Features for minimal input?

I am trying to create a simple input just to test the model and make sure I understand everything correctly. I just want to have 3 amnio acids as input and nothing else.

So here is my code so far:

```
from alphafold.model import model
from alphafold.model import config
from alphafold.model import data
import random
import sys

model_name = "model_1"
Nres=3
Nextra=1
features = {
'seq_length':[Nres]*Nres,
'aatype':[[1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]*Nres],
'sequence':[4,5,6],
'msa':[[[1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0]*Nres]*Nextra]
}
cfg = config.model_config(model_name)

params = data.get_model_haiku_params(model_name, 'alphafold/data')
model_runner = model.RunModel(cfg, params)

processed_feature_dict = model_runner.process_features(features , random_seed=0)
prediction = model_runner.predict(processed_feature_dict, random_seed=random.randrange(sys.maxsize))
print(prediction)
```

Obviously this doesn't work as I don't have the features struct correct. Can you suggest some minimal features to get this working?