608926580007e-05,
|
2415 |
+
"loss": 0.2541,
|
2416 |
+
"step": 802
|
2417 |
+
},
|
2418 |
+
{
|
2419 |
+
"epoch": 0.6437149719775821,
|
2420 |
+
"learning_rate": 1.9825169503239885e-05,
|
2421 |
+
"loss": 0.4126,
|
2422 |
+
"step": 804
|
2423 |
+
},
|
2424 |
+
{
|
2425 |
+
"epoch": 0.6453162530024019,
|
2426 |
+
"learning_rate": 1.9835423060346892e-05,
|
2427 |
+
"loss": 0.2168,
|
2428 |
+
"step": 806
|
2429 |
+
},
|
2430 |
+
{
|
2431 |
+
"epoch": 0.6469175340272217,
|
2432 |
+
"learning_rate": 1.9845369277495102e-05,
|
2433 |
+
"loss": 0.2062,
|
2434 |
+
"step": 808
|
2435 |
+
},
|
2436 |
+
{
|
2437 |
+
"epoch": 0.6485188150520417,
|
2438 |
+
"learning_rate": 1.985500784388244e-05,
|
2439 |
+
"loss": 0.3649,
|
2440 |
+
"step": 810
|
2441 |
+
},
|
2442 |
+
{
|
2443 |
+
"epoch": 0.6501200960768615,
|
2444 |
+
"learning_rate": 1.9864338458320366e-05,
|
2445 |
+
"loss": 0.3101,
|
2446 |
+
"step": 812
|
2447 |
+
},
|
2448 |
+
{
|
2449 |
+
"epoch": 0.6517213771016813,
|
2450 |
+
"learning_rate": 1.9873360829243323e-05,
|
2451 |
+
"loss": 0.0315,
|
2452 |
+
"step": 814
|
2453 |
+
},
|
2454 |
+
{
|
2455 |
+
"epoch": 0.6533226581265013,
|
2456 |
+
"learning_rate": 1.9882074674717836e-05,
|
2457 |
+
"loss": 0.2392,
|
2458 |
+
"step": 816
|
2459 |
+
},
|
2460 |
+
{
|
2461 |
+
"epoch": 0.6549239391513211,
|
2462 |
+
"learning_rate": 1.989047972245129e-05,
|
2463 |
+
"loss": 0.0814,
|
2464 |
+
"step": 818
|
2465 |
+
},
|
2466 |
+
{
|
2467 |
+
"epoch": 0.6565252201761409,
|
2468 |
+
"learning_rate": 1.989857570980049e-05,
|
2469 |
+
"loss": 0.1961,
|
2470 |
+
"step": 820
|
2471 |
+
},
|
2472 |
+
{
|
2473 |
+
"epoch": 0.6581265012009607,
|
2474 |
+
"learning_rate": 1.9906362383779826e-05,
|
2475 |
+
"loss": 0.0243,
|
2476 |
+
"step": 822
|
2477 |
+
},
|
2478 |
+
{
|
2479 |
+
"epoch": 0.6597277822257807,
|
2480 |
+
"learning_rate": 1.9913839501069213e-05,
|
2481 |
+
"loss": 0.3149,
|
2482 |
+
"step": 824
|
2483 |
+
},
|
2484 |
+
{
|
2485 |
+
"epoch": 0.6613290632506005,
|
2486 |
+
"learning_rate": 1.9921006828021666e-05,
|
2487 |
+
"loss": 0.0862,
|
2488 |
+
"step": 826
|
2489 |
+
},
|
2490 |
+
{
|
2491 |
+
"epoch": 0.6629303442754203,
|
2492 |
+
"learning_rate": 1.9927864140670615e-05,
|
2493 |
+
"loss": 0.2755,
|
2494 |
+
"step": 828
|
2495 |
+
},
|
2496 |
+
{
|
2497 |
+
"epoch": 0.6645316253002402,
|
2498 |
+
"learning_rate": 1.99344112247369e-05,
|
2499 |
+
"loss": 0.2877,
|
2500 |
+
"step": 830
|
2501 |
+
},
|
2502 |
+
{
|
2503 |
+
"epoch": 0.6661329063250601,
|
2504 |
+
"learning_rate": 1.9940647875635463e-05,
|
2505 |
+
"loss": 0.2791,
|
2506 |
+
"step": 832
|
2507 |
+
},
|
2508 |
+
{
|
2509 |
+
"epoch": 0.6677341873498799,
|
2510 |
+
"learning_rate": 1.994657389848176e-05,
|
2511 |
+
"loss": 0.1761,
|
2512 |
+
"step": 834
|
2513 |
+
},
|
2514 |
+
{
|
2515 |
+
"epoch": 0.6693354683746997,
|
2516 |
+
"learning_rate": 1.9952189108097825e-05,
|
2517 |
+
"loss": 0.2275,
|
2518 |
+
"step": 836
|
2519 |
+
},
|
2520 |
+
{
|
2521 |
+
"epoch": 0.6709367493995196,
|
2522 |
+
"learning_rate": 1.9957493329018064e-05,
|
2523 |
+
"loss": 0.0233,
|
2524 |
+
"step": 838
|
2525 |
+
},
|
2526 |
+
{
|
2527 |
+
"epoch": 0.6725380304243395,
|
2528 |
+
"learning_rate": 1.996248639549475e-05,
|
2529 |
+
"loss": 0.2653,
|
2530 |
+
"step": 840
|
2531 |
+
},
|
2532 |
+
{
|
2533 |
+
"epoch": 0.6741393114491593,
|
2534 |
+
"learning_rate": 1.9967168151503196e-05,
|
2535 |
+
"loss": 0.3435,
|
2536 |
+
"step": 842
|
2537 |
+
},
|
2538 |
+
{
|
2539 |
+
"epoch": 0.6757405924739792,
|
2540 |
+
"learning_rate": 1.997153845074662e-05,
|
2541 |
+
"loss": 0.0327,
|
2542 |
+
"step": 844
|
2543 |
+
},
|
2544 |
+
{
|
2545 |
+
"epoch": 0.677341873498799,
|
2546 |
+
"learning_rate": 1.997559715666073e-05,
|
2547 |
+
"loss": 0.2651,
|
2548 |
+
"step": 846
|
2549 |
+
},
|
2550 |
+
{
|
2551 |
+
"epoch": 0.6789431545236189,
|
2552 |
+
"learning_rate": 1.9979344142417986e-05,
|
2553 |
+
"loss": 0.3056,
|
2554 |
+
"step": 848
|
2555 |
+
},
|
2556 |
+
{
|
2557 |
+
"epoch": 0.6805444355484388,
|
2558 |
+
"learning_rate": 1.998277929093157e-05,
|
2559 |
+
"loss": 0.1792,
|
2560 |
+
"step": 850
|
2561 |
+
},
|
2562 |
+
{
|
2563 |
+
"epoch": 0.6821457165732586,
|
2564 |
+
"learning_rate": 1.9985902494859023e-05,
|
2565 |
+
"loss": 0.1139,
|
2566 |
+
"step": 852
|
2567 |
+
},
|
2568 |
+
{
|
2569 |
+
"epoch": 0.6837469975980784,
|
2570 |
+
"learning_rate": 1.9988713656605635e-05,
|
2571 |
+
"loss": 0.548,
|
2572 |
+
"step": 854
|
2573 |
+
},
|
2574 |
+
{
|
2575 |
+
"epoch": 0.6853482786228983,
|
2576 |
+
"learning_rate": 1.9991212688327456e-05,
|
2577 |
+
"loss": 0.1276,
|
2578 |
+
"step": 856
|
2579 |
+
},
|
2580 |
+
{
|
2581 |
+
"epoch": 0.6869495596477182,
|
2582 |
+
"learning_rate": 1.999339951193407e-05,
|
2583 |
+
"loss": 0.0769,
|
2584 |
+
"step": 858
|
2585 |
+
},
|
2586 |
+
{
|
2587 |
+
"epoch": 0.688550840672538,
|
2588 |
+
"learning_rate": 1.9995274059091018e-05,
|
2589 |
+
"loss": 0.2463,
|
2590 |
+
"step": 860
|
2591 |
+
},
|
2592 |
+
{
|
2593 |
+
"epoch": 0.6901521216973578,
|
2594 |
+
"learning_rate": 1.999683627122195e-05,
|
2595 |
+
"loss": 0.5896,
|
2596 |
+
"step": 862
|
2597 |
+
},
|
2598 |
+
{
|
2599 |
+
"epoch": 0.6917534027221778,
|
2600 |
+
"learning_rate": 1.9998086099510433e-05,
|
2601 |
+
"loss": 0.0092,
|
2602 |
+
"step": 864
|
2603 |
+
},
|
2604 |
+
{
|
2605 |
+
"epoch": 0.6933546837469976,
|
2606 |
+
"learning_rate": 1.99990235049015e-05,
|
2607 |
+
"loss": 0.3945,
|
2608 |
+
"step": 866
|
2609 |
+
},
|
2610 |
+
{
|
2611 |
+
"epoch": 0.6949559647718174,
|
2612 |
+
"learning_rate": 1.999964845810285e-05,
|
2613 |
+
"loss": 0.3901,
|
2614 |
+
"step": 868
|
2615 |
+
},
|
2616 |
+
{
|
2617 |
+
"epoch": 0.6965572457966374,
|
2618 |
+
"learning_rate": 1.999996093958578e-05,
|
2619 |
+
"loss": 0.0916,
|
2620 |
+
"step": 870
|
2621 |
+
},
|
2622 |
+
{
|
2623 |
+
"epoch": 0.6981585268214572,
|
2624 |
+
"learning_rate": 1.999996093958578e-05,
|
2625 |
+
"loss": 0.2292,
|
2626 |
+
"step": 872
|
2627 |
+
},
|
2628 |
+
{
|
2629 |
+
"epoch": 0.699759807846277,
|
2630 |
+
"learning_rate": 1.999964845810285e-05,
|
2631 |
+
"loss": 0.1592,
|
2632 |
+
"step": 874
|
2633 |
+
},
|
2634 |
+
{
|
2635 |
+
"epoch": 0.7013610888710968,
|
2636 |
+
"learning_rate": 1.99990235049015e-05,
|
2637 |
+
"loss": 0.4008,
|
2638 |
+
"step": 876
|
2639 |
+
},
|
2640 |
+
{
|
2641 |
+
"epoch": 0.7029623698959168,
|
2642 |
+
"learning_rate": 1.9998086099510433e-05,
|
2643 |
+
"loss": 0.1462,
|
2644 |
+
"step": 878
|
2645 |
+
},
|
2646 |
+
{
|
2647 |
+
"epoch": 0.7045636509207366,
|
2648 |
+
"learning_rate": 1.999683627122195e-05,
|
2649 |
+
"loss": 0.0401,
|
2650 |
+
"step": 880
|
2651 |
+
},
|
2652 |
+
{
|
2653 |
+
"epoch": 0.7061649319455564,
|
2654 |
+
"learning_rate": 1.999527405909102e-05,
|
2655 |
+
"loss": 0.1604,
|
2656 |
+
"step": 882
|
2657 |
+
},
|
2658 |
+
{
|
2659 |
+
"epoch": 0.7077662129703763,
|
2660 |
+
"learning_rate": 1.999339951193407e-05,
|
2661 |
+
"loss": 0.1657,
|
2662 |
+
"step": 884
|
2663 |
+
},
|
2664 |
+
{
|
2665 |
+
"epoch": 0.7093674939951962,
|
2666 |
+
"learning_rate": 1.9991212688327456e-05,
|
2667 |
+
"loss": 0.2581,
|
2668 |
+
"step": 886
|
2669 |
+
},
|
2670 |
+
{
|
2671 |
+
"epoch": 0.710968775020016,
|
2672 |
+
"learning_rate": 1.9988713656605635e-05,
|
2673 |
+
"loss": 0.0265,
|
2674 |
+
"step": 888
|
2675 |
+
},
|
2676 |
+
{
|
2677 |
+
"epoch": 0.7125700560448359,
|
2678 |
+
"learning_rate": 1.9985902494859026e-05,
|
2679 |
+
"loss": 0.0464,
|
2680 |
+
"step": 890
|
2681 |
+
},
|
2682 |
+
{
|
2683 |
+
"epoch": 0.7141713370696557,
|
2684 |
+
"learning_rate": 1.9982779290931572e-05,
|
2685 |
+
"loss": 0.0836,
|
2686 |
+
"step": 892
|
2687 |
+
},
|
2688 |
+
{
|
2689 |
+
"epoch": 0.7157726180944756,
|
2690 |
+
"learning_rate": 1.997934414241799e-05,
|
2691 |
+
"loss": 0.0439,
|
2692 |
+
"step": 894
|
2693 |
+
},
|
2694 |
+
{
|
2695 |
+
"epoch": 0.7173738991192954,
|
2696 |
+
"learning_rate": 1.997559715666073e-05,
|
2697 |
+
"loss": 0.0822,
|
2698 |
+
"step": 896
|
2699 |
+
},
|
2700 |
+
{
|
2701 |
+
"epoch": 0.7189751801441153,
|
2702 |
+
"learning_rate": 1.997153845074662e-05,
|
2703 |
+
"loss": 0.2426,
|
2704 |
+
"step": 898
|
2705 |
+
},
|
2706 |
+
{
|
2707 |
+
"epoch": 0.7205764611689351,
|
2708 |
+
"learning_rate": 1.9967168151503193e-05,
|
2709 |
+
"loss": 0.0023,
|
2710 |
+
"step": 900
|
2711 |
+
},
|
2712 |
+
{
|
2713 |
+
"epoch": 0.722177742193755,
|
2714 |
+
"learning_rate": 1.9962486395494753e-05,
|
2715 |
+
"loss": 0.114,
|
2716 |
+
"step": 902
|
2717 |
+
},
|
2718 |
+
{
|
2719 |
+
"epoch": 0.7237790232185749,
|
2720 |
+
"learning_rate": 1.9957493329018064e-05,
|
2721 |
+
"loss": 0.532,
|
2722 |
+
"step": 904
|
2723 |
+
},
|
2724 |
+
{
|
2725 |
+
"epoch": 0.7253803042433947,
|
2726 |
+
"learning_rate": 1.9952189108097825e-05,
|
2727 |
+
"loss": 0.4572,
|
2728 |
+
"step": 906
|
2729 |
+
},
|
2730 |
+
{
|
2731 |
+
"epoch": 0.7269815852682145,
|
2732 |
+
"learning_rate": 1.994657389848176e-05,
|
2733 |
+
"loss": 0.3566,
|
2734 |
+
"step": 908
|
2735 |
+
},
|
2736 |
+
{
|
2737 |
+
"epoch": 0.7285828662930345,
|
2738 |
+
"learning_rate": 1.9940647875635466e-05,
|
2739 |
+
"loss": 0.205,
|
2740 |
+
"step": 910
|
2741 |
+
},
|
2742 |
+
{
|
2743 |
+
"epoch": 0.7301841473178543,
|
2744 |
+
"learning_rate": 1.99344112247369e-05,
|
2745 |
+
"loss": 0.4722,
|
2746 |
+
"step": 912
|
2747 |
+
},
|
2748 |
+
{
|
2749 |
+
"epoch": 0.7317854283426741,
|
2750 |
+
"learning_rate": 1.9927864140670618e-05,
|
2751 |
+
"loss": 0.021,
|
2752 |
+
"step": 914
|
2753 |
+
},
|
2754 |
+
{
|
2755 |
+
"epoch": 0.733386709367494,
|
2756 |
+
"learning_rate": 1.9921006828021666e-05,
|
2757 |
+
"loss": 0.1926,
|
2758 |
+
"step": 916
|
2759 |
+
},
|
2760 |
+
{
|
2761 |
+
"epoch": 0.7349879903923139,
|
2762 |
+
"learning_rate": 1.9913839501069213e-05,
|
2763 |
+
"loss": 0.2962,
|
2764 |
+
"step": 918
|
2765 |
+
},
|
2766 |
+
{
|
2767 |
+
"epoch": 0.7365892714171337,
|
2768 |
+
"learning_rate": 1.9906362383779826e-05,
|
2769 |
+
"loss": 0.0869,
|
2770 |
+
"step": 920
|
2771 |
+
},
|
2772 |
+
{
|
2773 |
+
"epoch": 0.7381905524419535,
|
2774 |
+
"learning_rate": 1.989857570980049e-05,
|
2775 |
+
"loss": 0.0743,
|
2776 |
+
"step": 922
|
2777 |
+
},
|
2778 |
+
{
|
2779 |
+
"epoch": 0.7397918334667735,
|
2780 |
+
"learning_rate": 1.9890479722451292e-05,
|
2781 |
+
"loss": 0.3135,
|
2782 |
+
"step": 924
|
2783 |
+
},
|
2784 |
+
{
|
2785 |
+
"epoch": 0.7413931144915933,
|
2786 |
+
"learning_rate": 1.9882074674717832e-05,
|
2787 |
+
"loss": 1.2724,
|
2788 |
+
"step": 926
|
2789 |
+
},
|
2790 |
+
{
|
2791 |
+
"epoch": 0.7429943955164131,
|
2792 |
+
"learning_rate": 1.987336082924333e-05,
|
2793 |
+
"loss": 0.0381,
|
2794 |
+
"step": 928
|
2795 |
+
},
|
2796 |
+
{
|
2797 |
+
"epoch": 0.7445956765412329,
|
2798 |
+
"learning_rate": 1.986433845832037e-05,
|
2799 |
+
"loss": 0.6571,
|
2800 |
+
"step": 930
|
2801 |
+
},
|
2802 |
+
{
|
2803 |
+
"epoch": 0.7461969575660529,
|
2804 |
+
"learning_rate": 1.9855007843882437e-05,
|
2805 |
+
"loss": 0.1995,
|
2806 |
+
"step": 932
|
2807 |
+
},
|
2808 |
+
{
|
2809 |
+
"epoch": 0.7477982385908727,
|
2810 |
+
"learning_rate": 1.9845369277495105e-05,
|
2811 |
+
"loss": 0.1977,
|
2812 |
+
"step": 934
|
2813 |
+
},
|
2814 |
+
{
|
2815 |
+
"epoch": 0.7493995196156925,
|
2816 |
+
"learning_rate": 1.9835423060346892e-05,
|
2817 |
+
"loss": 0.3571,
|
2818 |
+
"step": 936
|
2819 |
+
},
|
2820 |
+
{
|
2821 |
+
"epoch": 0.7510008006405124,
|
2822 |
+
"learning_rate": 1.9825169503239885e-05,
|
2823 |
+
"loss": 0.1503,
|
2824 |
+
"step": 938
|
2825 |
+
},
|
2826 |
+
{
|
2827 |
+
"epoch": 0.7526020816653323,
|
2828 |
+
"learning_rate": 1.9814608926580007e-05,
|
2829 |
+
"loss": 0.2454,
|
2830 |
+
"step": 940
|
2831 |
+
},
|
2832 |
+
{
|
2833 |
+
"epoch": 0.7542033626901521,
|
2834 |
+
"learning_rate": 1.9803741660367018e-05,
|
2835 |
+
"loss": 0.2311,
|
2836 |
+
"step": 942
|
2837 |
+
},
|
2838 |
+
{
|
2839 |
+
"epoch": 0.755804643714972,
|
2840 |
+
"learning_rate": 1.979256804418418e-05,
|
2841 |
+
"loss": 0.0518,
|
2842 |
+
"step": 944
|
2843 |
+
},
|
2844 |
+
{
|
2845 |
+
"epoch": 0.7574059247397918,
|
2846 |
+
"learning_rate": 1.9781088427187677e-05,
|
2847 |
+
"loss": 0.3416,
|
2848 |
+
"step": 946
|
2849 |
+
},
|
2850 |
+
{
|
2851 |
+
"epoch": 0.7590072057646117,
|
2852 |
+
"learning_rate": 1.976930316809569e-05,
|
2853 |
+
"loss": 0.0772,
|
2854 |
+
"step": 948
|
2855 |
+
},
|
2856 |
+
{
|
2857 |
+
"epoch": 0.7606084867894315,
|
2858 |
+
"learning_rate": 1.9757212635177177e-05,
|
2859 |
+
"loss": 0.4751,
|
2860 |
+
"step": 950
|
2861 |
+
},
|
2862 |
+
{
|
2863 |
+
"epoch": 0.7622097678142514,
|
2864 |
+
"learning_rate": 1.9744817206240374e-05,
|
2865 |
+
"loss": 0.4526,
|
2866 |
+
"step": 952
|
2867 |
+
},
|
2868 |
+
{
|
2869 |
+
"epoch": 0.7638110488390712,
|
2870 |
+
"learning_rate": 1.9732117268621005e-05,
|
2871 |
+
"loss": 0.3096,
|
2872 |
+
"step": 954
|
2873 |
+
},
|
2874 |
+
{
|
2875 |
+
"epoch": 0.7654123298638911,
|
2876 |
+
"learning_rate": 1.9719113219170152e-05,
|
2877 |
+
"loss": 0.2254,
|
2878 |
+
"step": 956
|
2879 |
+
},
|
2880 |
+
{
|
2881 |
+
"epoch": 0.767013610888711,
|
2882 |
+
"learning_rate": 1.970580546424186e-05,
|
2883 |
+
"loss": 0.1723,
|
2884 |
+
"step": 958
|
2885 |
+
},
|
2886 |
+
{
|
2887 |
+
"epoch": 0.7686148919135308,
|
2888 |
+
"learning_rate": 1.9692194419680463e-05,
|
2889 |
+
"loss": 0.6533,
|
2890 |
+
"step": 960
|
2891 |
+
},
|
2892 |
+
{
|
2893 |
+
"epoch": 0.7702161729383507,
|
2894 |
+
"learning_rate": 1.9678280510807552e-05,
|
2895 |
+
"loss": 0.281,
|
2896 |
+
"step": 962
|
2897 |
+
},
|
2898 |
+
{
|
2899 |
+
"epoch": 0.7718174539631706,
|
2900 |
+
"learning_rate": 1.966406417240872e-05,
|
2901 |
+
"loss": 0.399,
|
2902 |
+
"step": 964
|
2903 |
+
},
|
2904 |
+
{
|
2905 |
+
"epoch": 0.7734187349879904,
|
2906 |
+
"learning_rate": 1.964954584871995e-05,
|
2907 |
+
"loss": 0.228,
|
2908 |
+
"step": 966
|
2909 |
+
},
|
2910 |
+
{
|
2911 |
+
"epoch": 0.7750200160128102,
|
2912 |
+
"learning_rate": 1.9634725993413744e-05,
|
2913 |
+
"loss": 0.5152,
|
2914 |
+
"step": 968
|
2915 |
+
},
|
2916 |
+
{
|
2917 |
+
"epoch": 0.77662129703763,
|
2918 |
+
"learning_rate": 1.9619605069584954e-05,
|
2919 |
+
"loss": 0.1713,
|
2920 |
+
"step": 970
|
2921 |
+
},
|
2922 |
+
{
|
2923 |
+
"epoch": 0.77822257806245,
|
2924 |
+
"learning_rate": 1.9604183549736287e-05,
|
2925 |
+
"loss": 0.2849,
|
2926 |
+
"step": 972
|
2927 |
+
},
|
2928 |
+
{
|
2929 |
+
"epoch": 0.7798238590872698,
|
2930 |
+
"learning_rate": 1.958846191576357e-05,
|
2931 |
+
"loss": 0.2035,
|
2932 |
+
"step": 974
|
2933 |
+
},
|
2934 |
+
{
|
2935 |
+
"epoch": 0.7814251401120896,
|
2936 |
+
"learning_rate": 1.9572440658940667e-05,
|
2937 |
+
"loss": 0.0795,
|
2938 |
+
"step": 976
|
2939 |
+
},
|
2940 |
+
{
|
2941 |
+
"epoch": 0.7830264211369096,
|
2942 |
+
"learning_rate": 1.955612027990415e-05,
|
2943 |
+
"loss": 0.177,
|
2944 |
+
"step": 978
|
2945 |
+
},
|
2946 |
+
{
|
2947 |
+
"epoch": 0.7846277021617294,
|
2948 |
+
"learning_rate": 1.953950128863763e-05,
|
2949 |
+
"loss": 0.2624,
|
2950 |
+
"step": 980
|
2951 |
+
},
|
2952 |
+
{
|
2953 |
+
"epoch": 0.7862289831865492,
|
2954 |
+
"learning_rate": 1.9522584204455835e-05,
|
2955 |
+
"loss": 0.2466,
|
2956 |
+
"step": 982
|
2957 |
+
},
|
2958 |
+
{
|
2959 |
+
"epoch": 0.7878302642113691,
|
2960 |
+
"learning_rate": 1.9505369555988395e-05,
|
2961 |
+
"loss": 0.1225,
|
2962 |
+
"step": 984
|
2963 |
+
},
|
2964 |
+
{
|
2965 |
+
"epoch": 0.789431545236189,
|
2966 |
+
"learning_rate": 1.9487857881163295e-05,
|
2967 |
+
"loss": 0.1407,
|
2968 |
+
"step": 986
|
2969 |
+
},
|
2970 |
+
{
|
2971 |
+
"epoch": 0.7910328262610088,
|
2972 |
+
"learning_rate": 1.947004972719008e-05,
|
2973 |
+
"loss": 0.5051,
|
2974 |
+
"step": 988
|
2975 |
+
},
|
2976 |
+
{
|
2977 |
+
"epoch": 0.7926341072858286,
|
2978 |
+
"learning_rate": 1.945194565054276e-05,
|
2979 |
+
"loss": 0.1115,
|
2980 |
+
"step": 990
|
2981 |
+
},
|
2982 |
+
{
|
2983 |
+
"epoch": 0.7942353883106485,
|
2984 |
+
"learning_rate": 1.9433546216942433e-05,
|
2985 |
+
"loss": 0.0047,
|
2986 |
+
"step": 992
|
2987 |
+
},
|
2988 |
+
{
|
2989 |
+
"epoch": 0.7958366693354684,
|
2990 |
+
"learning_rate": 1.941485200133955e-05,
|
2991 |
+
"loss": 0.4239,
|
2992 |
+
"step": 994
|
2993 |
+
},
|
2994 |
+
{
|
2995 |
+
"epoch": 0.7974379503602882,
|
2996 |
+
"learning_rate": 1.9395863587896025e-05,
|
2997 |
+
"loss": 0.4847,
|
2998 |
+
"step": 996
|
2999 |
+
},
|
3000 |
+
{
|
3001 |
+
"epoch": 0.7990392313851081,
|
3002 |
+
"learning_rate": 1.937658156996694e-05,
|
3003 |
+
"loss": 0.3656,
|
3004 |
+
"step": 998
|
3005 |
+
},
|
3006 |
+
{
|
3007 |
+
"epoch": 0.800640512409928,
|
3008 |
+
"learning_rate": 1.9357006550082e-05,
|
3009 |
+
"loss": 0.1793,
|
3010 |
+
"step": 1000
|
3011 |
+
},
|
3012 |
+
{
|
3013 |
+
"epoch": 0.8022417934347478,
|
3014 |
+
"learning_rate": 1.933713913992671e-05,
|
3015 |
+
"loss": 0.1453,
|
3016 |
+
"step": 1002
|
3017 |
+
},
|
3018 |
+
{
|
3019 |
+
"epoch": 0.8038430744595677,
|
3020 |
+
"learning_rate": 1.9316979960323283e-05,
|
3021 |
+
"loss": 0.7052,
|
3022 |
+
"step": 1004
|
3023 |
+
},
|
3024 |
+
{
|
3025 |
+
"epoch": 0.8054443554843875,
|
3026 |
+
"learning_rate": 1.9296529641211226e-05,
|
3027 |
+
"loss": 0.1244,
|
3028 |
+
"step": 1006
|
3029 |
+
},
|
3030 |
+
{
|
3031 |
+
"epoch": 0.8070456365092074,
|
3032 |
+
"learning_rate": 1.9275788821627607e-05,
|
3033 |
+
"loss": 0.374,
|
3034 |
+
"step": 1008
|
3035 |
+
},
|
3036 |
+
{
|
3037 |
+
"epoch": 0.8086469175340272,
|
3038 |
+
"learning_rate": 1.9254758149687187e-05,
|
3039 |
+
"loss": 0.3294,
|
3040 |
+
"step": 1010
|
3041 |
+
},
|
3042 |
+
{
|
3043 |
+
"epoch": 0.8102481985588471,
|
3044 |
+
"learning_rate": 1.9233438282562095e-05,
|
3045 |
+
"loss": 0.0984,
|
3046 |
+
"step": 1012
|
3047 |
+
},
|
3048 |
+
{
|
3049 |
+
"epoch": 0.8118494795836669,
|
3050 |
+
"learning_rate": 1.9211829886461278e-05,
|
3051 |
+
"loss": 0.3529,
|
3052 |
+
"step": 1014
|
3053 |
+
},
|
3054 |
+
{
|
3055 |
+
"epoch": 0.8134507606084868,
|
3056 |
+
"learning_rate": 1.918993363660975e-05,
|
3057 |
+
"loss": 0.1602,
|
3058 |
+
"step": 1016
|
3059 |
+
},
|
3060 |
+
{
|
3061 |
+
"epoch": 0.8150520416333067,
|
3062 |
+
"learning_rate": 1.916775021722745e-05,
|
3063 |
+
"loss": 0.1997,
|
3064 |
+
"step": 1018
|
3065 |
+
},
|
3066 |
+
{
|
3067 |
+
"epoch": 0.8166533226581265,
|
3068 |
+
"learning_rate": 1.9145280321507872e-05,
|
3069 |
+
"loss": 0.5403,
|
3070 |
+
"step": 1020
|
3071 |
+
},
|
3072 |
+
{
|
3073 |
+
"epoch": 0.8182546036829463,
|
3074 |
+
"learning_rate": 1.9122524651596372e-05,
|
3075 |
+
"loss": 0.0639,
|
3076 |
+
"step": 1022
|
3077 |
+
},
|
3078 |
+
{
|
3079 |
+
"epoch": 0.8198558847077662,
|
3080 |
+
"learning_rate": 1.9099483918568287e-05,
|
3081 |
+
"loss": 0.1201,
|
3082 |
+
"step": 1024
|
3083 |
+
},
|
3084 |
+
{
|
3085 |
+
"epoch": 0.8214571657325861,
|
3086 |
+
"learning_rate": 1.907615884240668e-05,
|
3087 |
+
"loss": 0.0731,
|
3088 |
+
"step": 1026
|
3089 |
+
},
|
3090 |
+
{
|
3091 |
+
"epoch": 0.8230584467574059,
|
3092 |
+
"learning_rate": 1.905255015197982e-05,
|
3093 |
+
"loss": 0.1547,
|
3094 |
+
"step": 1028
|
3095 |
+
},
|
3096 |
+
{
|
3097 |
+
"epoch": 0.8246597277822257,
|
3098 |
+
"learning_rate": 1.902865858501845e-05,
|
3099 |
+
"loss": 0.1453,
|
3100 |
+
"step": 1030
|
3101 |
+
},
|
3102 |
+
{
|
3103 |
+
"epoch": 0.8262610088070457,
|
3104 |
+
"learning_rate": 1.9004484888092734e-05,
|
3105 |
+
"loss": 0.3729,
|
3106 |
+
"step": 1032
|
3107 |
+
},
|
3108 |
+
{
|
3109 |
+
"epoch": 0.8278622898318655,
|
3110 |
+
"learning_rate": 1.8980029816588863e-05,
|
3111 |
+
"loss": 0.2768,
|
3112 |
+
"step": 1034
|
3113 |
+
},
|
3114 |
+
{
|
3115 |
+
"epoch": 0.8294635708566853,
|
3116 |
+
"learning_rate": 1.8955294134685528e-05,
|
3117 |
+
"loss": 0.205,
|
3118 |
+
"step": 1036
|
3119 |
+
},
|
3120 |
+
{
|
3121 |
+
"epoch": 0.8310648518815053,
|
3122 |
+
"learning_rate": 1.893027861533003e-05,
|
3123 |
+
"loss": 0.9087,
|
3124 |
+
"step": 1038
|
3125 |
+
},
|
3126 |
+
{
|
3127 |
+
"epoch": 0.8326661329063251,
|
3128 |
+
"learning_rate": 1.8904984040214043e-05,
|
3129 |
+
"loss": 0.1169,
|
3130 |
+
"step": 1040
|
3131 |
+
},
|
3132 |
+
{
|
3133 |
+
"epoch": 0.8342674139311449,
|
3134 |
+
"learning_rate": 1.8879411199749306e-05,
|
3135 |
+
"loss": 0.629,
|
3136 |
+
"step": 1042
|
3137 |
+
},
|
3138 |
+
{
|
3139 |
+
"epoch": 0.8358686949559647,
|
3140 |
+
"learning_rate": 1.885356089304285e-05,
|
3141 |
+
"loss": 0.323,
|
3142 |
+
"step": 1044
|
3143 |
+
},
|
3144 |
+
{
|
3145 |
+
"epoch": 0.8374699759807847,
|
3146 |
+
"learning_rate": 1.882743392787207e-05,
|
3147 |
+
"loss": 0.1265,
|
3148 |
+
"step": 1046
|
3149 |
+
},
|
3150 |
+
{
|
3151 |
+
"epoch": 0.8390712570056045,
|
3152 |
+
"learning_rate": 1.8801031120659396e-05,
|
3153 |
+
"loss": 0.0609,
|
3154 |
+
"step": 1048
|
3155 |
+
},
|
3156 |
+
{
|
3157 |
+
"epoch": 0.8406725380304243,
|
3158 |
+
"learning_rate": 1.877435329644691e-05,
|
3159 |
+
"loss": 0.2982,
|
3160 |
+
"step": 1050
|
3161 |
+
},
|
3162 |
+
{
|
3163 |
+
"epoch": 0.8422738190552442,
|
3164 |
+
"learning_rate": 1.8747401288870482e-05,
|
3165 |
+
"loss": 0.2404,
|
3166 |
+
"step": 1052
|
3167 |
+
},
|
3168 |
+
{
|
3169 |
+
"epoch": 0.8438751000800641,
|
3170 |
+
"learning_rate": 1.8720175940133712e-05,
|
3171 |
+
"loss": 0.2915,
|
3172 |
+
"step": 1054
|
3173 |
+
},
|
3174 |
+
{
|
3175 |
+
"epoch": 0.8454763811048839,
|
3176 |
+
"learning_rate": 1.8692678100981663e-05,
|
3177 |
+
"loss": 0.0284,
|
3178 |
+
"step": 1056
|
3179 |
+
},
|
3180 |
+
{
|
3181 |
+
"epoch": 0.8470776621297038,
|
3182 |
+
"learning_rate": 1.8664908630674264e-05,
|
3183 |
+
"loss": 0.4157,
|
3184 |
+
"step": 1058
|
3185 |
+
},
|
3186 |
+
{
|
3187 |
+
"epoch": 0.8486789431545236,
|
3188 |
+
"learning_rate": 1.8636868396959406e-05,
|
3189 |
+
"loss": 0.2116,
|
3190 |
+
"step": 1060
|
3191 |
+
},
|
3192 |
+
{
|
3193 |
+
"epoch": 0.8502802241793435,
|
3194 |
+
"learning_rate": 1.8608558276045898e-05,
|
3195 |
+
"loss": 0.1604,
|
3196 |
+
"step": 1062
|
3197 |
+
},
|
3198 |
+
{
|
3199 |
+
"epoch": 0.8518815052041633,
|
3200 |
+
"learning_rate": 1.8579979152576076e-05,
|
3201 |
+
"loss": 0.196,
|
3202 |
+
"step": 1064
|
3203 |
+
},
|
3204 |
+
{
|
3205 |
+
"epoch": 0.8534827862289832,
|
3206 |
+
"learning_rate": 1.8551131919598084e-05,
|
3207 |
+
"loss": 0.1915,
|
3208 |
+
"step": 1066
|
3209 |
+
},
|
3210 |
+
{
|
3211 |
+
"epoch": 0.855084067253803,
|
3212 |
+
"learning_rate": 1.852201747853807e-05,
|
3213 |
+
"loss": 0.2589,
|
3214 |
+
"step": 1068
|
3215 |
+
},
|
3216 |
+
{
|
3217 |
+
"epoch": 0.8566853482786229,
|
3218 |
+
"learning_rate": 1.849263673917196e-05,
|
3219 |
+
"loss": 0.3002,
|
3220 |
+
"step": 1070
|
3221 |
+
},
|
3222 |
+
{
|
3223 |
+
"epoch": 0.8582866293034428,
|
3224 |
+
"learning_rate": 1.846299061959706e-05,
|
3225 |
+
"loss": 0.1926,
|
3226 |
+
"step": 1072
|
3227 |
+
},
|
3228 |
+
{
|
3229 |
+
"epoch": 0.8598879103282626,
|
3230 |
+
"learning_rate": 1.8433080046203293e-05,
|
3231 |
+
"loss": 0.4974,
|
3232 |
+
"step": 1074
|
3233 |
+
},
|
3234 |
+
{
|
3235 |
+
"epoch": 0.8614891913530824,
|
3236 |
+
"learning_rate": 1.8402905953644356e-05,
|
3237 |
+
"loss": 0.3675,
|
3238 |
+
"step": 1076
|
3239 |
+
},
|
3240 |
+
{
|
3241 |
+
"epoch": 0.8630904723779024,
|
3242 |
+
"learning_rate": 1.837246928480848e-05,
|
3243 |
+
"loss": 0.1309,
|
3244 |
+
"step": 1078
|
3245 |
+
},
|
3246 |
+
{
|
3247 |
+
"epoch": 0.8646917534027222,
|
3248 |
+
"learning_rate": 1.8341770990788874e-05,
|
3249 |
+
"loss": 0.1405,
|
3250 |
+
"step": 1080
|
3251 |
+
},
|
3252 |
+
{
|
3253 |
+
"epoch": 0.866293034427542,
|
3254 |
+
"learning_rate": 1.831081203085415e-05,
|
3255 |
+
"loss": 0.088,
|
3256 |
+
"step": 1082
|
3257 |
+
},
|
3258 |
+
{
|
3259 |
+
"epoch": 0.8678943154523618,
|
3260 |
+
"learning_rate": 1.8279593372418284e-05,
|
3261 |
+
"loss": 0.3894,
|
3262 |
+
"step": 1084
|
3263 |
+
},
|
3264 |
+
{
|
3265 |
+
"epoch": 0.8694955964771818,
|
3266 |
+
"learning_rate": 1.8248115991010303e-05,
|
3267 |
+
"loss": 0.1894,
|
3268 |
+
"step": 1086
|
3269 |
+
},
|
3270 |
+
{
|
3271 |
+
"epoch": 0.8710968775020016,
|
3272 |
+
"learning_rate": 1.8216380870243963e-05,
|
3273 |
+
"loss": 0.1194,
|
3274 |
+
"step": 1088
|
3275 |
+
},
|
3276 |
+
{
|
3277 |
+
"epoch": 0.8726981585268214,
|
3278 |
+
"learning_rate": 1.8184389001786912e-05,
|
3279 |
+
"loss": 0.4904,
|
3280 |
+
"step": 1090
|
3281 |
+
},
|
3282 |
+
{
|
3283 |
+
"epoch": 0.8742994395516414,
|
3284 |
+
"learning_rate": 1.815214138532966e-05,
|
3285 |
+
"loss": 0.3006,
|
3286 |
+
"step": 1092
|
3287 |
+
},
|
3288 |
+
{
|
3289 |
+
"epoch": 0.8759007205764612,
|
3290 |
+
"learning_rate": 1.8119639028554475e-05,
|
3291 |
+
"loss": 0.0661,
|
3292 |
+
"step": 1094
|
3293 |
+
},
|
3294 |
+
{
|
3295 |
+
"epoch": 0.877502001601281,
|
3296 |
+
"learning_rate": 1.808688294710378e-05,
|
3297 |
+
"loss": 0.4522,
|
3298 |
+
"step": 1096
|
3299 |
+
},
|
3300 |
+
{
|
3301 |
+
"epoch": 0.8791032826261009,
|
3302 |
+
"learning_rate": 1.805387416454849e-05,
|
3303 |
+
"loss": 1.0015,
|
3304 |
+
"step": 1098
|
3305 |
+
},
|
3306 |
+
{
|
3307 |
+
"epoch": 0.8807045636509208,
|
3308 |
+
"learning_rate": 1.802061371235592e-05,
|
3309 |
+
"loss": 0.0829,
|
3310 |
+
"step": 1100
|
3311 |
+
},
|
3312 |
+
{
|
3313 |
+
"epoch": 0.8823058446757406,
|
3314 |
+
"learning_rate": 1.7987102629857692e-05,
|
3315 |
+
"loss": 0.0227,
|
3316 |
+
"step": 1102
|
3317 |
+
},
|
3318 |
+
{
|
3319 |
+
"epoch": 0.8839071257005604,
|
3320 |
+
"learning_rate": 1.7953341964217196e-05,
|
3321 |
+
"loss": 0.0373,
|
3322 |
+
"step": 1104
|
3323 |
+
},
|
3324 |
+
{
|
3325 |
+
"epoch": 0.8855084067253803,
|
3326 |
+
"learning_rate": 1.7919332770396798e-05,
|
3327 |
+
"loss": 0.3724,
|
3328 |
+
"step": 1106
|
3329 |
+
},
|
3330 |
+
{
|
3331 |
+
"epoch": 0.8871096877502002,
|
3332 |
+
"learning_rate": 1.7885076111125e-05,
|
3333 |
+
"loss": 0.0994,
|
3334 |
+
"step": 1108
|
3335 |
+
},
|
3336 |
+
{
|
3337 |
+
"epoch": 0.88871096877502,
|
3338 |
+
"learning_rate": 1.7850573056863173e-05,
|
3339 |
+
"loss": 0.7036,
|
3340 |
+
"step": 1110
|
3341 |
+
},
|
3342 |
+
{
|
3343 |
+
"epoch": 0.8903122497998399,
|
3344 |
+
"learning_rate": 1.7815824685772042e-05,
|
3345 |
+
"loss": 0.1373,
|
3346 |
+
"step": 1112
|
3347 |
+
},
|
3348 |
+
{
|
3349 |
+
"epoch": 0.8919135308246597,
|
3350 |
+
"learning_rate": 1.7780832083678122e-05,
|
3351 |
+
"loss": 0.1676,
|
3352 |
+
"step": 1114
|
3353 |
+
},
|
3354 |
+
{
|
3355 |
+
"epoch": 0.8935148118494796,
|
3356 |
+
"learning_rate": 1.774559634403971e-05,
|
3357 |
+
"loss": 0.6738,
|
3358 |
+
"step": 1116
|
3359 |
+
},
|
3360 |
+
{
|
3361 |
+
"epoch": 0.8951160928742994,
|
3362 |
+
"learning_rate": 1.7710118567912732e-05,
|
3363 |
+
"loss": 0.1152,
|
3364 |
+
"step": 1118
|
3365 |
+
},
|
3366 |
+
{
|
3367 |
+
"epoch": 0.8967173738991193,
|
3368 |
+
"learning_rate": 1.7674399863916298e-05,
|
3369 |
+
"loss": 0.265,
|
3370 |
+
"step": 1120
|
3371 |
+
},
|
3372 |
+
{
|
3373 |
+
"epoch": 0.8983186549239391,
|
3374 |
+
"learning_rate": 1.7638441348198144e-05,
|
3375 |
+
"loss": 0.3219,
|
3376 |
+
"step": 1122
|
3377 |
+
},
|
3378 |
+
{
|
3379 |
+
"epoch": 0.899919935948759,
|
3380 |
+
"learning_rate": 1.7602244144399713e-05,
|
3381 |
+
"loss": 0.7533,
|
3382 |
+
"step": 1124
|
3383 |
+
},
|
3384 |
+
{
|
3385 |
+
"epoch": 0.9015212169735789,
|
3386 |
+
"learning_rate": 1.7565809383620966e-05,
|
3387 |
+
"loss": 0.3681,
|
3388 |
+
"step": 1126
|
3389 |
+
},
|
3390 |
+
{
|
3391 |
+
"epoch": 0.9031224979983987,
|
3392 |
+
"learning_rate": 1.7529138204385186e-05,
|
3393 |
+
"loss": 0.438,
|
3394 |
+
"step": 1128
|
3395 |
+
},
|
3396 |
+
{
|
3397 |
+
"epoch": 0.9047237790232185,
|
3398 |
+
"learning_rate": 1.7492231752603305e-05,
|
3399 |
+
"loss": 0.1269,
|
3400 |
+
"step": 1130
|
3401 |
+
},
|
3402 |
+
{
|
3403 |
+
"epoch": 0.9063250600480385,
|
3404 |
+
"learning_rate": 1.7455091181538094e-05,
|
3405 |
+
"loss": 0.246,
|
3406 |
+
"step": 1132
|
3407 |
+
},
|
3408 |
+
{
|
3409 |
+
"epoch": 0.9079263410728583,
|
3410 |
+
"learning_rate": 1.741771765176815e-05,
|
3411 |
+
"loss": 0.3602,
|
3412 |
+
"step": 1134
|
3413 |
+
},
|
3414 |
+
{
|
3415 |
+
"epoch": 0.9095276220976781,
|
3416 |
+
"learning_rate": 1.7380112331151657e-05,
|
3417 |
+
"loss": 0.3644,
|
3418 |
+
"step": 1136
|
3419 |
+
},
|
3420 |
+
{
|
3421 |
+
"epoch": 0.911128903122498,
|
3422 |
+
"learning_rate": 1.7342276394789825e-05,
|
3423 |
+
"loss": 0.1733,
|
3424 |
+
"step": 1138
|
3425 |
+
},
|
3426 |
+
{
|
3427 |
+
"epoch": 0.9127301841473179,
|
3428 |
+
"learning_rate": 1.7304211024990216e-05,
|
3429 |
+
"loss": 0.3662,
|
3430 |
+
"step": 1140
|
3431 |
+
},
|
3432 |
+
{
|
3433 |
+
"epoch": 0.9143314651721377,
|
3434 |
+
"learning_rate": 1.7265917411229803e-05,
|
3435 |
+
"loss": 0.2163,
|
3436 |
+
"step": 1142
|
3437 |
+
},
|
3438 |
+
{
|
3439 |
+
"epoch": 0.9159327461969575,
|
3440 |
+
"learning_rate": 1.7227396750117802e-05,
|
3441 |
+
"loss": 0.0666,
|
3442 |
+
"step": 1144
|
3443 |
+
},
|
3444 |
+
{
|
3445 |
+
"epoch": 0.9175340272217775,
|
3446 |
+
"learning_rate": 1.718865024535822e-05,
|
3447 |
+
"loss": 0.5219,
|
3448 |
+
"step": 1146
|
3449 |
+
},
|
3450 |
+
{
|
3451 |
+
"epoch": 0.9191353082465973,
|
3452 |
+
"learning_rate": 1.7149679107712317e-05,
|
3453 |
+
"loss": 0.459,
|
3454 |
+
"step": 1148
|
3455 |
+
},
|
3456 |
+
{
|
3457 |
+
"epoch": 0.9207365892714171,
|
3458 |
+
"learning_rate": 1.711048455496075e-05,
|
3459 |
+
"loss": 0.4002,
|
3460 |
+
"step": 1150
|
3461 |
+
},
|
3462 |
+
{
|
3463 |
+
"epoch": 0.922337870296237,
|
3464 |
+
"learning_rate": 1.7071067811865474e-05,
|
3465 |
+
"loss": 0.2244,
|
3466 |
+
"step": 1152
|
3467 |
+
},
|
3468 |
+
{
|
3469 |
+
"epoch": 0.9239391513210569,
|
3470 |
+
"learning_rate": 1.7031430110131566e-05,
|
3471 |
+
"loss": 0.3854,
|
3472 |
+
"step": 1154
|
3473 |
+
},
|
3474 |
+
{
|
3475 |
+
"epoch": 0.9255404323458767,
|
3476 |
+
"learning_rate": 1.699157268836863e-05,
|
3477 |
+
"loss": 0.1912,
|
3478 |
+
"step": 1156
|
3479 |
+
},
|
3480 |
+
{
|
3481 |
+
"epoch": 0.9271417133706965,
|
3482 |
+
"learning_rate": 1.6951496792052148e-05,
|
3483 |
+
"loss": 0.3037,
|
3484 |
+
"step": 1158
|
3485 |
+
},
|
3486 |
+
{
|
3487 |
+
"epoch": 0.9287429943955164,
|
3488 |
+
"learning_rate": 1.6911203673484583e-05,
|
3489 |
+
"loss": 0.3624,
|
3490 |
+
"step": 1160
|
3491 |
+
},
|
3492 |
+
{
|
3493 |
+
"epoch": 0.9303442754203363,
|
3494 |
+
"learning_rate": 1.687069459175619e-05,
|
3495 |
+
"loss": 0.2524,
|
3496 |
+
"step": 1162
|
3497 |
+
},
|
3498 |
+
{
|
3499 |
+
"epoch": 0.9319455564451561,
|
3500 |
+
"learning_rate": 1.682997081270568e-05,
|
3501 |
+
"loss": 0.2356,
|
3502 |
+
"step": 1164
|
3503 |
+
},
|
3504 |
+
{
|
3505 |
+
"epoch": 0.933546837469976,
|
3506 |
+
"learning_rate": 1.6789033608880742e-05,
|
3507 |
+
"loss": 0.4152,
|
3508 |
+
"step": 1166
|
3509 |
+
},
|
3510 |
+
{
|
3511 |
+
"epoch": 0.9351481184947958,
|
3512 |
+
"learning_rate": 1.6747884259498185e-05,
|
3513 |
+
"loss": 0.3348,
|
3514 |
+
"step": 1168
|
3515 |
+
},
|
3516 |
+
{
|
3517 |
+
"epoch": 0.9367493995196157,
|
3518 |
+
"learning_rate": 1.6706524050404006e-05,
|
3519 |
+
"loss": 0.2385,
|
3520 |
+
"step": 1170
|
3521 |
+
},
|
3522 |
+
{
|
3523 |
+
"epoch": 0.9383506805444356,
|
3524 |
+
"learning_rate": 1.6664954274033175e-05,
|
3525 |
+
"loss": 0.2656,
|
3526 |
+
"step": 1172
|
3527 |
+
},
|
3528 |
+
{
|
3529 |
+
"epoch": 0.9399519615692554,
|
3530 |
+
"learning_rate": 1.662317622936933e-05,
|
3531 |
+
"loss": 0.1864,
|
3532 |
+
"step": 1174
|
3533 |
+
},
|
3534 |
+
{
|
3535 |
+
"epoch": 0.9415532425940752,
|
3536 |
+
"learning_rate": 1.6581191221904098e-05,
|
3537 |
+
"loss": 0.0389,
|
3538 |
+
"step": 1176
|
3539 |
+
},
|
3540 |
+
{
|
3541 |
+
"epoch": 0.9431545236188951,
|
3542 |
+
"learning_rate": 1.6539000563596328e-05,
|
3543 |
+
"loss": 0.3571,
|
3544 |
+
"step": 1178
|
3545 |
+
},
|
3546 |
+
{
|
3547 |
+
"epoch": 0.944755804643715,
|
3548 |
+
"learning_rate": 1.6496605572831127e-05,
|
3549 |
+
"loss": 0.2997,
|
3550 |
+
"step": 1180
|
3551 |
+
},
|
3552 |
+
{
|
3553 |
+
"epoch": 0.9463570856685348,
|
3554 |
+
"learning_rate": 1.6454007574378657e-05,
|
3555 |
+
"loss": 0.0873,
|
3556 |
+
"step": 1182
|
3557 |
+
},
|
3558 |
+
{
|
3559 |
+
"epoch": 0.9479583666933546,
|
3560 |
+
"learning_rate": 1.6411207899352633e-05,
|
3561 |
+
"loss": 0.3684,
|
3562 |
+
"step": 1184
|
3563 |
+
},
|
3564 |
+
{
|
3565 |
+
"epoch": 0.9495596477181746,
|
3566 |
+
"learning_rate": 1.6368207885168904e-05,
|
3567 |
+
"loss": 0.8584,
|
3568 |
+
"step": 1186
|
3569 |
+
},
|
3570 |
+
{
|
3571 |
+
"epoch": 0.9511609287429944,
|
3572 |
+
"learning_rate": 1.6325008875503563e-05,
|
3573 |
+
"loss": 0.0825,
|
3574 |
+
"step": 1188
|
3575 |
+
},
|
3576 |
+
{
|
3577 |
+
"epoch": 0.9527622097678142,
|
3578 |
+
"learning_rate": 1.628161222025089e-05,
|
3579 |
+
"loss": 0.3448,
|
3580 |
+
"step": 1190
|
3581 |
+
},
|
3582 |
+
{
|
3583 |
+
"epoch": 0.9543634907926342,
|
3584 |
+
"learning_rate": 1.623801927548132e-05,
|
3585 |
+
"loss": 0.0754,
|
3586 |
+
"step": 1192
|
3587 |
+
},
|
3588 |
+
{
|
3589 |
+
"epoch": 0.955964771817454,
|
3590 |
+
"learning_rate": 1.6194231403398987e-05,
|
3591 |
+
"loss": 0.2999,
|
3592 |
+
"step": 1194
|
3593 |
+
},
|
3594 |
+
{
|
3595 |
+
"epoch": 0.9575660528422738,
|
3596 |
+
"learning_rate": 1.6150249972299173e-05,
|
3597 |
+
"loss": 0.0235,
|
3598 |
+
"step": 1196
|
3599 |
+
},
|
3600 |
+
{
|
3601 |
+
"epoch": 0.9591673338670936,
|
3602 |
+
"learning_rate": 1.6106076356525484e-05,
|
3603 |
+
"loss": 0.3246,
|
3604 |
+
"step": 1198
|
3605 |
+
},
|
3606 |
+
{
|
3607 |
+
"epoch": 0.9607686148919136,
|
3608 |
+
"learning_rate": 1.6061711936427028e-05,
|
3609 |
+
"loss": 0.3452,
|
3610 |
+
"step": 1200
|
3611 |
+
},
|
3612 |
+
{
|
3613 |
+
"epoch": 0.9623698959167334,
|
3614 |
+
"learning_rate": 1.6017158098315224e-05,
|
3615 |
+
"loss": 0.1659,
|
3616 |
+
"step": 1202
|
3617 |
+
},
|
3618 |
+
{
|
3619 |
+
"epoch": 0.9639711769415532,
|
3620 |
+
"learning_rate": 1.5972416234420404e-05,
|
3621 |
+
"loss": 0.1711,
|
3622 |
+
"step": 1204
|
3623 |
+
},
|
3624 |
+
{
|
3625 |
+
"epoch": 0.9655724579663731,
|
3626 |
+
"learning_rate": 1.592748774284844e-05,
|
3627 |
+
"loss": 0.2315,
|
3628 |
+
"step": 1206
|
3629 |
+
},
|
3630 |
+
{
|
3631 |
+
"epoch": 0.967173738991193,
|
3632 |
+
"learning_rate": 1.588237402753703e-05,
|
3633 |
+
"loss": 0.3712,
|
3634 |
+
"step": 1208
|
3635 |
+
},
|
3636 |
+
{
|
3637 |
+
"epoch": 0.9687750200160128,
|
3638 |
+
"learning_rate": 1.5837076498211673e-05,
|
3639 |
+
"loss": 0.2155,
|
3640 |
+
"step": 1210
|
3641 |
+
},
|
3642 |
+
{
|
3643 |
+
"epoch": 0.9703763010408326,
|
3644 |
+
"learning_rate": 1.579159657034185e-05,
|
3645 |
+
"loss": 0.1708,
|
3646 |
+
"step": 1212
|
3647 |
+
},
|
3648 |
+
{
|
3649 |
+
"epoch": 0.9719775820656525,
|
3650 |
+
"learning_rate": 1.574593566509664e-05,
|
3651 |
+
"loss": 0.218,
|
3652 |
+
"step": 1214
|
3653 |
+
},
|
3654 |
+
{
|
3655 |
+
"epoch": 0.9735788630904724,
|
3656 |
+
"learning_rate": 1.5700095209300386e-05,
|
3657 |
+
"loss": 0.3061,
|
3658 |
+
"step": 1216
|
3659 |
+
},
|
3660 |
+
{
|
3661 |
+
"epoch": 0.9751801441152922,
|
3662 |
+
"learning_rate": 1.5654076635387976e-05,
|
3663 |
+
"loss": 0.3219,
|
3664 |
+
"step": 1218
|
3665 |
+
},
|
3666 |
+
{
|
3667 |
+
"epoch": 0.9767814251401121,
|
3668 |
+
"learning_rate": 1.560788138136029e-05,
|
3669 |
+
"loss": 0.2066,
|
3670 |
+
"step": 1220
|
3671 |
+
},
|
3672 |
+
{
|
3673 |
+
"epoch": 0.978382706164932,
|
3674 |
+
"learning_rate": 1.5561510890739137e-05,
|
3675 |
+
"loss": 0.0483,
|
3676 |
+
"step": 1222
|
3677 |
+
},
|
3678 |
+
{
|
3679 |
+
"epoch": 0.9799839871897518,
|
3680 |
+
"learning_rate": 1.5514966612522088e-05,
|
3681 |
+
"loss": 0.338,
|
3682 |
+
"step": 1224
|
3683 |
+
},
|
3684 |
+
{
|
3685 |
+
"epoch": 0.9815852682145717,
|
3686 |
+
"learning_rate": 1.546825000113736e-05,
|
3687 |
+
"loss": 0.1178,
|
3688 |
+
"step": 1226
|
3689 |
+
},
|
3690 |
+
{
|
3691 |
+
"epoch": 0.9831865492393915,
|
3692 |
+
"learning_rate": 1.5421362516398285e-05,
|
3693 |
+
"loss": 0.0207,
|
3694 |
+
"step": 1228
|
3695 |
+
},
|
3696 |
+
{
|
3697 |
+
"epoch": 0.9847878302642114,
|
3698 |
+
"learning_rate": 1.5374305623457605e-05,
|
3699 |
+
"loss": 0.0918,
|
3700 |
+
"step": 1230
|
3701 |
+
},
|
3702 |
+
{
|
3703 |
+
"epoch": 0.9863891112890312,
|
3704 |
+
"learning_rate": 1.532708079276186e-05,
|
3705 |
+
"loss": 0.4725,
|
3706 |
+
"step": 1232
|
3707 |
+
},
|
3708 |
+
{
|
3709 |
+
"epoch": 0.9879903923138511,
|
3710 |
+
"learning_rate": 1.5279689500005353e-05,
|
3711 |
+
"loss": 0.1675,
|
3712 |
+
"step": 1234
|
3713 |
+
},
|
3714 |
+
{
|
3715 |
+
"epoch": 0.9895916733386709,
|
3716 |
+
"learning_rate": 1.5232133226083962e-05,
|
3717 |
+
"loss": 0.0769,
|
3718 |
+
"step": 1236
|
3719 |
+
},
|
3720 |
+
{
|
3721 |
+
"epoch": 0.9911929543634908,
|
3722 |
+
"learning_rate": 1.5184413457049014e-05,
|
3723 |
+
"loss": 0.1771,
|
3724 |
+
"step": 1238
|
3725 |
+
},
|
3726 |
+
{
|
3727 |
+
"epoch": 0.9927942353883107,
|
3728 |
+
"learning_rate": 1.5136531684060753e-05,
|
3729 |
+
"loss": 0.1841,
|
3730 |
+
"step": 1240
|
3731 |
+
},
|
3732 |
+
{
|
3733 |
+
"epoch": 0.9943955164131305,
|
3734 |
+
"learning_rate": 1.50884894033418e-05,
|
3735 |
+
"loss": 0.1286,
|
3736 |
+
"step": 1242
|
3737 |
+
},
|
3738 |
+
{
|
3739 |
+
"epoch": 0.9959967974379503,
|
3740 |
+
"learning_rate": 1.504028811613027e-05,
|
3741 |
+
"loss": 0.1521,
|
3742 |
+
"step": 1244
|
3743 |
+
},
|
3744 |
+
{
|
3745 |
+
"epoch": 0.9975980784627703,
|
3746 |
+
"learning_rate": 1.4991929328633043e-05,
|
3747 |
+
"loss": 0.0993,
|
3748 |
+
"step": 1246
|
3749 |
+
},
|
3750 |
+
{
|
3751 |
+
"epoch": 0.9991993594875901,
|
3752 |
+
"learning_rate": 1.4943414551978622e-05,
|
3753 |
+
"loss": 1.0781,
|
3754 |
+
"step": 1248
|
3755 |
+
},
|
3756 |
+
{
|
3757 |
+
"epoch": 1.0,
|
3758 |
+
"step": 1249,
|
3759 |
+
"total_flos": 7716239844573184.0,
|
3760 |
+
"train_loss": 0.25484217935285286,
|
3761 |
+
"train_runtime": 2013.0633,
|
3762 |
+
"train_samples_per_second": 9.927,
|
3763 |
+
"train_steps_per_second": 0.62
|
3764 |
+
}
|
3765 |
+
],
|
3766 |
+
"logging_steps": 2,
|
3767 |
+
"max_steps": 1249,
|
3768 |
+
"num_input_tokens_seen": 0,
|
3769 |
+
"num_train_epochs": 1,
|
3770 |
+
"save_steps": 500,
|
3771 |
+
"stateful_callbacks": {},
|
3772 |
+
"total_flos": 7716239844573184.0,
|
3773 |
+
"train_batch_size": 1,
|
3774 |
+
"trial_name": null,
|
3775 |
+
"trial_params": null
|
3776 |
+
}
|
client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_0625_seed1/server_model_round0.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:869e2e62bc69ef9c3492bbcedbd2ab2239d0c34a87e1be44dea1be9a5549a443
|
3 |
+
size 639793378
|
client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_0625_seed1/server_model_round1.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8dbd491d2a6d2cfdd487bbd2499573b31c5a4366c7ea2b890d67aab5efdcca87
|
3 |
+
size 639793378
|
client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_0625_seed1/server_model_round2.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8369f289e19c0d10b1b825b4ded821e564f892cacc4eef187078f79bc56b740d
|
3 |
+
size 639793378
|
client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_sft_scenario12_new_10000_random0_0625_seed1/server_model_round3.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a0ef5c39ebe1bb987f228f5bccd58021a45d7193f451c8f3bf7de1fc398d6e5c
|
3 |
+
size 639793378
|