utils.callbacks.swanlab
utils.callbacks.swanlab
Callbacks for SwanLab integration
Classes
| Name | Description |
|---|---|
| CustomSwanLabCallback | Lightweight SwanLab callback that directly logs metrics without using |
| SaveAxolotlConfigtoSwanLabCallback | Callback to save axolotl config to SwanLab |
CustomSwanLabCallback
utils.callbacks.swanlab.CustomSwanLabCallback()Lightweight SwanLab callback that directly logs metrics without using SwanLab’s transformers integration (which requires omegaconf).
This avoids the antlr4 version conflict between omegaconf and axolotl.
Methods
| Name | Description |
|---|---|
| on_log | Called when logging metrics |
| on_train_begin | Called at the beginning of training |
| on_train_end | Called at the end of training |
| setup | Lazy initialization of SwanLab |
on_log
utils.callbacks.swanlab.CustomSwanLabCallback.on_log(
args,
state,
control,
logs=None,
**kwargs,
)Called when logging metrics
on_train_begin
utils.callbacks.swanlab.CustomSwanLabCallback.on_train_begin(
args,
state,
control,
**kwargs,
)Called at the beginning of training
on_train_end
utils.callbacks.swanlab.CustomSwanLabCallback.on_train_end(
args,
state,
control,
**kwargs,
)Called at the end of training
setup
utils.callbacks.swanlab.CustomSwanLabCallback.setup()Lazy initialization of SwanLab
SaveAxolotlConfigtoSwanLabCallback
utils.callbacks.swanlab.SaveAxolotlConfigtoSwanLabCallback(axolotl_config_path)Callback to save axolotl config to SwanLab