Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add larger GPT models #213

Merged
merged 1 commit into from
Jan 10, 2022
Merged

Add larger GPT models #213

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
97 changes: 97 additions & 0 deletions composer/yamls/models/gpt2_1,3b.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,97 @@
train_dataset:
lm:
split: train
datadir: /datasets/openwebtext_saved
tokenizer_name: gpt2
seed: 17
shuffle: false
drop_last: false
num_tokens: 7340032000 # 7000 (batches) * 1024 (batch_size) * 1024 (seq_len)
val_dataset:
lm:
split: validation
datadir: /datasets/openwebtext_saved
tokenizer_name: gpt2
seed: 17
shuffle: false
drop_last: false
num_tokens: 102400000 # 100000 (seq) * 1024 (seq_len)
model:
gpt2:
use_pretrained: false
tokenizer_name: gpt2
model_config:
activation_function: gelu_new
architectures:
- GPT2LMHeadModel
attn_pdrop: 0.1
bos_token_id: 50256
embd_pdrop: 0.1
eos_token_id: 50256
initializer_range: 0.02
layer_norm_epsilon: 1.0e-05
model_type: gpt2
n_ctx: 1024
n_embd: 2048
n_head: 16
n_inner: 8192
n_layer: 24
n_positions: 1024
resid_pdrop: 0.1
scale_attn_weights: true
summary_activation: null
summary_first_dropout: 0.1
summary_proj_to_labels: true
summary_type: cls_index
summary_use_proj: true
task_specific_params:
text-generation:
do_sample: true
max_length: 50
transformers_version: 4.11.0.dev0
use_cache: true
vocab_size: 50257
optimizer:
adamw:
lr: 2.0e-4
betas:
- 0.9
- 0.999
eps: 1.0e-08
weight_decay: 0.0
schedulers:
- warmup:
warmup_method: linear
warmup_factor: 0
interval: step
warmup_iters: 140ba
- cosine_decay:
interval: step
eta_min: 0
verbose: false
T_max: 13860ba
loggers:
- file:
log_level: batch
filename: stdout
buffer_size: 1
flush_every_n_batches: 100
every_n_batches: 100
every_n_epochs: 1
max_epochs: 1
train_batch_size: 1024
eval_batch_size: 8
seed: 17
device:
gpu: {}
dataloader:
pin_memory: true
persistent_workers: true
num_workers: 8
timeout: 0
prefetch_factor: 2
precision: amp
grad_clip_norm: 1.0
grad_accum: 128
validate_every_n_batches: 1000
validate_every_n_epochs: 1
97 changes: 97 additions & 0 deletions composer/yamls/models/gpt2_13b.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,97 @@
train_dataset:
lm:
split: train
datadir: /datasets/openwebtext_saved
tokenizer_name: gpt2
seed: 17
shuffle: false
drop_last: false
num_tokens: 7340032000 # 3500 (batches) * 2048 (batch_size) * 1024 (seq_len)
val_dataset:
lm:
split: validation
datadir: /datasets/openwebtext_saved
tokenizer_name: gpt2
seed: 17
shuffle: false
drop_last: false
num_tokens: 102400000 # 100000 (seq) * 1024 (seq_len)
model:
gpt2:
use_pretrained: false
tokenizer_name: gpt2
model_config:
activation_function: gelu_new
architectures:
- GPT2LMHeadModel
attn_pdrop: 0.1
bos_token_id: 50256
embd_pdrop: 0.1
eos_token_id: 50256
initializer_range: 0.02
layer_norm_epsilon: 1.0e-05
model_type: gpt2
n_ctx: 1024
n_embd: 5120
n_head: 40
n_inner: 20480
n_layer: 40
n_positions: 1024
resid_pdrop: 0.1
scale_attn_weights: true
summary_activation: null
summary_first_dropout: 0.1
summary_proj_to_labels: true
summary_type: cls_index
summary_use_proj: true
task_specific_params:
text-generation:
do_sample: true
max_length: 50
transformers_version: 4.11.0.dev0
use_cache: true
vocab_size: 50257
optimizer:
adamw:
lr: 1.0e-4
betas:
- 0.9
- 0.999
eps: 1.0e-08
weight_decay: 0.0
schedulers:
- warmup:
warmup_method: linear
warmup_factor: 0
interval: step
warmup_iters: 140ba
- cosine_decay:
interval: step
eta_min: 0
verbose: false
T_max: 13860ba
loggers:
- file:
log_level: batch
filename: stdout
buffer_size: 1
flush_every_n_batches: 100
every_n_batches: 100
every_n_epochs: 1
max_epochs: 1
train_batch_size: 2048
eval_batch_size: 8
seed: 17
device:
gpu: {}
dataloader:
pin_memory: true
persistent_workers: true
num_workers: 8
timeout: 0
prefetch_factor: 2
precision: amp
grad_clip_norm: 1.0
grad_accum: 256
validate_every_n_batches: 1000
validate_every_n_epochs: 1
97 changes: 97 additions & 0 deletions composer/yamls/models/gpt2_2,7b.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,97 @@
train_dataset:
lm:
split: train
datadir: /datasets/openwebtext_saved
tokenizer_name: gpt2
seed: 17
shuffle: false
drop_last: false
num_tokens: 7340032000 # 7000 (batches) * 1024 (batch_size) * 1024 (seq_len)
val_dataset:
lm:
split: validation
datadir: /datasets/openwebtext_saved
tokenizer_name: gpt2
seed: 17
shuffle: false
drop_last: false
num_tokens: 102400000 # 100000 (seq) * 1024 (seq_len)
model:
gpt2:
use_pretrained: false
tokenizer_name: gpt2
model_config:
activation_function: gelu_new
architectures:
- GPT2LMHeadModel
attn_pdrop: 0.1
bos_token_id: 50256
embd_pdrop: 0.1
eos_token_id: 50256
initializer_range: 0.02
layer_norm_epsilon: 1.0e-05
model_type: gpt2
n_ctx: 1024
n_embd: 2560
n_head: 32
n_inner: 10240
n_layer: 32
n_positions: 1024
resid_pdrop: 0.1
scale_attn_weights: true
summary_activation: null
summary_first_dropout: 0.1
summary_proj_to_labels: true
summary_type: cls_index
summary_use_proj: true
task_specific_params:
text-generation:
do_sample: true
max_length: 50
transformers_version: 4.11.0.dev0
use_cache: true
vocab_size: 50257
optimizer:
adamw:
lr: 1.6e-4
betas:
- 0.9
- 0.999
eps: 1.0e-08
weight_decay: 0.0
schedulers:
- warmup:
warmup_method: linear
warmup_factor: 0
interval: step
warmup_iters: 140ba
- cosine_decay:
interval: step
eta_min: 0
verbose: false
T_max: 13860ba
loggers:
- file:
log_level: batch
filename: stdout
buffer_size: 1
flush_every_n_batches: 100
every_n_batches: 100
every_n_epochs: 1
max_epochs: 1
train_batch_size: 1024
eval_batch_size: 8
seed: 17
device:
gpu: {}
dataloader:
pin_memory: true
persistent_workers: true
num_workers: 8
timeout: 0
prefetch_factor: 2
precision: amp
grad_clip_norm: 1.0
grad_accum: 128
validate_every_n_batches: 1000
validate_every_n_epochs: 1
97 changes: 97 additions & 0 deletions composer/yamls/models/gpt2_350m.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,97 @@
train_dataset:
lm:
split: train
datadir: /datasets/openwebtext_saved
tokenizer_name: gpt2
seed: 17
shuffle: false
drop_last: false
num_tokens: 5767168000 # 11000 (batches) * 512 (batch_size) * 1024 (seq_len)
val_dataset:
lm:
split: validation
datadir: /datasets/openwebtext_saved
tokenizer_name: gpt2
seed: 17
shuffle: false
drop_last: false
num_tokens: 102400000 # 100000 (seq) * 1024 (seq_len)
model:
gpt2:
use_pretrained: false
tokenizer_name: gpt2
model_config:
activation_function: gelu_new
architectures:
- GPT2LMHeadModel
attn_pdrop: 0.1
bos_token_id: 50256
embd_pdrop: 0.1
eos_token_id: 50256
initializer_range: 0.02
layer_norm_epsilon: 1.0e-05
model_type: gpt2
n_ctx: 1024
n_embd: 1024
n_head: 16
n_inner: 4096
n_layer: 24
n_positions: 1024
resid_pdrop: 0.1
scale_attn_weights: true
summary_activation: null
summary_first_dropout: 0.1
summary_proj_to_labels: true
summary_type: cls_index
summary_use_proj: true
task_specific_params:
text-generation:
do_sample: true
max_length: 50
transformers_version: 4.11.0.dev0
use_cache: true
vocab_size: 50257
optimizer:
adamw:
lr: 3.0e-4
betas:
- 0.9
- 0.999
eps: 1.0e-08
weight_decay: 0.0
schedulers:
- warmup:
warmup_method: linear
warmup_factor: 0
interval: step
warmup_iters: 110ba
- cosine_decay:
interval: step
eta_min: 0
verbose: false
T_max: 10890ba
loggers:
- file:
log_level: batch
filename: stdout
buffer_size: 1
flush_every_n_batches: 100
every_n_batches: 100
every_n_epochs: 1
max_epochs: 1
train_batch_size: 512
eval_batch_size: 8
seed: 17
device:
gpu: {}
dataloader:
pin_memory: true
persistent_workers: true
num_workers: 8
timeout: 0
prefetch_factor: 2
precision: amp
grad_clip_norm: 1.0
grad_accum: 64
validate_every_n_batches: 1000
validate_every_n_epochs: 1
Loading