# Generator object issue help

**URL:** <https://support.prodi.gy/t/generator-object-issue-help/887>\
**Category:** Uncategorized\
**Tags:** usage, solved\
**Created:** [October 16, 2018, 2:49pm UTC](https://support.prodi.gy/t/generator-object-issue-help/887 "2018-10-16T14:49:51Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![yuskey](https://avatars.discourse-cdn.com/v4/letter/y/ce73a5/32.png) [@yuskey](https://support.prodi.gy/u/yuskey)\
**Post date:** [October 16, 2018, 2:49pm UTC](https://support.prodi.gy/t/generator-object-issue-help/887/1 "2018-10-16T14:49:51Z")

</div>

So I’m trying to write a custom loader for my text. The text is in the form of a csv document and my loader function is below:

```
def load_data(path): 
	with open(path) as f:
		reader = csv.DictReader(f)
		for row in reader:
			text = row.get('Description of the Event')
			task = {'text': text}
			yield task

```

The issue I’m running into is when running the recipe I get an error stating object of type ‘generator’ has no length. The recipe is below:

```
@recipe('sent.teach',
		dataset=recipe_args['dataset'],
		spacy_model=recipe_args['spacy_model'],
		source=recipe_args['source'],
		label=recipe_args['label_set'],
		long_text=("Long text", "flag", "L", bool))
def teach(dataset, spacy_model, source=None, label='',long_text=False):
	log('RECIPE: Starting recipe sent.teach', locals())
	DB = connect()
	path = os.getcwd()
	if spacy_model:
		print("loading model")
		nlp = spacy.load(spacy_model) 

	stream = load_data(os.path.join(path, 'data', 'sims_2018.csv'))
	
	# Rank the stream. Note this is continuous, as model() is a generator.
	# As we call model.update(), the ranking of examples changes.
	stream = prefer_uncertain(nlp(stream))
	
	return {
		'view_id': 'classification',
		'dataset': dataset,
		'stream': stream,
		'update': nlp.update,
		'config': {'lang': nlp.lang, 'labels': nlp.textcat.labels}
	}

```

I used the textcat.teach as a template for creating my recipe. I know the error arises in  
`stream = prefer_uncertain(nlp(stream))`, however, I thought the input into the predict function was supposed to be a generator? I’m not sure what I’m doing wrong here.

---

<div class="post-metadata">

**Author:** ![yuskey](https://avatars.discourse-cdn.com/v4/letter/y/ce73a5/32.png) [@yuskey](https://support.prodi.gy/u/yuskey)\
**Post date:** [October 16, 2018, 3:26pm UTC](https://support.prodi.gy/t/generator-object-issue-help/887/2 "2018-10-16T15:26:23Z")

</div>

Ah never mind, I figured out what I did wrong. I didn’t realize that the call to the nlp only took in text so I just created a predict method that takes in the nlp model and the stream and returns the score and example.

```
def predict(nlp, stream):
	for eg in stream:
		doc = nlp(eg['text'])
		eg['label'] = max(doc.cats)
		score = max(doc.cats.values())
		yield (score, eg)
```

---

<div class="post-metadata">

**Author:** ![ines](https://sea2.discourse-cdn.com/flex020/user_avatar/support.prodi.gy/ines/32/3_2.png) [@ines](https://support.prodi.gy/u/ines)\
**Post date:** [October 16, 2018, 5:13pm UTC](https://support.prodi.gy/t/generator-object-issue-help/887/3 "2018-10-16T17:13:59Z")

</div>

Yes, that makes a lot of sense – thanks for the update and glad you got it working! 👍