Eleutherai-oslo's group workspace
Group: 8uMZBmV5EbWDap3b2fe4h3_4c4wzuk1
State
Notes
User
Tags
Created
Runtime
Sweep
activation
adlr_autoresume
adlr_autoresume_interval
apply_query_key_layer_scaling
attention_config
attention_dropout
attention_softmax_in_fp32
batch_size
bias_dropout_fusion
bias_gelu_fusion
char_level_ppl
checkpoint_activations
checkpoint_in_cpu
checkpoint_num_layers
checkpoint_validation_with_forward_pass
clip_grad
config_files.13B_ko.yml
contiguous_checkpointing
data_impl
data_path
deepscale
deepspeed
deepspeed_activation_checkpointing
deepspeed_mpi
detect_nvlink_pairs
distributed_backend
dump_state
dynamic_loss_scale
eod_mask_loss
eval_interval
eval_iters
eval_results_prefix
eval_tasks
eval_tasks_interval
finetune
fp16.enabled
fp16.fp16
fp16.hysteresis
fp16.initial_scale_power
fp16.loss_scale_window
fp16.min_loss_scale
fp16_lm_cross_entropy
fp32_allreduce
gas
Crashed
eleutherai-oslo
5m 29s
-
gelu
false
1000
false
["global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global"]
0
false
2
false
true
false
true
false
1
false
1
# GPT-2 pretraining setup
{
# Tokenizer / checkpoint settings - you will need to change these to the location you have them saved in
"tokenizer-type": "HFTokenizer",
"vocab-file": "./tokenizer/MBBPE/tokenizer.json",
"save": "/fsx/multi-lingual-6b/gpt-neox/checkpoints/13B_scratch",
# "load": "/fsx/multi-lingual-6b/gpt-neox/checkpoints/6B_scratch",
# wandb config
"wandb_team": "eleutherai-oslo",
# If finetuning, edit the following to the location of your finetuning dataset:
"data-path": "/fsx/multi-lingual-6b/gpt-neox/processed/multi_ko_13b_text_document",
# parallelism settings ( you will want to change these based on your cluster setup, ideally scheduling pipeline stages
# across the node boundaries )
"pipe-parallel-size": 1,
"model-parallel-size": 4,
# model settings
"num-layers": 40,
"hidden-size": 5120,
"num-attention-heads": 40,
"seq-length": 2048,
"max-position-embeddings": 2048,
"norm": "layernorm",
"pos-emb": "rotary",
"no-weight-tying": true,
"rotary_ndims": 64,
"gpt_j_residual": true,
"output_layer_parallelism": "column",
# these should provide some speedup but takes a while to build, set to true if desired
"scaled-upper-triang-masked-softmax-fusion": true,
"bias-gelu-fusion": true,
# init methods
"init_method": "small_init",
"output_layer_init_method": "wang_init",
# optimizer settings
"optimizer": {
"type": "Adam",
"params": {
"lr": 0.0001,
"betas": [0.9, 0.95],
"eps": 1.0e-8,
}
},
"min_lr": 0.00001,
"zero_optimization": {
"stage": 1,
"allgather_partitions": True,
"allgather_bucket_size": 500000000,
"overlap_comm": True,
"reduce_scatter": True,
"reduce_bucket_size": 500000000,
"contiguous_gradients": True,
"cpu_offload": False
},
# batch / data settings
"train_micro_batch_size_per_gpu": 2,
"gradient_accumulation_steps": 8,
"data-impl": "mmap",
# "split": "949,50,1",
# activation checkpointing
"checkpoint-activations": true,
"checkpoint-num-layers": 1,
"partition-activations": true,
"synchronize-each-layer": true,
# regularization
"gradient_clipping": 1.0,
"weight-decay": 0.1,
"hidden-dropout": 0,
"attention-dropout": 0,
# precision settings
# "attention_softmax_in_fp32": true,
"fp16": {
"fp16": true,
"enabled": true,
"initial_scale_power": 32,
"loss_scale_window": 1000,
"hysteresis": 2,
"min_loss_scale": 1
},
# misc. training settings
"train-iters": 500000,
"lr-decay-iters": 500000,
"distributed-backend": "nccl",
"lr-decay-style": "cosine",
"warmup": 0.01,
"save-interval": 1000,
"eval-interval": 1000,
"eval-tasks-interval": 50000000000,
"eval-iters": 10,
# logging
"log-interval": 100,
"steps_per_print": 10,
"keep-last-n-checkpoints": 5,
"wall_clock_breakdown": true,
# wandb
"use_wandb": true,
# "wandb_init_all_ranks": true,
"wandb_project": "polyglot-ko-12_8b",
"eval_tasks": ["nsmc"],
# deepspeed launcher
"launcher": "openmpi",
"deepspeed_mpi": true
}
false
mmap
/fsx/multi-lingual-6b/gpt-neox/processed/multi_ko_13b_text_document
false
true
true
true
false
nccl
false
true
false
1000
10
["nsmc"]
50000000000
false
true
true
2
32
1000
1
false
false
8
Failed
eleutherai-oslo
3m 5s
-
gelu
false
1000
false
["global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global","global"]
0
false
2
false
true
false
true
false
1
false
1
# GPT-2 pretraining setup
{
# Tokenizer / checkpoint settings - you will need to change these to the location you have them saved in
"tokenizer-type": "HFTokenizer",
"vocab-file": "./tokenizer/MBBPE/tokenizer.json",
"save": "/fsx/multi-lingual-6b/gpt-neox/checkpoints/13B_scratch",
# "load": "/fsx/multi-lingual-6b/gpt-neox/checkpoints/6B_scratch",
# wandb config
"wandb_team": "eleutherai-oslo",
# If finetuning, edit the following to the location of your finetuning dataset:
"data-path": "/fsx/multi-lingual-6b/gpt-neox/processed/multi_ko_13b_text_document",
# parallelism settings ( you will want to change these based on your cluster setup, ideally scheduling pipeline stages
# across the node boundaries )
"pipe-parallel-size": 1,
"model-parallel-size": 4,
# model settings
"num-layers": 40,
"hidden-size": 5120,
"num-attention-heads": 40,
"seq-length": 2048,
"max-position-embeddings": 2048,
"norm": "layernorm",
"pos-emb": "rotary",
"no-weight-tying": true,
"rotary_ndims": 64,
"gpt_j_residual": true,
"output_layer_parallelism": "column",
# these should provide some speedup but takes a while to build, set to true if desired
"scaled-upper-triang-masked-softmax-fusion": true,
"bias-gelu-fusion": true,
# init methods
"init_method": "small_init",
"output_layer_init_method": "wang_init",
# optimizer settings
"optimizer": {
"type": "Adam",
"params": {
"lr": 0.0001,
"betas": [0.9, 0.95],
"eps": 1.0e-8,
}
},
"min_lr": 0.00001,
"zero_optimization": {
"stage": 1,
"allgather_partitions": True,
"allgather_bucket_size": 500000000,
"overlap_comm": True,
"reduce_scatter": True,
"reduce_bucket_size": 500000000,
"contiguous_gradients": True,
"cpu_offload": False
},
# batch / data settings
"train_micro_batch_size_per_gpu": 2,
"gradient_accumulation_steps": 8,
"data-impl": "mmap",
# "split": "949,50,1",
# activation checkpointing
"checkpoint-activations": true,
"checkpoint-num-layers": 1,
"partition-activations": true,
"synchronize-each-layer": true,
# regularization
"gradient_clipping": 1.0,
"weight-decay": 0.1,
"hidden-dropout": 0,
"attention-dropout": 0,
# precision settings
# "attention_softmax_in_fp32": true,
"fp16": {
"fp16": true,
"enabled": true,
"initial_scale_power": 32,
"loss_scale_window": 1000,
"hysteresis": 2,
"min_loss_scale": 1
},
# misc. training settings
"train-iters": 500000,
"lr-decay-iters": 500000,
"distributed-backend": "nccl",
"lr-decay-style": "cosine",
"warmup": 0.01,
"save-interval": 1000,
"eval-interval": 1000,
"eval-tasks-interval": 50000000000,
"eval-iters": 10,
# logging
"log-interval": 100,
"steps_per_print": 10,
"keep-last-n-checkpoints": 5,
"wall_clock_breakdown": true,
# wandb
"use_wandb": true,
# "wandb_init_all_ranks": true,
"wandb_project": "polyglot-ko-12_8b",
"eval_tasks": ["nsmc"],
# deepspeed launcher
"launcher": "openmpi",
"deepspeed_mpi": true
}
false
mmap
/fsx/multi-lingual-6b/gpt-neox/processed/multi_ko_13b_text_document
false
true
true
true
false
nccl
false
true
false
1000
10
["nsmc"]
50000000000
false
true
true
2
32
1000
1
false
false
8
1-2
of 2