Biomedical Stanza models error during loading pos - KeyError: 'bert_finetune'

[mh-n]

Describe the bug Trying to make a Stanza pipeline with any biomedical package and receive the error 'KeyError: bert_finetune' while loading the POS tagger (occurs with or without i2b2 NER processor).

To Reproduce Steps to reproduce the behavior:

1. Running nlp = stanza.Pipeline('en', package='mimic', processors={'ner': 'i2b2'}) or nlp = stanza.Pipeline('en', package='craft', processors={'ner': 'i2b2'}) or nlp = stanza.Pipeline('en', package='genia', processors={'ner': 'i2b2'})
2. Error, example for 'mimic' package:

   
   `2024-02-29 16:41:08 INFO: Checking for updates to resources.json in case models have been updated.  Note: this behavior can be turned off with download_method=None or download_method=DownloadMethod.REUSE_RESOURCES
   INFO:stanza:Checking for updates to resources.json in case models have been updated.  Note: this behavior can be turned off with download_method=None or download_method=DownloadMethod.REUSE_RESOURCES
   Downloading https://raw.githubusercontent.com/stanfordnlp/stanza-resources/main/resources_1.8.0.json:
   373k/? [00:00<00:00, 53.2MB/s]
   2024-02-29 16:41:08 INFO: Downloaded file to /home/idies/stanza_resources/resources.json
   INFO:stanza:Downloaded file to /home/idies/stanza_resources/resources.json
   2024-02-29 16:41:09 INFO: Loading these models for language: en (English):
   ==============================
   | Processor | Package        |
   ------------------------------
   | tokenize  | mimic          |
   | pos       | mimic_charlm   |
   | lemma     | mimic_nocharlm |
   | depparse  | mimic_charlm   |
   | ner       | i2b2           |
   ==============================

INFO:stanza:Loading these models for language: en (English):

| Processor | Package |

| tokenize | mimic | | pos | mimic_charlm | | lemma | mimic_nocharlm | | depparse | mimic_charlm | | ner | i2b2 |

2024-02-29 16:41:09 INFO: Using device: cpu INFO:stanza:Using device: cpu 2024-02-29 16:41:09 INFO: Loading: tokenize INFO:stanza:Loading: tokenize 2024-02-29 16:41:09 INFO: Loading: pos INFO:stanza:Loading: pos

KeyError Traceback (most recent call last) Cell In[95], line 2 1 #nlp = stanza.Pipeline('nl', processors={'ner': 'conll02'}) ----> 2 nlp = stanza.Pipeline('en', package='mimic', processors={'ner': 'i2b2'})

File ~/miniconda3/lib/python3.9/site-packages/stanza/pipeline/core.py:305, in Pipeline.init(self, lang, dir, package, processors, logging_level, verbose, use_gpu, model_dir, download_method, resources_url, resources_branch, resources_version, resources_filepath, proxies, foundation_cache, device, allow_unknown_language, **kwargs) 302 logger.debug(curr_processor_config) 303 try: 304 # try to build processor, throw an exception if there is a requirements issue --> 305 self.processors[processor_name] = NAME_TO_PROCESSOR_CLASS[processor_name](config=curr_processor_config, 306 pipeline=self, 307 device=self.device) 308 except ProcessorRequirementsException as e: 309 # if there was a requirements issue, add it to list which will be printed at end 310 pipeline_reqs_exceptions.append(e)

File ~/miniconda3/lib/python3.9/site-packages/stanza/pipeline/processor.py:193, in UDProcessor.init(self, config, pipeline, device) 191 self._vocab = None 192 if not hasattr(self, '_variant'): --> 193 self._set_up_model(config, pipeline, device) 195 # build the final config for the processor 196 self._set_up_final_config(config)

File ~/miniconda3/lib/python3.9/site-packages/stanza/pipeline/pos_processor.py:32, in POSProcessor._set_up_model(self, config, pipeline, device) 29 args = {'charlm_forward_file': config.get('forward_charlm_path', None), 30 'charlm_backward_file': config.get('backward_charlm_path', None)} 31 # set up trainer ---> 32 self._trainer = Trainer(pretrain=self.pretrain, model_file=config['model_path'], device=device, args=args, foundation_cache=pipeline.foundation_cache) 33 self._tqdm = 'tqdm' in config and config['tqdm']

File ~/miniconda3/lib/python3.9/site-packages/stanza/models/pos/trainer.py:44, in Trainer.init(self, args, vocab, pretrain, model_file, device, foundation_cache) 40 self.optimizers = utils.get_split_optimizer(self.args['optim'], self.model, self.args['lr'], betas=(0.9, self.args['beta2']), eps=1e-6, weight_decay=self.args.get('initial_weight_decay', None), bert_learning_rate=self.args.get('bert_learning_rate', 0.0), is_peft=self.args.get("peft", False)) 42 self.schedulers = {} ---> 44 if self.args["bert_finetune"]: 45 import transformers 46 warmup_scheduler = transformers.get_linear_schedule_with_warmup( 47 self.optimizers["bert_optimizer"], 48 # todo late starting? 49 0, self.args["max_steps"])

KeyError: 'bert_finetune'`

**Expected behavior**
No issues running basic Stanza model - i.e. `nlp = stanza.Pipeline('nl', processors={'ner': 'conll02'})`
Output: 

`2024-02-29 16:59:36 INFO: Checking for updates to resources.json in case models have been updated. Note: this behavior can be turned off with download_method=None or download_method=DownloadMethod.REUSE_RESOURCES INFO:stanza:Checking for updates to resources.json in case models have been updated. Note: this behavior can be turned off with download_method=None or download_method=DownloadMethod.REUSE_RESOURCES Downloading https://raw.githubusercontent.com/stanfordnlp/stanza-resources/main/resources_1.8.0.json: 373k/? [00:00<00:00, 37.7MB/s] 2024-02-29 16:59:36 INFO: Downloaded file to /home/idies/stanza_resources/resources.json INFO:stanza:Downloaded file to /home/idies/stanza_resources/resources.json 2024-02-29 16:59:37 INFO: Loading these models for language: nl (Dutch):

| Processor | Package |

| tokenize | alpino | | mwt | alpino | | pos | alpino_charlm | | lemma | alpino_nocharlm | | depparse | alpino_charlm | | ner | conll02 |

INFO:stanza:Loading these models for language: nl (Dutch):

| Processor | Package |

| tokenize | alpino | | mwt | alpino | | pos | alpino_charlm | | lemma | alpino_nocharlm | | depparse | alpino_charlm | | ner | conll02 |

2024-02-29 16:59:37 INFO: Using device: cpu INFO:stanza:Using device: cpu 2024-02-29 16:59:37 INFO: Loading: tokenize INFO:stanza:Loading: tokenize 2024-02-29 16:59:37 INFO: Loading: mwt INFO:stanza:Loading: mwt 2024-02-29 16:59:37 INFO: Loading: pos INFO:stanza:Loading: pos 2024-02-29 16:59:39 INFO: Loading: lemma INFO:stanza:Loading: lemma 2024-02-29 16:59:39 INFO: Loading: depparse INFO:stanza:Loading: depparse 2024-02-29 16:59:40 INFO: Loading: ner INFO:stanza:Loading: ner 2024-02-29 16:59:41 INFO: Done loading processors! INFO:stanza:Done loading processors!`

**Environment (please complete the following information):**
 - OS: Windows 
 - Python version: 3.9.17
 - Stanza version:  1.8.0

Not sure if there are some obvious dependencies missing on my end. Appreciate any help you can provide. Thank you!

[AngledLuffa]

Yikes, that's not good. I just added a fix to the dev branch. I can push that fix at a new version 1.8.1 either later tonight or tomorrow

[AngledLuffa]

This should now be fixed in v1.8.1

[AngledLuffa]

Fixed?

[mh-n]

Sorry for the delay! Just got to check and looks good on my end now. Thanks!