imdatta0's picture
End of training
8b1b653 verified
raw
history blame
7.69 kB
{
"best_metric": 1.2975215911865234,
"best_model_checkpoint": "/home/datta0/models/lora_final/gemma-2-9b_magiccoder_ortho/checkpoint-4",
"epoch": 0.3135653771562723,
"eval_steps": 4,
"global_step": 48,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.006532612024089007,
"grad_norm": 5.831324577331543,
"learning_rate": 2.5e-05,
"loss": 1.3342,
"step": 1
},
{
"epoch": 0.013065224048178014,
"grad_norm": 5.2910685539245605,
"learning_rate": 5e-05,
"loss": 1.2199,
"step": 2
},
{
"epoch": 0.026130448096356028,
"grad_norm": 3.488215684890747,
"learning_rate": 0.0001,
"loss": 1.2682,
"step": 4
},
{
"epoch": 0.026130448096356028,
"eval_loss": 1.2975215911865234,
"eval_runtime": 866.0894,
"eval_samples_per_second": 0.565,
"eval_steps_per_second": 0.565,
"step": 4
},
{
"epoch": 0.03919567214453404,
"grad_norm": 3.667466163635254,
"learning_rate": 9.995555091232516e-05,
"loss": 1.3348,
"step": 6
},
{
"epoch": 0.052260896192712056,
"grad_norm": 8.944449424743652,
"learning_rate": 9.982228267815643e-05,
"loss": 1.6701,
"step": 8
},
{
"epoch": 0.052260896192712056,
"eval_loss": 4.235434532165527,
"eval_runtime": 252.6713,
"eval_samples_per_second": 1.935,
"eval_steps_per_second": 1.935,
"step": 8
},
{
"epoch": 0.06532612024089007,
"grad_norm": 8.121936798095703,
"learning_rate": 9.96004322435508e-05,
"loss": 5.8059,
"step": 10
},
{
"epoch": 0.07839134428906808,
"grad_norm": 1.232266902923584,
"learning_rate": 9.929039405048501e-05,
"loss": 10.2087,
"step": 12
},
{
"epoch": 0.07839134428906808,
"eval_loss": 11.443126678466797,
"eval_runtime": 295.202,
"eval_samples_per_second": 1.656,
"eval_steps_per_second": 1.656,
"step": 12
},
{
"epoch": 0.0914565683372461,
"grad_norm": 20.134733200073242,
"learning_rate": 9.889271933555213e-05,
"loss": 11.4156,
"step": 14
},
{
"epoch": 0.10452179238542411,
"grad_norm": 1.036009669303894,
"learning_rate": 9.840811514988294e-05,
"loss": 11.4501,
"step": 16
},
{
"epoch": 0.10452179238542411,
"eval_loss": 11.910863876342773,
"eval_runtime": 288.3269,
"eval_samples_per_second": 1.696,
"eval_steps_per_second": 1.696,
"step": 16
},
{
"epoch": 0.11758701643360213,
"grad_norm": 0.3910887539386749,
"learning_rate": 9.783744310203491e-05,
"loss": 11.9102,
"step": 18
},
{
"epoch": 0.13065224048178015,
"grad_norm": 0.7469679713249207,
"learning_rate": 9.718171782608356e-05,
"loss": 11.8815,
"step": 20
},
{
"epoch": 0.13065224048178015,
"eval_loss": 11.953683853149414,
"eval_runtime": 513.8765,
"eval_samples_per_second": 0.952,
"eval_steps_per_second": 0.952,
"step": 20
},
{
"epoch": 0.14371746452995815,
"grad_norm": 0.6554788947105408,
"learning_rate": 9.644210517764014e-05,
"loss": 11.9806,
"step": 22
},
{
"epoch": 0.15678268857813615,
"grad_norm": 0.6563255786895752,
"learning_rate": 9.561992016100293e-05,
"loss": 12.0113,
"step": 24
},
{
"epoch": 0.15678268857813615,
"eval_loss": 11.999762535095215,
"eval_runtime": 627.957,
"eval_samples_per_second": 0.779,
"eval_steps_per_second": 0.779,
"step": 24
},
{
"epoch": 0.16984791262631418,
"grad_norm": 0.5697500109672546,
"learning_rate": 9.471662459112747e-05,
"loss": 11.9923,
"step": 26
},
{
"epoch": 0.1829131366744922,
"grad_norm": 0.4122526943683624,
"learning_rate": 9.373382449457304e-05,
"loss": 11.9021,
"step": 28
},
{
"epoch": 0.1829131366744922,
"eval_loss": 11.957951545715332,
"eval_runtime": 666.2787,
"eval_samples_per_second": 0.734,
"eval_steps_per_second": 0.734,
"step": 28
},
{
"epoch": 0.1959783607226702,
"grad_norm": 0.15934322774410248,
"learning_rate": 9.267326725404599e-05,
"loss": 11.9592,
"step": 30
},
{
"epoch": 0.20904358477084822,
"grad_norm": 0.49380967020988464,
"learning_rate": 9.153683850161706e-05,
"loss": 11.9273,
"step": 32
},
{
"epoch": 0.20904358477084822,
"eval_loss": 11.900312423706055,
"eval_runtime": 678.6572,
"eval_samples_per_second": 0.721,
"eval_steps_per_second": 0.721,
"step": 32
},
{
"epoch": 0.22210880881902623,
"grad_norm": 0.49335020780563354,
"learning_rate": 9.032655876613636e-05,
"loss": 11.8792,
"step": 34
},
{
"epoch": 0.23517403286720426,
"grad_norm": 0.333467960357666,
"learning_rate": 8.904457988080681e-05,
"loss": 11.833,
"step": 36
},
{
"epoch": 0.23517403286720426,
"eval_loss": 11.815201759338379,
"eval_runtime": 652.6748,
"eval_samples_per_second": 0.749,
"eval_steps_per_second": 0.749,
"step": 36
},
{
"epoch": 0.24823925691538226,
"grad_norm": 0.2453756183385849,
"learning_rate": 8.76931811573033e-05,
"loss": 11.8354,
"step": 38
},
{
"epoch": 0.2613044809635603,
"grad_norm": 0.22134803235530853,
"learning_rate": 8.627476533323957e-05,
"loss": 11.7118,
"step": 40
},
{
"epoch": 0.2613044809635603,
"eval_loss": 11.792489051818848,
"eval_runtime": 659.3874,
"eval_samples_per_second": 0.742,
"eval_steps_per_second": 0.742,
"step": 40
},
{
"epoch": 0.2743697050117383,
"grad_norm": 0.22551162540912628,
"learning_rate": 8.479185430018858e-05,
"loss": 11.7013,
"step": 42
},
{
"epoch": 0.2874349290599163,
"grad_norm": 0.24199481308460236,
"learning_rate": 8.324708461985124e-05,
"loss": 11.5874,
"step": 44
},
{
"epoch": 0.2874349290599163,
"eval_loss": 11.56563663482666,
"eval_runtime": 557.8414,
"eval_samples_per_second": 0.877,
"eval_steps_per_second": 0.877,
"step": 44
},
{
"epoch": 0.3005001531080943,
"grad_norm": 0.1317184865474701,
"learning_rate": 8.164320283634585e-05,
"loss": 11.5482,
"step": 46
},
{
"epoch": 0.3135653771562723,
"grad_norm": 0.183132141828537,
"learning_rate": 7.998306059295301e-05,
"loss": 11.4794,
"step": 48
},
{
"epoch": 0.3135653771562723,
"eval_loss": 11.47742748260498,
"eval_runtime": 674.8709,
"eval_samples_per_second": 0.725,
"eval_steps_per_second": 0.725,
"step": 48
}
],
"logging_steps": 2,
"max_steps": 153,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 4,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 9.33894962284462e+16,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}