Diplomovka/hruska_experiments/hruska_experiment_summary.csv
2026-09-30 21:22:53 +02:00

12 lines
562 B
Plaintext

rank,experiment,eval_loss,train_loss,train_runtime
1,10_r16_a32_lr1e4,0.6918,0.8612,7:15:07.38
2,06_lr1e4_r16_a16,0.7184,0.889,7:17:30.28
3,04_rank32_a32_lr5e5,0.7292,0.9003,7:22:32.83
4,05_r16_alpha32_lr5e5,0.7323,0.9026,7:17:43.75
5,02_optimizer_adamw8bit_r16_a16_lr5e5,0.7521,0.9248,7:20:23.21
6,09_batch2_accum4_r16_a16_lr5e5,0.7563,0.929,8:24:57.09
7,01_baseline_r16_a16_lr5e5_adamw,0.7567,0.929,7:18:05.66
8,03_rank8_a16_lr5e5,0.7586,0.9308,7:16:23.07
9,07_lr2e5_r16_a16,0.7912,0.9674,7:18:24.50
10,08_attention_only_r16_a16_lr5e5,0.8094,0.9808,6:51:24.56