output/pretrain_memory/output_pretraining_llama_1b_c4_batch-size-1_adamw+lomo_2025-04-21_11-22-09.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,0.1949940818361938,8.52999936,9.382658048,18.956207275390625,17.633771896362305,5.8321003913879395

output/pretrain_memory/output_pretraining_llama_1b_c4_batch-size-1_adamw+lora+lomo_2025-04-21_11-35-11.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,0.19572750213555992,4.304766464,4.36207616,9.8455228805542,9.900449752807617,2.6913557052612305

output/pretrain_memory/output_pretraining_llama_1b_c4_batch-size-1_adamw+lora_2025-04-21_11-35-18.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,0.16241615107282997,4.79674112,4.869586944,10.328532218933105,9.720293998718262,2.643303394317627

output/pretrain_memory/output_pretraining_llama_1b_c4_batch-size-1_adamw_2025-04-21_11-35-13.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,0.1519155834801495,13.70144512,14.891876352,15.858454704284668,13.794758796691895,3.9728190898895264

output/pretrain_memory/output_pretraining_llama_1b_c4_batch-size-1_galore+lomo_2025-04-21_11-43-06.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,7.01321501750499,5.338831872,5.528092672,17.581697463989258,13.727916717529297,3.946352005004883

output/pretrain_memory/output_pretraining_llama_1b_c4_batch-size-1_galore+lora+lomo_2025-04-21_11-44-13.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,0.16601467011496424,4.304766464,4.36207616,9.690999031066895,9.362051963806152,2.5502851009368896

output/pretrain_memory/output_pretraining_llama_1b_c4_batch-size-1_galore+lora_2025-04-21_11-44-51.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,0.17468192339874805,4.41575168,4.471128064,10.142722129821777,9.87549877166748,2.6846487522125244

output/pretrain_memory/output_pretraining_llama_1b_c4_batch-size-1_galore8bit+lomo_2025-04-21_12-10-32.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,7.09518618634902,4.217449984,4.5613056,12.897409439086914,19.605955123901367,7.103556156158447

output/pretrain_memory/output_pretraining_llama_1b_c4_batch-size-1_galore8bit+lora+lomo_2025-04-21_11-54-08.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,0.17671572077088057,3.70452224,3.75390208,10.959427833557129,16.56402015686035,5.240421772003174

output/pretrain_memory/output_pretraining_llama_1b_c4_batch-size-1_galore8bit+lora_2025-04-21_11-54-46.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,0.17641670098528267,3.81491968,3.85875968,9.408828735351562,9.908549308776855,2.6935362815856934

output/pretrain_memory/output_pretraining_llama_1b_c4_batch-size-1_galore8bit_2025-04-21_12-19-49.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,7.148737526452169,6.77276416,7.595884544,13.972885131835938,12.764690399169922,3.5839626789093018

output/pretrain_memory/output_pretraining_llama_1b_c4_batch-size-1_galore_2025-04-21_11-53-14.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,6.961088717542589,8.009071104,8.269070336,14.665997505187988,14.047356605529785,4.07444953918457

output/pretrain_memory/output_pretraining_llama_350m_c4_batch-size-1_adamw+lomo_2025-04-19_10-48-25.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,0.17898882757872342,2.410153472,2.678063104,14.150588035583496,13.938347816467285,4.030276298522949

output/pretrain_memory/output_pretraining_llama_350m_c4_batch-size-1_adamw+lora+lomo_2025-04-19_11-58-23.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,0.18357930895872415,1.238236672,1.361051648,10.421347618103027,10.199911117553711,2.773170232772827

output/pretrain_memory/output_pretraining_llama_350m_c4_batch-size-1_adamw+lora_2025-04-19_10-19-44.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,0.1526119064539671,1.290989056,1.503657984,9.291796684265137,9.653308868408203,2.6256563663482666

output/pretrain_memory/output_pretraining_llama_350m_c4_batch-size-1_adamw_2025-04-19_09-23-59.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,0.13244848777540028,3.74104832,4.0370176,10.23366641998291,9.12821102142334,2.4913408756256104

output/pretrain_memory/output_pretraining_llama_350m_c4_batch-size-1_galore+lomo_2025-04-19_11-03-57.csv

+epoch,training_step,compute_time,peak_memory_usage_allocated_GB,peak_memory_usage_reserved_GB,training_loss,evaluation_loss,evaluation_metric
+1,10,1.792762536276132,1.658086912,1.843396608,12.309712409973145,11.6229829788208,3.1972732543945312