Ebs's workspace
Runs
4
State
Notes
User
Tags
Created
Runtime
Sweep
batch_size
checkpointer._component_
checkpointer.checkpoint_dir
checkpointer.checkpoint_files
checkpointer.model_type
checkpointer.output_dir
compile
dataset._component_
device
dtype
enable_activation_checkpointing
epochs
gradient_accumulation_steps
log_every_n_steps
log_peak_memory_stats
loss._component_
lr_scheduler._component_
lr_scheduler.num_warmup_steps
max_steps_per_epoch
metric_logger._component_
metric_logger.log_dir
metric_logger.name
metric_logger.project
model._component_
optimizer._component_
optimizer.lr
optimizer_in_bwd
output_dir
resume_from_checkpoint
shuffle
tokenizer._component_
tokenizer.path
global_step
loss
lr
peak_memory_active
peak_memory_alloc
peak_memory_reserved
tokens_per_second_per_gpu
Finished
-
ebs
2m 2s
-
4
torchtune.training.FullModelMetaCheckpointer
/tmp/Meta-Llama-3-8B-Instruct/original/
["consolidated.00.pth"]
LLAMA3
/tmp/Meta-Llama-3-8B-Instruct/
false
torchtune.datasets.alpaca_dataset
cuda
bf16
true
1
1
1
true
torchtune.modules.loss.CEWithChunkedOutputLoss
torchtune.training.lr_schedulers.get_cosine_schedule_with_warmup
100
100
torchtune.training.metric_logging.WandBLogger
/tmp/full-llama3-finetune
new-bsz4-ga1
llama3-8b-full-single
torchtune.models.llama3.llama3_8b
bitsandbytes.optim.PagedAdamW8bit
0.00001
false
/tmp/full-llama3-finetune
false
true
torchtune.models.llama3.llama3_tokenizer
/tmp/Meta-Llama-3-8B-Instruct/original/tokenizer.model
100
1.16348
0.00001
30.29896
30.29896
31.69336
910.98041
Finished
-
ebs
3m
-
1
torchtune.training.FullModelMetaCheckpointer
/tmp/Meta-Llama-3-8B-Instruct/original/
["consolidated.00.pth"]
LLAMA3
/tmp/Meta-Llama-3-8B-Instruct/
false
torchtune.datasets.alpaca_dataset
cuda
bf16
true
1
4
1
true
torchtune.modules.loss.CEWithChunkedOutputLoss
torchtune.training.lr_schedulers.get_cosine_schedule_with_warmup
100
100
torchtune.training.metric_logging.WandBLogger
/tmp/full-llama3-finetune
baseline-bsz1-ga4
llama3-8b-full-single
torchtune.models.llama3.llama3_8b
bitsandbytes.optim.PagedAdamW8bit
0.00001
false
/tmp/full-llama3-finetune
false
true
torchtune.models.llama3.llama3_tokenizer
/tmp/Meta-Llama-3-8B-Instruct/original/tokenizer.model
100
1.17437
0.00001
33.33807
33.33807
34.41992
74.47246
Finished
-
ebs
3m 4s
-
1
torchtune.training.FullModelMetaCheckpointer
/tmp/Meta-Llama-3-8B-Instruct/original/
["consolidated.00.pth"]
LLAMA3
/tmp/Meta-Llama-3-8B-Instruct/
false
torchtune.datasets.alpaca_dataset
cuda
bf16
true
1
4
1
true
torchtune.modules.loss.CEWithChunkedOutputLoss
torchtune.training.lr_schedulers.get_cosine_schedule_with_warmup
100
100
torchtune.training.metric_logging.WandBLogger
/tmp/full-llama3-finetune
new-bsz1-ga4
llama3-8b-full-single
torchtune.models.llama3.llama3_8b
bitsandbytes.optim.PagedAdamW8bit
0.00001
false
/tmp/full-llama3-finetune
false
true
torchtune.models.llama3.llama3_tokenizer
/tmp/Meta-Llama-3-8B-Instruct/original/tokenizer.model
100
1.16044
0.00001
31.64735
31.64735
32.63086
499.9202
Finished
-
ebs
2m 19s
-
4
torchtune.training.FullModelMetaCheckpointer
/tmp/Meta-Llama-3-8B-Instruct/original/
["consolidated.00.pth"]
LLAMA3
/tmp/Meta-Llama-3-8B-Instruct/
false
torchtune.datasets.alpaca_dataset
cuda
bf16
true
1
1
1
true
torchtune.modules.loss.CEWithChunkedOutputLoss
torchtune.training.lr_schedulers.get_cosine_schedule_with_warmup
100
100
torchtune.training.metric_logging.WandBLogger
/tmp/full-llama3-finetune
baseline-bsz4-ga1
llama3-8b-full-single
torchtune.models.llama3.llama3_8b
bitsandbytes.optim.PagedAdamW8bit
0.00001
false
/tmp/full-llama3-finetune
false
true
torchtune.models.llama3.llama3_tokenizer
/tmp/Meta-Llama-3-8B-Instruct/original/tokenizer.model
100
1.16454
0.00001
30.29896
30.29896
31.69336
914.93134
1-4
of 4