If I remember correctly, it used to be possible to use specific, older models, like so:
from comet import download_model, load_from_checkpoint
if __name__ == "__main__":
model_path = download_model("eamt22-cometinho-da")
model = load_from_checkpoint(model_path)
data = [
{
"src": "Dem Feuer konnte Einhalt geboten werden",
"mt": "The fire could be stopped",
"ref": "They were able to control the fire."
},
{
"src": "Schulen und Kindergärten wurden eröffnet.",
"mt": "Schools and kindergartens were open",
"ref": "Schools and kindergartens opened"
}
]
model_output = model.predict(data, batch_size=8, gpus=0)
print(model_output)
It seems that in v2.0, you have made the switch to the hugging face hub. I think that's a good thing but it is unclear how I can still use the older models now. If I try the code above, an error is thrown stating that that repo cannot be found on the HF hub.
Maybe the older checkpoints can also be uploaded to the hub? Is there an alternative, auto-download, way to use the older checkpoints? This file can be updated accordingly to tell users how to load/use which model.
❓ Questions and Help
If I remember correctly, it used to be possible to use specific, older models, like so:
It seems that in v2.0, you have made the switch to the hugging face hub. I think that's a good thing but it is unclear how I can still use the older models now. If I try the code above, an error is thrown stating that that repo cannot be found on the HF hub.
Maybe the older checkpoints can also be uploaded to the hub? Is there an alternative, auto-download, way to use the older checkpoints? This file can be updated accordingly to tell users how to load/use which model.