Simpler auto log hparams

[eracah]

What does this PR do?

logs just hparams passed directly to Trainer. Any non-[int, float, string, bool, None] objects are just logged as obj.__class__.__name__ instead of crawling them for haprams

What issue(s) does this change relate to?

fix CO-586
fix CO-203

Manual Test:

transform = transforms.Compose([transforms.ToTensor()])
train_set = datasets.MNIST("data", train=True, download=True, transform=transform)
val_set = datasets.MNIST("data", train=False, download=True, transform=transform)
train_dataloader = DataLoader(train_set, batch_size=128)
eval_dataloader = DataLoader(val_set, batch_size=64)
model=mnist_model(num_classes=10)


trainer = Trainer(
    model=model,
    train_dataloader=train_dataloader,
    eval_dataloader=eval_dataloader,
    max_duration="3ep",
    train_subset_num_batches=4,
    optimizers=[Adam(model.parameters(), eps=1e-3)],
    progress_bar=False,
    log_traces=False,
    log_to_console=True,
    console_log_interval='1ep',
    auto_log_hparams=True,
    algorithms=[ChannelsLast(), ColOut(), BlurPool()],
    callbacks=[SpeedMonitor(), MemoryMonitor(), ImageVisualizer()],
    loggers=[WandBLogger()],
)
trainer.fit()

Wandb Result
Console result:

Config:
algorithm_passes: null
algorithms:
- ChannelsLast
- ColOut
- BlurPool
auto_log_hparams: true
auto_microbatching: false
autoresume: false
blurpool/num_blurconv_layers: 0
blurpool/num_blurpool_layers: 0
callbacks:
- SpeedMonitor
- MemoryMonitor
- ImageVisualizer
console_log_interval: 1ep
console_stream: stderr
ddp_sync_strategy: null
deepspeed_config: null
deterministic_mode: false
device: DeviceCPU
device_train_microbatch_size: null
dist_timeout: 1800.0
eval_batch_split: 1
eval_dataloader: DataLoader
eval_interval: 1
eval_subset_num_batches: -1
fsdp_config: null
grad_accum: 1
latest_remote_file_name: null
load_exclude_algorithms: null
load_ignore_keys: null
load_object_store: null
load_path: null
load_progress_bar: true
load_strict_model_weights: false
load_weights_only: false
log_to_console: true
log_traces: false
loggers:
- WandBLogger
- ConsoleLogger
max_duration: 3ep
model: ComposerClassifier
num_cpus_per_node: 1
num_nodes: 1
num_optimizers: 1
optimizers:
- Adam
precision: Precision
profiler: null
progress_bar: false
python_log_level: null
rank_zero_seed: 791585187
remote_ud_has_format_string:
- false
- false
run_name: 1672804048-impossible-viper
save_filename: ep{epoch}-ba{batch}-rank{rank}.pt
save_folder: null
save_interval: 1ep
save_latest_filename: latest-rank{rank}.pt
save_num_checkpoints_to_keep: -1
save_overwrite: false
save_weights_only: false
scale_schedule_ratio: 1.0
schedulers: null
seed: 791585187
step_schedulers_every_batch: null
train_dataloader: DataLoader
train_dataloader_label: train
train_subset_num_batches: 4
using_device_microbatch_size: false

[dakinggg]

LGTM