> ## Documentation Index
> Fetch the complete documentation index at: https://training-docs.cerebras.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Trainer Configuration Overview
The Cerebras Model Zoo comes packaged with a few useful utilities. Namely, It features a way to configure a [`Trainer`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html "cerebras.modelzoo.Trainer") class using a YAML configuration file.
On this page you will learn how to write a YAML file to configure an instance of the [`Trainer`](../model-zoo/api/trainer-api "cerebras.modelzoo.Trainer") class. By the end, you should be comfortable enough with the YAML specification to write your own configuration files from scratch.
## Prerequisites
Please ensure that you have read through the [Trainer Overview](../model-zoo/trainer-overview) beforehand. The rest of this page assumes that you already have at least a cursory understanding of what the Cerebras Model Zoo Trainer is and how to use the python API.
## Base Specification
The YAML specification is intentionally designed to map almost exactly one-to-one with the Trainer’s python API.
The Trainer’s constructor can be specified via a YAML configuration file as follows:
```Yaml theme={null}
trainer:
init:
device: "CSX"
model_dir: "./model_dir"
model:
# The remaining arguments to the model class
vocab_size: 1024
max_position_embeddings: 1024
...
optimizer:
# Corresponds to cstorch.optim.SGD
SGD:
lr: 0.01
momentum: 0.9
loop:
num_steps: 1000
eval_steps: 100
eval_frequency: 100
checkpoint:
steps: 100
fit:
train_dataloader:
data_processor: GptHDF5MapDataProcessor
data_dir: "/path/to/train/data"
batch_size: 64
...
val_dataloader:
data_processor: GptHDF5MapDataProcessor
data_dir: "/path/to/validation/data"
batch_size: 64
...
```
```Python theme={null}
import cerebras.pytorch as cstorch
from cerebras.modelzoo import Trainer
from cerebras.modelzoo.trainer.callbacks import Checkpoint, TrainingLoop
from cerebras.modelzoo.models.nlp.gpt2.model import Gpt2Model
from cerebras.modelzoo.data.nlp.gpt.GptHDF5MapDataProcessor import GptHDF5MapDataProcessor
trainer = Trainer(
device="CSX", # The device to run on
model_dir="./model_dir", # The directory at which to store artifacts
model=lambda: Gpt2Model(
vocab_size=1024,
max_position_embeddings=1024,
...
),
optimizer=lambda model: cstorch.optim.SGD(
model.parameters(), lr=0.01, momentum=0.9
),
loop=TrainingLoop(num_steps=1000, eval_steps=100, eval_frequency=100),
checkpoint=Checkpoint(steps=100),
)
trainer.fit(
train_dataloader=cstorch.utils.data.DataLoader(
GptHDF5MapDataProcessor,
data_dir="/path/to/train/data",
batch_size=64,
...,
),
val_dataloader=cstorch.utils.data.DataLoader(
GptHDF5MapDataProcessor,
data_dir="/path/to/validation/data",
batch_size=64,
...,
),
)
```
If you open the Python tab above, you can see the equivalent Python code that the YAML configuration corresponds to. As can be seen, the YAML specification almost directly mirrors the python API. This was an intentional design choice to make it easy to use one if you are familiar with the other.
## Breaking It Down
The YAML specification starts with the top level `trainer` key.
```Bash theme={null}
trainer:
...
```
If this key is not present, then the configuration is not valid.
The `trainer` accepts the following subkeys:
* [`init`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html#cerebras.modelzoo.Trainer)
* [`fit`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html?highlight=fit#cerebras.modelzoo.Trainer.fit)
* [`validate`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html?highlight=fit#cerebras.modelzoo.Trainer.validate)
* [`validate_all`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html?highlight=fit#cerebras.modelzoo.Trainer.validate_all)
### init
The [`init`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html#cerebras.modelzoo.Trainer)key is used to specify the arguments to the Trainer’s (/model-zoo/api/trainer-api) constructor
The arguments are passed as key-value pairs, where the key is the argument name and the value is the argument value.
Below are all of the accepted keys alongside YAML examples and their equivalent Python counterparts:
#### device
The device to train the model on. If provided, it must be one of `"CSX"`, `"CPU"`, or `"GPU"`.
```yaml theme={null}
trainer:
init:
device: "CSX"
...
...
```
```python theme={null}
from cerebras.modelzoo import Trainer
trainer = Trainer(
device="CSX",
...,
)
...
```
#### backend
Configures the backend used to train the model. If provided, it is expected to be a dictionary whose keys will be used to construct a [`cerebras.pytorch.backend`](https://training-api.cerebras.ai/en/latest/wsc/api/cerebras_pytorch/cstorch.html#backend) instance.
```yaml theme={null}
trainer:
init:
backend:
backend_type: "CSX"
cluster_config:
num_csx: 4
mount_dirs:
- /path/to/dir1
- /path/to/dir2
...
...
...
...
```
```python theme={null}
import cerebras.pytorch as cstorch
from cerebras.modelzoo import Trainer
trainer = Trainer(
backend=cstorch.backend(
backend_type="CSX",
cluster_config=cstorch.distributed.ClusterConfig(
num_csx=4,
mount_dirs=["/path/to/dir1", "/path/to/dir2"],
...,
),
...,
)
...
)
...
```
The `backend` argument is mutually exclusive with `device`. The functionality it provides is a strict superset of the functionality provided by `device`. To use a certain backend with all default parameters, you may specify `device`. To configure anything about the backend, you must specify those parameters via the `backend` key.
To learn more about the backend argument, you can check out [Trainer Backend](../model-zoo/components/trainer-components/backend).
#### model\_dir
The directory where the model artifacts are saved. Some of the artifacts that may be dumped into the `model_dir` include (but are not limited to):
* Client-side logs
* Checkpoints
* TensorBoard event files
* Tensor summaries
* Validation results
```yaml theme={null}
trainer:
init:
...
model_dir: "./model_dir"
...
...
```
```python theme={null}
from cerebras.modelzoo import Trainer
trainer = Trainer(
...,
model_dir="./model_dir",
...,
)
...
```
#### model
Configures the [`Module`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html) to train/validate using the constructed Trainer. All subkeys are passed as arguments to the model class.
```yaml theme={null}
trainer:
init:
...
model:
vocab_size: 1024
max_position_embeddings: 1024
...
...
...
```
```python theme={null}
from cerebras.modelzoo import Trainer
from cerebras.modelzoo.models.nlp.gpt2.model import Gpt2Model
trainer = Trainer(
...,
model=lambda: Gpt2Model(
vocab_size=1024,
max_position_embeddings=1024,
...,
),
...,
)
...
```
To learn more about the model argument, you can check out [Trainer Model](../model-zoo/components/trainer-components/model).
#### optimizer
Configures the [`Optimizer`](../model-zoo/components/trainer-components/optimizer-and-scheduler) to use to optimize the model’s weights during training.
The value at this key is expected to be a dictionary. This dictionary is expected to contain a single key that specifies the name of the Cerebras optimizer to construct. That is to say, it must be the name of a subclass of [`Optimizer`](../model-zoo/components/trainer-components/optimizer-and-scheduler) subclasses that come packaged in `cerebras.pytorch`)
The value of the Optimizer name key is expected to be dictionary of key-value pairs that correspond to the arguments of the optimizer subclass.
```yaml theme={null}
trainer:
init:
...
optimizer:
# Corresponds to cstorch.optim.SGD
SGD:
lr: 0.01
momentum: 0.9
...
...
```
The `params` argument to the optimizer is automatically passed in and thus is not required.
```python theme={null}
import cerebras.pytorch as cstorch
from cerebras.modelzoo import Trainer
trainer = Trainer(
...,
optimizer=lambda model: cstorch.optim.SGD(
model.parameters(),
lr=0.01,
momentum=0.9,
),
...,
)
...
```
To learn more about the optimizer argument, you can check out [Trainer Optimizer](../model-zoo/components/trainer-components/optimizer-and-scheduler).
#### schedulers
Configures the [`Scheduler`](../model-zoo/components/trainer-components/optimizer-and-scheduler) instances to use during the training run.
The value at this key is expected to be a dictionary or a list of dictionaries. Each dictionary is expected to have a single key specifying the name of the `Scheduler` to use.
The corresponding value of the Scheduler name key is expected to be mapping of key-value pairs that are passed as keyword arguments to the Scheduler.
```yaml theme={null}
trainer:
init:
...
schedulers:
- LinearLR:
initial_learning_rate: 0.01
end_learning_rate: 0.001
total_iters: 100
...
...
```
The optimizer argument to the Scheduler is automatically passed in and thus is not required.
```python theme={null}
import cerebras.pytorch as cstorch
from cerebras.modelzoo import Trainer
trainer = Trainer(
...,
schedulers=[
lambda optimizer: cstorch.optim.lr_scheduler.LinearLR(
optimizer,
initial_learning_rate=0.01,
end_learning_rate=0.001,
total_iters=100,
),
...
],
...,
)
...
...
```
To learn more about the schedulers argument, you can check out [Trainer Schedulers](../model-zoo/components/trainer-components/optimizer-and-scheduler).
#### precision
Configures [`Precision`](https://training-api.cerebras.ai/en/latest/wsc/api/cerebras_pytorch/optim.html#cerebras.pytorch.optim.Optimizer).
Today, the only supported `Precision` type is [`MixedPrecision`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/generated/cerebras.modelzoo.trainer.callbacks.html?highlight=mixedprecision#cerebras.modelzoo.trainer.callbacks.MixedPrecision).
So, the value of the `precision` key is expected to be a dictionary corresponding to the arguments of `MixedPrecision`.
```yaml theme={null}
trainer:
init:
...
precision:
fp16_type: float16
precision_opt_level: 1
loss_scaling_factor: dynamic
max_gradient_norm: 1.0
...
...
...
```
```python theme={null}
from cerebras.modelzoo import Trainer
from cerebras.modelzoo.trainer.callbacks import MixedPrecision
trainer = Trainer(
...,
precision=MixedPrecision(
fp16_type="float16",
precision_opt_level=1,
loss_scaling_factor="dynamic",
max_gradient_norm=1.0,
...,
),
...,
)
...
```
To learn more about the precision argument, you can check out [Trainer Precision](../model-zoo/api/cerebras-modelzoo/cerebras-modelzoo).
#### sparsity
Configures the [`SparsityAlgorithm`](https://training-api.cerebras.ai/en/latest/wsc/api/cerebras_pytorch/sparse.html?highlight=sparsityalgorithm#cerebras.pytorch.sparse.SparsityAlgorithm) to use to sparsity the model’s weights and optimizer state.
The value at this key is expected to be a dictionary. At a minimum, this dictionary is expected to contain an `algorithm` key that specifies the name of the sparsity algorithm to apply as well as a `sparsity` that specifies the level of sparsity to apply.
```YAML theme={null}
trainer:
init:
...
sparsity:
algorithm: Static
sparsity: 0.5
...
...
```
```python theme={null}
import cerebras.pytorch as cstorch
from cerebras.modelzoo import Trainer
trainer = Trainer(
...,
sparsity=cstorch.sparse.Static(
sparsity=0.5,
),
...,
)
...
```
To learn more about how sparsity can be configured, see [Train a Model with Weight Sparsity](../model-zoo/tutorials/train-a-model-with-weight-sparsity).
#### loop
Configures a [`TrainingLoop`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/generated/cerebras.modelzoo.trainer.callbacks.html?highlight=trainingloop#cerebras.modelzoo.trainer.callbacks.TrainingLoop) instance that specifies how many steps to train and validate for.
```YAML theme={null}
trainer:
init:
...
loop:
num_steps: 1000
eval_steps: 100
eval_frequency: 100
...
...
```
```python theme={null}
from cerebras.modelzoo import Trainer
from cerebras.modelzoo.trainer.callbacks import TrainingLoop
trainer = Trainer(
...,
loop=TrainingLoop(
num_steps=1000,
eval_steps=100,
eval_frequency=100,
),
...,
)
...
```
To learn more about the loop argument, you can check out [Training Loop](../model-zoo/components/trainer-components/loop).
#### checkpoint
Configures a [`Checkpoint`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/generated/cerebras.modelzoo.trainer.callbacks.html#checkpoint) instance that specifies how frequently the trainer should save checkpoints during training.
```YAML theme={null}
trainer:
init:
...
checkpoint:
steps: 100
...
...
```
```python theme={null}
from cerebras.modelzoo import Trainer
from cerebras.modelzoo.trainer.callbacks import Checkpoint
trainer = Trainer(
...,
loop=Checkpoint(
steps=1000,
),
...,
)
...
```
To learn more about the checkpoint argument, you can check out [Checkpointing](../model-zoo/components/trainer-components/checkpointing).
#### logging
Configures a [`Logging`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/generated/cerebras.modelzoo.trainer.callbacks.html#cerebras.modelzoo.trainer.callbacks.Logging) instance that configures the Python logger as well as specify how frequently the trainer should be writing logs.
```YAML theme={null}
trainer:
init:
...
logging:
log_steps: 10
log_level: INFO
...
...
```
```python theme={null}
from cerebras.modelzoo import Trainer
from cerebras.modelzoo.trainer.callbacks import Logging
trainer = Trainer(
...,
logging=Logging(
log_steps=10,
log_level="INFO",
),
...,
)
...
```
In the above example, the Python logger is configured to allow info logs to be printed and to print logs at every 10 steps.
To learn more about the logging argument, you can check out [Trainer Logging](../model-zoo/components/trainer-components/logging).
#### callbacks
This key accepts a list of dictionaries, each of which specifies the configuration for some [`Callback`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/generated/cerebras.modelzoo.trainer.callbacks.html#callback) class.
Each dictionary is expected to have a single key specifying the name of the `Callback`
The value at this key are passed in as keyword arguments to the subclass’s constructor.
```YAML theme={null}
trainer:
init:
...
callbacks:
- CheckLoss: {}
- ComputeNorm: {}
- RateProfiler: {}
- LogOptimizerParamGroup:
keys:
- lr
...
...
```
```python theme={null}
from cerebras.modelzoo import Trainer
from cerebras.modelzoo.trainer.callbacks import (
CheckLoss,
ComputeNorm,
RateProfiler,
LogOptimizerParamGroup,
)
trainer = Trainer(
...,
callbacks=[
CheckLoss(),
ComputeNorm(),
RateProfiler(),
LogOptimizerParamGroup(keys=["lr"])
],
...,
)
...
```
You can even include your own custom callbacks here. To learn more, you can read [Customizing the Trainer with Callbacks](../model-zoo/components/trainer-components/customizing-the-trainer-with-callbacks)
#### loggers
This key accepts a list of dictionaries, each of which specifies the configuration for some [`Logger`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/generated/cerebras.modelzoo.trainer.loggers.html#cerebras.modelzoo.trainer.loggers.Logger) class.
Each dictionary is expected to have a single key specifying the name of the `Logger`.
The value at this key are passed in as keyword arguments to the subclass’s constructor.
```YAML theme={null}
trainer:
init:
...
logger:
- ProgressLogger: {}
- TensorboardLogger: {}
...
...
```
```python theme={null}
from cerebras.modelzoo import Trainer
from cerebras.modelzoo.trainer.loggers import (
ProgressLogger,
TensorboardLogger,
)
trainer = Trainer(
...,
loggers=[
ProgressLogger(),
Tensorboardlogger(),
],
...,
)
...
```
You can even include your own custom loggers here. To learn more, you can read [Loggers](../model-zoo/components/trainer-components/logging).
#### seed
This key accepts a single integer value to seed the random number generator.
Setting this parameter will seed the PyTorch generator via a call to [`torch.manual_seed`](https://pytorch.org/docs/stable/generated/torch.manual_seed.html#torch.manual_seed "(in PyTorch v2.4)").
```YAML theme={null}
trainer:
init:
...
seed: 2024
...
```
```python theme={null}
from cerebras.modelzoo import Trainer
trainer = Trainer(
...,
seed=2024,
)
...
```
### fit
The [`fit`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html#cerebras.modelzoo.Trainer.fit) key is used to specify the arguments to the [`Trainer`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html) method.
The arguments are passed as key-value pairs, where the key is the argument name and the value is the argument value.
Below are all of the accepted keys alongside YAML examples and their equivalent Python counterparts:
#### train\_dataloader
This key is used to configure the training dataloader used to train the model.
This value at this key is expected to be a dictionary containing at a minimum the `data_processor` key which specifies the name of the data processors to use.
All other key-values in the dictionary are passed as argument to [`cerebras.pytorch.utils.data.DataLoader`](https://training-api.cerebras.ai/en/latest/wsc/api/cerebras_pytorch/cstorch.html?highlight=dataloader#cerebras.pytorch.utils.data.DataLoader).
```YAML theme={null}
trainer:
init:
...
fit:
train_dataloader:
data_processor: GptHDF5MapDataProcessor
data_dir: "/path/to/train/data"
batch_size: 64
...
...
```
```python theme={null}
import cerebras.pytorch as cstorch
from cerebras.modelzoo import Trainer
from cerebras.modelzoo.data.nlp.gpt.GptHDF5MapDataProcessor import GptHDF5MapDataProcessor
trainer = Trainer(...)
trainer.fit(
train_dataloader=cstorch.utils.data.DataLoader(
GptHDF5MapDataProcessor,
data_dir="/path/to/train/data",
batch_size=64,
...,
),
...
)
```
#### val\_dataloader
This key is used to configure the validation dataloader(s) used to validate the model.
The dataloader configured here gets run for `eval_steps` every `eval_frequency` training steps.
This value at this key is expected to be a dictionary or a list of dictionaries. Each dictionary is expected to contain at a minimum the `data_processor` key which specifies the name of the data processors to use.
All other key-values in the dictionary are passed as argument to [`cerebras.pytorch.utils.data.DataLoader`](https://training-api.cerebras.ai/en/latest/wsc/api/cerebras_pytorch/cstorch.html?highlight=dataloader#cerebras.pytorch.utils.data.DataLoader).
```YAML theme={null}
trainer:
init:
...
fit:
...
val_dataloader:
- data_processor: GptHDF5MapDataProcessor
data_dir: "/path/to/validation/data"
batch_size: 64
...
...
```
```python theme={null}
import cerebras.pytorch as cstorch
from cerebras.modelzoo import Trainer
from cerebras.modelzoo.data.nlp.gpt.GptHDF5MapDataProcessor import GptHDF5MapDataProcessor
trainer = Trainer(...)
trainer.fit(
...,
val_dataloader=[
cstorch.utils.data.DataLoader(
GptHDF5MapDataProcessor,
data_dir="/path/to/validation/data",
batch_size=64,
...,
),
]
...
)
```
#### ckpt\_path
Specifies the path to the checkpoint to load.
```YAML theme={null}
trainer:
init:
...
fit:
...
ckpt_path: /path/to/checkpoint
...
```
```python theme={null}
from cerebras.modelzoo import Trainer
trainer = Trainer(...)
trainer.fit(
...,
ckpt_path="/path/to/checkpoint",
)
```
### validate
The [`validate`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html#cerebras.modelzoo.Trainer.validate) method.
The arguments are passed as key-value pairs, where the key is the argument name and the value is the argument value.
Below are all of the accepted keys alongside YAML examples and their equivalent Python counterparts:
#### val\_dataloader
This key is used to configure the validation dataloader used to validate the model.
This value at this key is expected to be a dictionary that contains at a minimum the `data_processor` key which specifies the name of the data processors to use.
All other key-values in the dictionary are passed as argument to [`cerebras.pytorch.utils.data.DataLoader`](https://training-api.cerebras.ai/en/latest/wsc/api/cerebras_pytorch/cstorch.html?highlight=dataloader#cerebras.pytorch.utils.data.DataLoader).
```YAML theme={null}
trainer:
init:
...
validate:
val_dataloader:
data_processor: GptHDF5MapDataProcessor
data_dir: "/path/to/validation/data"
batch_size: 64
...
...
```
```python theme={null}
import cerebras.pytorch as cstorch
from cerebras.modelzoo import Trainer
from cerebras.modelzoo.data.nlp.gpt.GptHDF5MapDataProcessor import GptHDF5MapDataProcessor
trainer = Trainer(...)
trainer.validate(
val_dataloader=cstorch.utils.data.DataLoader(
GptHDF5MapDataProcessor,
data_dir="/path/to/validation/data",
batch_size=64,
...,
),
...
)
```
The validation dataloader is intended to be used alongside the validation metrics classes.
#### ckpt\_path
Specifies the path to the checkpoint to load.
```YAML theme={null}
trainer:
init:
...
validate:
...
ckpt_path: /path/to/checkpoint
...
```
```python theme={null}
from cerebras.modelzoo import Trainer
trainer = Trainer(...)
trainer.validate(
...,
ckpt_path="/path/to/checkpoint",
)
```
### validate\_all
The [`validate_all`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html#cerebras.modelzoo.Trainer.validate_all) method.
The arguments are passed as key-value pairs, where the key is the argument name and the value is the argument value.
Below are all of the accepted keys alongside YAML examples and their equivalent Python counterparts:
#### val\_dataloaders
This key is used to configure the validation dataloader(s) used to validate the model.
This value at this key is expected to be a dictionary or a list of dictionaries. Each dictionary is expected to contain at a minimum the `data_processor` key which specifies the name of the data processors to use.
All other key-values in the dictionary are passed as argument to [`cerebras.pytorch.utils.data.DataLoader`](https://training-api.cerebras.ai/en/latest/wsc/api/cerebras_pytorch/cstorch.html?highlight=dataloader#cerebras.pytorch.utils.data.DataLoader).
```YAML theme={null}
trainer:
init:
...
validate_all:
val_dataloaders:
- data_processor: GptHDF5MapDataProcessor
data_dir: "/path/to/validation/data1"
batch_size: 64
...
- data_processor: GptHDF5MapDataProcessor
data_dir: "/path/to/validation/data2"
batch_size: 64
...
...
...
```
```python theme={null}
Python Python
import cerebras.pytorch as cstorch
from cerebras.modelzoo import Trainer
from cerebras.modelzoo.data.nlp.gpt.GptHDF5MapDataProcessor import GptHDF5MapDataProcessor
trainer = Trainer(...)
trainer.validate_all(
val_dataloaders=[
cstorch.utils.data.DataLoader(
GptHDF5MapDataProcessor,
data_dir="/path/to/validation/data1",
batch_size=64,
...,
),
cstorch.utils.data.DataLoader(
GptHDF5MapDataProcessor,
data_dir="/path/to/validation/data2",
batch_size=64,
...,
),
]
...
)
```
#### ckpt\_paths
Specifies the paths to the checkpoints to load.
```YAML theme={null}
trainer:
init:
...
validate_all:
...
ckpt_paths:
- /path/to/checkpoint1
- /glob/path/to/checkpoint*
...
```
```python theme={null}
from cerebras.modelzoo import Trainer
trainer = Trainer(...)
trainer.validate(
...,
ckpt_paths=[
"/path/to/checkpoint1",
"/glob/path/to/checkpoint*",
]
)
```
Globs are accepted as well.
All validation dataloaders are used to run validation for every checkpoint. So, effectively, [`validate_all`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html#cerebras.modelzoo.Trainer.validate_all) is doing
```python theme={null}
from cerebras.modelzoo import Trainer
trainer = Trainer(...)
for ckpt_path in ckpt_paths:
trainer.load_checkpoint(ckpt_path)
for val_dataloader in val_dataloaders:
trainer.validate(val_dataloader)
```
## Legacy Specification
Versions 2.2 and earlier used the following YAML specification, now referred to as the legacy YAML specification:
```Bash theme={null}
model:
...
optimizer:
...
train_input:
...
eval_input:
...
runconfig:
...
```
We've updated it to make full use of the [`Trainer`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html) class. The training scripts in Model Zoo will detect if you've passed in a legacy configuration and will automatically invoke a converter tool before constructing and using the trainer.
However, if you'd like to manually convert a legacy YAML or learn how to specify a specific parameter in the Trainer YAML, learn more in [Convert Legacy to Trainer YAML](../model-zoo/trainer-configuration-overview/correspondance-from-legacy-to-trainer).
## What’s next?
To learn more about how you can use the [`Trainer`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html) in some core workflows, you can check out [Pretraining with Upstream Validation](../model-zoo/core-workflows/pretraining-with-upstream-validation).
To learn more about how you can extend the capabilities of the [`Trainer`](https://training-api.cerebras.ai/en/latest/wsc/Model-zoo/api/index.html) class, you can check out [Trainer Components](../model-zoo/components/trainer-components/trainer-components).