Using Model.Predict()

[ayush714]

Hi, I trained my model, Now, I am using model.predict() to generate predictions, so here is my command for prediction:-

test_input = f"""what did the client ask for? \\n Client said: Hello, how are you? Can you please reset my account"""
answer = f"reset my account" 
# model predict by test_input  
var = None 
print(model.predict(test_input, output_file=var)) 

I am getting below error:-

ValueError: No Task or Mixture found with name 'all_mix'. Available:
 - qa_t5_meshs

[ayush714]

@craffel Plz Have a look at this, I am waiting for 3 days

[adarob]

Can you share additional context for how you're loading the model?

[ayush714]

Hi @adarob I am fine tuning the model using below code,

    model = t5.models.MtfModel(
        model_dir=MODEL_DIR,
        tpu=None,
        model_parallelism=model_parallelism,
        batch_size=train_batch_size,
        sequence_length={"inputs": 128, "targets": 32},
        learning_rate_schedule=0.003,
        save_checkpoints_steps=5000,
        keep_checkpoint_max=keep_checkpoint_max if ON_CLOUD else None,
        iterations_per_loop=100,
    )
    FINETUNE_STEPS = 4

    model.finetune(
        mixture_or_task_name="qa_t5_meshs",
        pretrained_model_dir=PRETRAINED_DIR,
        finetune_steps=FINETUNE_STEPS
    )
    print("Done")

and then trying to make predictions from the model

[adarob]

When you call model.predict, try setting vocabulary=t5.data.get_default_vocabulary().

This will avoid the need to load the Task used during pretraining (from the operative config) just to access the vocabulary.

[ayush714]

Hey @adarob That Error got resolved but a new error game while making predictions which is below:-

NotFoundError: what did the client ask for? \n Client said: Hello, how are you? Can you please reset my account; No such file or directory
  In call to configurable 'decode_from_file' (<function decode_from_file at 0x7f8e022ab820>)
  In call to configurable 'infer_model' (<function infer_model at 0x7f8e022ae0d0>)